diff --git a/.clang-tidy b/.clang-tidy index 2ca1402ddf1..860e7b3189f 100644 --- a/.clang-tidy +++ b/.clang-tidy @@ -1,6 +1,14 @@ +# To run clang-tidy from CMake, build ClickHouse with -DENABLE_CLANG_TIDY=1. To show all warnings, it is +# recommended to pass "-k0" to Ninja. + # Enable all checks + disale selected checks. Feel free to remove disabled checks from below list if # a) the new check is not controversial (this includes many checks in readability-* and google-*) or # b) too noisy (checks with > 100 new warnings are considered noisy, this includes e.g. cppcoreguidelines-*). + +# TODO Let clang-tidy check headers in further directories +# --> HeaderFilterRegex: '^.*/(src|base|programs|utils)/.*(h|hpp)$' +HeaderFilterRegex: '^.*/(base)/.*(h|hpp)$' + Checks: '*, -abseil-*, diff --git a/.github/workflows/backport_branches.yml b/.github/workflows/backport_branches.yml index e1b2b1fad01..a1086452184 100644 --- a/.github/workflows/backport_branches.yml +++ b/.github/workflows/backport_branches.yml @@ -437,7 +437,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateless_debug REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateless tests (address) + CHECK_NAME=Stateless tests (asan) REPO_COPY=${{runner.temp}}/stateless_debug/ClickHouse KILL_TIMEOUT=10800 EOF @@ -521,7 +521,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stress_thread REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stress test (thread) + CHECK_NAME=Stress test (tsan) REPO_COPY=${{runner.temp}}/stress_thread/ClickHouse EOF - name: Download json reports diff --git a/.github/workflows/master.yml b/.github/workflows/master.yml index 2acc1468328..e8e3deceef5 100644 --- a/.github/workflows/master.yml +++ b/.github/workflows/master.yml @@ -1287,7 +1287,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateless_debug REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateless tests (address) + CHECK_NAME=Stateless tests (asan) REPO_COPY=${{runner.temp}}/stateless_debug/ClickHouse KILL_TIMEOUT=10800 RUN_BY_HASH_NUM=0 @@ -1326,7 +1326,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateless_debug REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateless tests (address) + CHECK_NAME=Stateless tests (asan) REPO_COPY=${{runner.temp}}/stateless_debug/ClickHouse KILL_TIMEOUT=10800 RUN_BY_HASH_NUM=1 @@ -1365,7 +1365,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateless_tsan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateless tests (thread) + CHECK_NAME=Stateless tests (tsan) REPO_COPY=${{runner.temp}}/stateless_tsan/ClickHouse KILL_TIMEOUT=10800 RUN_BY_HASH_NUM=0 @@ -1404,7 +1404,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateless_tsan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateless tests (thread) + CHECK_NAME=Stateless tests (tsan) REPO_COPY=${{runner.temp}}/stateless_tsan/ClickHouse KILL_TIMEOUT=10800 RUN_BY_HASH_NUM=1 @@ -1443,7 +1443,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateless_tsan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateless tests (thread) + CHECK_NAME=Stateless tests (tsan) REPO_COPY=${{runner.temp}}/stateless_tsan/ClickHouse KILL_TIMEOUT=10800 RUN_BY_HASH_NUM=2 @@ -1519,7 +1519,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateless_memory REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateless tests (memory) + CHECK_NAME=Stateless tests (msan) REPO_COPY=${{runner.temp}}/stateless_memory/ClickHouse KILL_TIMEOUT=10800 RUN_BY_HASH_NUM=0 @@ -1558,7 +1558,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateless_memory REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateless tests (memory) + CHECK_NAME=Stateless tests (msan) REPO_COPY=${{runner.temp}}/stateless_memory/ClickHouse KILL_TIMEOUT=10800 RUN_BY_HASH_NUM=1 @@ -1597,7 +1597,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateless_memory REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateless tests (memory) + CHECK_NAME=Stateless tests (msan) REPO_COPY=${{runner.temp}}/stateless_memory/ClickHouse KILL_TIMEOUT=10800 RUN_BY_HASH_NUM=2 @@ -1830,7 +1830,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateful_debug REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateful tests (address) + CHECK_NAME=Stateful tests (asan) REPO_COPY=${{runner.temp}}/stateful_debug/ClickHouse KILL_TIMEOUT=3600 EOF @@ -1867,7 +1867,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateful_tsan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateful tests (thread) + CHECK_NAME=Stateful tests (tsan) REPO_COPY=${{runner.temp}}/stateful_tsan/ClickHouse KILL_TIMEOUT=3600 EOF @@ -1904,7 +1904,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateful_msan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateful tests (memory) + CHECK_NAME=Stateful tests (msan) REPO_COPY=${{runner.temp}}/stateful_msan/ClickHouse KILL_TIMEOUT=3600 EOF @@ -2018,7 +2018,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stress_thread REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stress test (address) + CHECK_NAME=Stress test (asan) REPO_COPY=${{runner.temp}}/stress_thread/ClickHouse EOF - name: Download json reports @@ -2058,7 +2058,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stress_thread REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stress test (thread) + CHECK_NAME=Stress test (tsan) REPO_COPY=${{runner.temp}}/stress_thread/ClickHouse EOF - name: Download json reports @@ -2094,7 +2094,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stress_memory REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stress test (memory) + CHECK_NAME=Stress test (msan) REPO_COPY=${{runner.temp}}/stress_memory/ClickHouse EOF - name: Download json reports @@ -2130,7 +2130,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stress_undefined REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stress test (undefined) + CHECK_NAME=Stress test (ubsan) REPO_COPY=${{runner.temp}}/stress_undefined/ClickHouse EOF - name: Download json reports @@ -2319,7 +2319,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/integration_tests_tsan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Integration tests (thread) + CHECK_NAME=Integration tests (tsan) REPO_COPY=${{runner.temp}}/integration_tests_tsan/ClickHouse RUN_BY_HASH_NUM=0 RUN_BY_HASH_TOTAL=4 @@ -2357,7 +2357,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/integration_tests_tsan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Integration tests (thread) + CHECK_NAME=Integration tests (tsan) REPO_COPY=${{runner.temp}}/integration_tests_tsan/ClickHouse RUN_BY_HASH_NUM=1 RUN_BY_HASH_TOTAL=4 @@ -2395,7 +2395,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/integration_tests_tsan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Integration tests (thread) + CHECK_NAME=Integration tests (tsan) REPO_COPY=${{runner.temp}}/integration_tests_tsan/ClickHouse RUN_BY_HASH_NUM=2 RUN_BY_HASH_TOTAL=4 @@ -2433,7 +2433,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/integration_tests_tsan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Integration tests (thread) + CHECK_NAME=Integration tests (tsan) REPO_COPY=${{runner.temp}}/integration_tests_tsan/ClickHouse RUN_BY_HASH_NUM=3 RUN_BY_HASH_TOTAL=4 @@ -2550,7 +2550,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/ast_fuzzer_asan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=AST fuzzer (ASan) + CHECK_NAME=AST fuzzer (asan) REPO_COPY=${{runner.temp}}/ast_fuzzer_asan/ClickHouse EOF - name: Download json reports @@ -2586,7 +2586,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/ast_fuzzer_tsan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=AST fuzzer (TSan) + CHECK_NAME=AST fuzzer (tsan) REPO_COPY=${{runner.temp}}/ast_fuzzer_tsan/ClickHouse EOF - name: Download json reports @@ -2622,7 +2622,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/ast_fuzzer_ubsan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=AST fuzzer (UBSan) + CHECK_NAME=AST fuzzer (ubsan) REPO_COPY=${{runner.temp}}/ast_fuzzer_ubsan/ClickHouse EOF - name: Download json reports @@ -2658,7 +2658,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/ast_fuzzer_msan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=AST fuzzer (MSan) + CHECK_NAME=AST fuzzer (msan) REPO_COPY=${{runner.temp}}/ast_fuzzer_msan/ClickHouse EOF - name: Download json reports diff --git a/.github/workflows/pull_request.yml b/.github/workflows/pull_request.yml index 513df8487c4..7901008a7db 100644 --- a/.github/workflows/pull_request.yml +++ b/.github/workflows/pull_request.yml @@ -1300,7 +1300,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateless_debug REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateless tests (address) + CHECK_NAME=Stateless tests (asan) REPO_COPY=${{runner.temp}}/stateless_debug/ClickHouse KILL_TIMEOUT=10800 RUN_BY_HASH_NUM=0 @@ -1339,7 +1339,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateless_debug REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateless tests (address) + CHECK_NAME=Stateless tests (asan) REPO_COPY=${{runner.temp}}/stateless_debug/ClickHouse KILL_TIMEOUT=10800 RUN_BY_HASH_NUM=1 @@ -1378,7 +1378,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateless_tsan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateless tests (thread) + CHECK_NAME=Stateless tests (tsan) REPO_COPY=${{runner.temp}}/stateless_tsan/ClickHouse KILL_TIMEOUT=10800 RUN_BY_HASH_NUM=0 @@ -1417,7 +1417,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateless_tsan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateless tests (thread) + CHECK_NAME=Stateless tests (tsan) REPO_COPY=${{runner.temp}}/stateless_tsan/ClickHouse KILL_TIMEOUT=10800 RUN_BY_HASH_NUM=1 @@ -1456,7 +1456,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateless_tsan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateless tests (thread) + CHECK_NAME=Stateless tests (tsan) REPO_COPY=${{runner.temp}}/stateless_tsan/ClickHouse KILL_TIMEOUT=10800 RUN_BY_HASH_NUM=2 @@ -1532,7 +1532,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateless_memory REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateless tests (memory) + CHECK_NAME=Stateless tests (msan) REPO_COPY=${{runner.temp}}/stateless_memory/ClickHouse KILL_TIMEOUT=10800 RUN_BY_HASH_NUM=0 @@ -1571,7 +1571,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateless_memory REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateless tests (memory) + CHECK_NAME=Stateless tests (msan) REPO_COPY=${{runner.temp}}/stateless_memory/ClickHouse KILL_TIMEOUT=10800 RUN_BY_HASH_NUM=1 @@ -1610,7 +1610,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateless_memory REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateless tests (memory) + CHECK_NAME=Stateless tests (msan) REPO_COPY=${{runner.temp}}/stateless_memory/ClickHouse KILL_TIMEOUT=10800 RUN_BY_HASH_NUM=2 @@ -1766,7 +1766,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateless_flaky_asan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateless tests flaky check (address) + CHECK_NAME=Stateless tests flaky check (asan) REPO_COPY=${{runner.temp}}/stateless_flaky_asan/ClickHouse KILL_TIMEOUT=3600 EOF @@ -1927,7 +1927,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateful_debug REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateful tests (address) + CHECK_NAME=Stateful tests (asan) REPO_COPY=${{runner.temp}}/stateful_debug/ClickHouse KILL_TIMEOUT=3600 EOF @@ -1964,7 +1964,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateful_tsan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateful tests (thread) + CHECK_NAME=Stateful tests (tsan) REPO_COPY=${{runner.temp}}/stateful_tsan/ClickHouse KILL_TIMEOUT=3600 EOF @@ -2001,7 +2001,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateful_msan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateful tests (memory) + CHECK_NAME=Stateful tests (msan) REPO_COPY=${{runner.temp}}/stateful_msan/ClickHouse KILL_TIMEOUT=3600 EOF @@ -2115,7 +2115,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stress_thread REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stress test (address) + CHECK_NAME=Stress test (asan) REPO_COPY=${{runner.temp}}/stress_thread/ClickHouse EOF - name: Download json reports @@ -2155,7 +2155,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stress_thread REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stress test (thread) + CHECK_NAME=Stress test (tsan) REPO_COPY=${{runner.temp}}/stress_thread/ClickHouse EOF - name: Download json reports @@ -2191,7 +2191,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stress_memory REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stress test (memory) + CHECK_NAME=Stress test (msan) REPO_COPY=${{runner.temp}}/stress_memory/ClickHouse EOF - name: Download json reports @@ -2227,7 +2227,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stress_undefined REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stress test (undefined) + CHECK_NAME=Stress test (ubsan) REPO_COPY=${{runner.temp}}/stress_undefined/ClickHouse EOF - name: Download json reports @@ -2302,7 +2302,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/ast_fuzzer_asan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=AST fuzzer (ASan) + CHECK_NAME=AST fuzzer (asan) REPO_COPY=${{runner.temp}}/ast_fuzzer_asan/ClickHouse EOF - name: Download json reports @@ -2338,7 +2338,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/ast_fuzzer_tsan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=AST fuzzer (TSan) + CHECK_NAME=AST fuzzer (tsan) REPO_COPY=${{runner.temp}}/ast_fuzzer_tsan/ClickHouse EOF - name: Download json reports @@ -2374,7 +2374,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/ast_fuzzer_ubsan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=AST fuzzer (UBSan) + CHECK_NAME=AST fuzzer (ubsan) REPO_COPY=${{runner.temp}}/ast_fuzzer_ubsan/ClickHouse EOF - name: Download json reports @@ -2410,7 +2410,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/ast_fuzzer_msan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=AST fuzzer (MSan) + CHECK_NAME=AST fuzzer (msan) REPO_COPY=${{runner.temp}}/ast_fuzzer_msan/ClickHouse EOF - name: Download json reports @@ -2599,7 +2599,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/integration_tests_tsan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Integration tests (thread) + CHECK_NAME=Integration tests (tsan) REPO_COPY=${{runner.temp}}/integration_tests_tsan/ClickHouse RUN_BY_HASH_NUM=0 RUN_BY_HASH_TOTAL=4 @@ -2637,7 +2637,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/integration_tests_tsan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Integration tests (thread) + CHECK_NAME=Integration tests (tsan) REPO_COPY=${{runner.temp}}/integration_tests_tsan/ClickHouse RUN_BY_HASH_NUM=1 RUN_BY_HASH_TOTAL=4 @@ -2675,7 +2675,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/integration_tests_tsan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Integration tests (thread) + CHECK_NAME=Integration tests (tsan) REPO_COPY=${{runner.temp}}/integration_tests_tsan/ClickHouse RUN_BY_HASH_NUM=2 RUN_BY_HASH_TOTAL=4 @@ -2713,7 +2713,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/integration_tests_tsan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Integration tests (thread) + CHECK_NAME=Integration tests (tsan) REPO_COPY=${{runner.temp}}/integration_tests_tsan/ClickHouse RUN_BY_HASH_NUM=3 RUN_BY_HASH_TOTAL=4 diff --git a/.github/workflows/release_branches.yml b/.github/workflows/release_branches.yml index afeebf3c2b7..6403d00157f 100644 --- a/.github/workflows/release_branches.yml +++ b/.github/workflows/release_branches.yml @@ -591,7 +591,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateless_debug REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateless tests (address) + CHECK_NAME=Stateless tests (asan) REPO_COPY=${{runner.temp}}/stateless_debug/ClickHouse KILL_TIMEOUT=10800 RUN_BY_HASH_NUM=0 @@ -630,7 +630,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateless_debug REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateless tests (address) + CHECK_NAME=Stateless tests (asan) REPO_COPY=${{runner.temp}}/stateless_debug/ClickHouse KILL_TIMEOUT=10800 RUN_BY_HASH_NUM=1 @@ -669,7 +669,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateless_tsan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateless tests (thread) + CHECK_NAME=Stateless tests (tsan) REPO_COPY=${{runner.temp}}/stateless_tsan/ClickHouse KILL_TIMEOUT=10800 RUN_BY_HASH_NUM=0 @@ -708,7 +708,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateless_tsan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateless tests (thread) + CHECK_NAME=Stateless tests (tsan) REPO_COPY=${{runner.temp}}/stateless_tsan/ClickHouse KILL_TIMEOUT=10800 RUN_BY_HASH_NUM=1 @@ -747,7 +747,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateless_tsan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateless tests (thread) + CHECK_NAME=Stateless tests (tsan) REPO_COPY=${{runner.temp}}/stateless_tsan/ClickHouse KILL_TIMEOUT=10800 RUN_BY_HASH_NUM=2 @@ -823,7 +823,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateless_memory REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateless tests (memory) + CHECK_NAME=Stateless tests (msan) REPO_COPY=${{runner.temp}}/stateless_memory/ClickHouse KILL_TIMEOUT=10800 RUN_BY_HASH_NUM=0 @@ -862,7 +862,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateless_memory REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateless tests (memory) + CHECK_NAME=Stateless tests (msan) REPO_COPY=${{runner.temp}}/stateless_memory/ClickHouse KILL_TIMEOUT=10800 RUN_BY_HASH_NUM=1 @@ -901,7 +901,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateless_memory REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateless tests (memory) + CHECK_NAME=Stateless tests (msan) REPO_COPY=${{runner.temp}}/stateless_memory/ClickHouse KILL_TIMEOUT=10800 RUN_BY_HASH_NUM=2 @@ -1134,7 +1134,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateful_debug REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateful tests (address) + CHECK_NAME=Stateful tests (asan) REPO_COPY=${{runner.temp}}/stateful_debug/ClickHouse KILL_TIMEOUT=3600 EOF @@ -1171,7 +1171,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateful_tsan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateful tests (thread) + CHECK_NAME=Stateful tests (tsan) REPO_COPY=${{runner.temp}}/stateful_tsan/ClickHouse KILL_TIMEOUT=3600 EOF @@ -1208,7 +1208,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stateful_msan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stateful tests (memory) + CHECK_NAME=Stateful tests (msan) REPO_COPY=${{runner.temp}}/stateful_msan/ClickHouse KILL_TIMEOUT=3600 EOF @@ -1322,7 +1322,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stress_thread REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stress test (address) + CHECK_NAME=Stress test (asan) REPO_COPY=${{runner.temp}}/stress_thread/ClickHouse EOF - name: Download json reports @@ -1362,7 +1362,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stress_thread REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stress test (thread) + CHECK_NAME=Stress test (tsan) REPO_COPY=${{runner.temp}}/stress_thread/ClickHouse EOF - name: Download json reports @@ -1398,7 +1398,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stress_memory REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stress test (memory) + CHECK_NAME=Stress test (msan) REPO_COPY=${{runner.temp}}/stress_memory/ClickHouse EOF - name: Download json reports @@ -1434,7 +1434,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/stress_undefined REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Stress test (undefined) + CHECK_NAME=Stress test (ubsan) REPO_COPY=${{runner.temp}}/stress_undefined/ClickHouse EOF - name: Download json reports @@ -1623,7 +1623,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/integration_tests_tsan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Integration tests (thread) + CHECK_NAME=Integration tests (tsan) REPO_COPY=${{runner.temp}}/integration_tests_tsan/ClickHouse RUN_BY_HASH_NUM=0 RUN_BY_HASH_TOTAL=4 @@ -1661,7 +1661,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/integration_tests_tsan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Integration tests (thread) + CHECK_NAME=Integration tests (tsan) REPO_COPY=${{runner.temp}}/integration_tests_tsan/ClickHouse RUN_BY_HASH_NUM=1 RUN_BY_HASH_TOTAL=4 @@ -1699,7 +1699,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/integration_tests_tsan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Integration tests (thread) + CHECK_NAME=Integration tests (tsan) REPO_COPY=${{runner.temp}}/integration_tests_tsan/ClickHouse RUN_BY_HASH_NUM=2 RUN_BY_HASH_TOTAL=4 @@ -1737,7 +1737,7 @@ jobs: cat >> "$GITHUB_ENV" << 'EOF' TEMP_PATH=${{runner.temp}}/integration_tests_tsan REPORTS_PATH=${{runner.temp}}/reports_dir - CHECK_NAME=Integration tests (thread) + CHECK_NAME=Integration tests (tsan) REPO_COPY=${{runner.temp}}/integration_tests_tsan/ClickHouse RUN_BY_HASH_NUM=3 RUN_BY_HASH_TOTAL=4 diff --git a/base/base/Decimal.h b/base/base/Decimal.h index 1efb8ba8d92..22cb577b1b2 100644 --- a/base/base/Decimal.h +++ b/base/base/Decimal.h @@ -52,15 +52,15 @@ struct Decimal constexpr Decimal(Decimal &&) noexcept = default; constexpr Decimal(const Decimal &) = default; - constexpr Decimal(const T & value_): value(value_) {} + constexpr Decimal(const T & value_): value(value_) {} // NOLINT(google-explicit-constructor) template - constexpr Decimal(const Decimal & x): value(x.value) {} + constexpr Decimal(const Decimal & x): value(x.value) {} // NOLINT(google-explicit-constructor) constexpr Decimal & operator=(Decimal &&) noexcept = default; constexpr Decimal & operator = (const Decimal &) = default; - constexpr operator T () const { return value; } + constexpr operator T () const { return value; } // NOLINT(google-explicit-constructor) template constexpr U convertTo() const @@ -111,7 +111,7 @@ public: using Base::Base; using NativeType = Base::NativeType; - constexpr DateTime64(const Base & v): Base(v) {} + constexpr DateTime64(const Base & v): Base(v) {} // NOLINT(google-explicit-constructor) }; } diff --git a/base/base/DecomposedFloat.h b/base/base/DecomposedFloat.h index 652b28966b2..f152637b94e 100644 --- a/base/base/DecomposedFloat.h +++ b/base/base/DecomposedFloat.h @@ -36,14 +36,14 @@ struct DecomposedFloat { using Traits = FloatTraits; - DecomposedFloat(T x) + explicit DecomposedFloat(T x) { memcpy(&x_uint, &x, sizeof(x)); } typename Traits::UInt x_uint; - bool is_negative() const + bool isNegative() const { return x_uint >> (Traits::bits - 1); } @@ -53,7 +53,7 @@ struct DecomposedFloat { return (exponent() == 0 && mantissa() == 0) ? 0 - : (is_negative() + : (isNegative() ? -1 : 1); } @@ -63,7 +63,7 @@ struct DecomposedFloat return (x_uint >> (Traits::mantissa_bits)) & (((1ull << (Traits::exponent_bits + 1)) - 1) >> 1); } - int16_t normalized_exponent() const + int16_t normalizedExponent() const { return int16_t(exponent()) - ((1ull << (Traits::exponent_bits - 1)) - 1); } @@ -73,20 +73,20 @@ struct DecomposedFloat return x_uint & ((1ull << Traits::mantissa_bits) - 1); } - int64_t mantissa_with_sign() const + int64_t mantissaWithSign() const { - return is_negative() ? -mantissa() : mantissa(); + return isNegative() ? -mantissa() : mantissa(); } /// NOTE Probably floating point instructions can be better. - bool is_integer_in_representable_range() const + bool isIntegerInRepresentableRange() const { return x_uint == 0 - || (normalized_exponent() >= 0 /// The number is not less than one + || (normalizedExponent() >= 0 /// The number is not less than one /// The number is inside the range where every integer has exact representation in float - && normalized_exponent() <= static_cast(Traits::mantissa_bits) + && normalizedExponent() <= static_cast(Traits::mantissa_bits) /// After multiplying by 2^exp, the fractional part becomes zero, means the number is integer - && ((mantissa() & ((1ULL << (Traits::mantissa_bits - normalized_exponent())) - 1)) == 0)); + && ((mantissa() & ((1ULL << (Traits::mantissa_bits - normalizedExponent())) - 1)) == 0)); } @@ -102,15 +102,15 @@ struct DecomposedFloat return sign(); /// Different signs - if (is_negative() && rhs > 0) + if (isNegative() && rhs > 0) return -1; - if (!is_negative() && rhs < 0) + if (!isNegative() && rhs < 0) return 1; /// Fractional number with magnitude less than one - if (normalized_exponent() < 0) + if (normalizedExponent() < 0) { - if (!is_negative()) + if (!isNegative()) return rhs > 0 ? -1 : 1; else return rhs >= 0 ? -1 : 1; @@ -121,11 +121,11 @@ struct DecomposedFloat { if (rhs == std::numeric_limits::lowest()) { - assert(is_negative()); + assert(isNegative()); - if (normalized_exponent() < static_cast(8 * sizeof(Int) - is_signed_v)) + if (normalizedExponent() < static_cast(8 * sizeof(Int) - is_signed_v)) return 1; - if (normalized_exponent() > static_cast(8 * sizeof(Int) - is_signed_v)) + if (normalizedExponent() > static_cast(8 * sizeof(Int) - is_signed_v)) return -1; if (mantissa() == 0) @@ -136,44 +136,44 @@ struct DecomposedFloat } /// Too large number: abs(float) > abs(rhs). Also the case with infinities and NaN. - if (normalized_exponent() >= static_cast(8 * sizeof(Int) - is_signed_v)) - return is_negative() ? -1 : 1; + if (normalizedExponent() >= static_cast(8 * sizeof(Int) - is_signed_v)) + return isNegative() ? -1 : 1; using UInt = std::conditional_t<(sizeof(Int) > sizeof(typename Traits::UInt)), make_unsigned_t, typename Traits::UInt>; UInt uint_rhs = rhs < 0 ? -rhs : rhs; /// Smaller octave: abs(rhs) < abs(float) /// FYI, TIL: octave is also called "binade", https://en.wikipedia.org/wiki/Binade - if (uint_rhs < (static_cast(1) << normalized_exponent())) - return is_negative() ? -1 : 1; + if (uint_rhs < (static_cast(1) << normalizedExponent())) + return isNegative() ? -1 : 1; /// Larger octave: abs(rhs) > abs(float) - if (normalized_exponent() + 1 < static_cast(8 * sizeof(Int) - is_signed_v) - && uint_rhs >= (static_cast(1) << (normalized_exponent() + 1))) - return is_negative() ? 1 : -1; + if (normalizedExponent() + 1 < static_cast(8 * sizeof(Int) - is_signed_v) + && uint_rhs >= (static_cast(1) << (normalizedExponent() + 1))) + return isNegative() ? 1 : -1; /// The same octave - /// uint_rhs == 2 ^ normalized_exponent + mantissa * 2 ^ (normalized_exponent - mantissa_bits) + /// uint_rhs == 2 ^ normalizedExponent + mantissa * 2 ^ (normalizedExponent - mantissa_bits) - bool large_and_always_integer = normalized_exponent() >= static_cast(Traits::mantissa_bits); + bool large_and_always_integer = normalizedExponent() >= static_cast(Traits::mantissa_bits); UInt a = large_and_always_integer - ? static_cast(mantissa()) << (normalized_exponent() - Traits::mantissa_bits) - : static_cast(mantissa()) >> (Traits::mantissa_bits - normalized_exponent()); + ? static_cast(mantissa()) << (normalizedExponent() - Traits::mantissa_bits) + : static_cast(mantissa()) >> (Traits::mantissa_bits - normalizedExponent()); - UInt b = uint_rhs - (static_cast(1) << normalized_exponent()); + UInt b = uint_rhs - (static_cast(1) << normalizedExponent()); if (a < b) - return is_negative() ? 1 : -1; + return isNegative() ? 1 : -1; if (a > b) - return is_negative() ? -1 : 1; + return isNegative() ? -1 : 1; /// Float has no fractional part means that the numbers are equal. - if (large_and_always_integer || (mantissa() & ((1ULL << (Traits::mantissa_bits - normalized_exponent())) - 1)) == 0) + if (large_and_always_integer || (mantissa() & ((1ULL << (Traits::mantissa_bits - normalizedExponent())) - 1)) == 0) return 0; else /// Float has fractional part means its abs value is larger. - return is_negative() ? -1 : 1; + return isNegative() ? -1 : 1; } diff --git a/base/base/JSON.h b/base/base/JSON.h index 214e9f88e9b..850b74715c6 100644 --- a/base/base/JSON.h +++ b/base/base/JSON.h @@ -38,6 +38,7 @@ */ +// NOLINTBEGIN(google-explicit-constructor) #ifdef __clang__ # pragma clang diagnostic push # pragma clang diagnostic ignored "-Wdeprecated-dynamic-exception-spec" @@ -46,6 +47,7 @@ POCO_DECLARE_EXCEPTION(Foundation_API, JSONException, Poco::Exception) #ifdef __clang__ # pragma clang diagnostic pop #endif +// NOLINTEND(google-explicit-constructor) class JSON { @@ -61,7 +63,7 @@ public: checkInit(); } - JSON(const std::string & s) : ptr_begin(s.data()), ptr_end(s.data() + s.size()), level(0) + explicit JSON(std::string_view s) : ptr_begin(s.data()), ptr_end(s.data() + s.size()), level(0) { checkInit(); } @@ -71,13 +73,7 @@ public: *this = rhs; } - JSON & operator=(const JSON & rhs) - { - ptr_begin = rhs.ptr_begin; - ptr_end = rhs.ptr_end; - level = rhs.level; - return *this; - } + JSON & operator=(const JSON & rhs) = default; const char * data() const { return ptr_begin; } const char * dataEnd() const { return ptr_end; } @@ -169,7 +165,7 @@ public: /// Перейти к следующему элементу массива или следующей name-value паре объекта. iterator & operator++(); - iterator operator++(int); + iterator operator++(int); // NOLINT(cert-dcl21-cpp) /// Есть ли в строке escape-последовательности bool hasEscapes() const; diff --git a/base/base/arithmeticOverflow.h b/base/base/arithmeticOverflow.h index 9a0e27505e1..d7242058658 100644 --- a/base/base/arithmeticOverflow.h +++ b/base/base/arithmeticOverflow.h @@ -3,6 +3,7 @@ #include #include +// NOLINTBEGIN(google-runtime-int) namespace common { @@ -206,3 +207,5 @@ namespace common return false; } } + +// NOLINTEND(google-runtime-int) diff --git a/base/base/bit_cast.h b/base/base/bit_cast.h index 5b4b0931b62..d1246b45590 100644 --- a/base/base/bit_cast.h +++ b/base/base/bit_cast.h @@ -1,6 +1,6 @@ #pragma once -#include +#include #include #include diff --git a/base/base/defines.h b/base/base/defines.h index c8c408b9c93..671253ed9e8 100644 --- a/base/base/defines.h +++ b/base/base/defines.h @@ -143,8 +143,8 @@ /// Macros for suppressing TSA warnings for specific reads/writes (instead of suppressing it for the whole function) /// Consider adding a comment before using these macros. -# define TSA_SUPPRESS_WARNING_FOR_READ(x) [&]() TSA_NO_THREAD_SAFETY_ANALYSIS -> const auto & { return (x); }() -# define TSA_SUPPRESS_WARNING_FOR_WRITE(x) [&]() TSA_NO_THREAD_SAFETY_ANALYSIS -> auto & { return (x); }() +# define TSA_SUPPRESS_WARNING_FOR_READ(x) ([&]() TSA_NO_THREAD_SAFETY_ANALYSIS -> const auto & { return (x); }()) +# define TSA_SUPPRESS_WARNING_FOR_WRITE(x) ([&]() TSA_NO_THREAD_SAFETY_ANALYSIS -> auto & { return (x); }()) /// This macro is useful when only one thread writes to a member /// and you want to read this member from the same thread without locking a mutex. diff --git a/base/base/extended_types.h b/base/base/extended_types.h index 7deb8e17b8e..b58df45a97e 100644 --- a/base/base/extended_types.h +++ b/base/base/extended_types.h @@ -5,7 +5,6 @@ #include #include - using Int128 = wide::integer<128, signed>; using UInt128 = wide::integer<128, unsigned>; using Int256 = wide::integer<256, signed>; @@ -18,7 +17,7 @@ static_assert(sizeof(UInt256) == 32); /// (std::common_type), are "set in stone". Attempting to specialize them causes undefined behavior. /// So instead of using the std type_traits, we use our own version which allows extension. template -struct is_signed +struct is_signed // NOLINT(readability-identifier-naming) { static constexpr bool value = std::is_signed_v; }; @@ -30,7 +29,7 @@ template inline constexpr bool is_signed_v = is_signed::value; template -struct is_unsigned +struct is_unsigned // NOLINT(readability-identifier-naming) { static constexpr bool value = std::is_unsigned_v; }; @@ -51,7 +50,7 @@ template concept is_integer = template concept is_floating_point = std::is_floating_point_v; template -struct is_arithmetic +struct is_arithmetic // NOLINT(readability-identifier-naming) { static constexpr bool value = std::is_arithmetic_v; }; @@ -66,9 +65,9 @@ template inline constexpr bool is_arithmetic_v = is_arithmetic::value; template -struct make_unsigned +struct make_unsigned // NOLINT(readability-identifier-naming) { - typedef std::make_unsigned_t type; + using type = std::make_unsigned_t; }; template <> struct make_unsigned { using type = UInt128; }; @@ -79,9 +78,9 @@ template <> struct make_unsigned { using type = UInt256; }; template using make_unsigned_t = typename make_unsigned::type; template -struct make_signed +struct make_signed // NOLINT(readability-identifier-naming) { - typedef std::make_signed_t type; + using type = std::make_signed_t; }; template <> struct make_signed { using type = Int128; }; @@ -92,7 +91,7 @@ template <> struct make_signed { using type = Int256; }; template using make_signed_t = typename make_signed::type; template -struct is_big_int +struct is_big_int // NOLINT(readability-identifier-naming) { static constexpr bool value = false; }; @@ -104,4 +103,3 @@ template <> struct is_big_int { static constexpr bool value = true; }; template inline constexpr bool is_big_int_v = is_big_int::value; - diff --git a/base/base/find_symbols.h b/base/base/find_symbols.h index b28749afda6..83f53773ae7 100644 --- a/base/base/find_symbols.h +++ b/base/base/find_symbols.h @@ -15,7 +15,7 @@ * * Allow to search for next character from the set of 'symbols...' in a string. * It is similar to 'strpbrk', 'strcspn' (and 'strchr', 'memchr' in the case of one symbol and '\0'), - * but with the following differencies: + * but with the following differences: * - works with any memory ranges, including containing zero bytes; * - doesn't require terminating zero byte: end of memory range is passed explicitly; * - if not found, returns pointer to end instead of nullptr; diff --git a/base/base/iostream_debug_helpers.h b/base/base/iostream_debug_helpers.h index 3a3f1a741ad..db974c911df 100644 --- a/base/base/iostream_debug_helpers.h +++ b/base/base/iostream_debug_helpers.h @@ -120,6 +120,7 @@ Out & dumpDispatchPriorities(Out & out, T && x, std::decay_t(out, x); } +// NOLINTNEXTLINE(google-explicit-constructor) struct LowPriority { LowPriority(void *) {} }; template diff --git a/base/base/itoa.h b/base/base/itoa.h index da7c2ffc73e..5e0b18d50c0 100644 --- a/base/base/itoa.h +++ b/base/base/itoa.h @@ -91,10 +91,10 @@ template using DivisionBy10PowN = typename SelectType < N, - Division, /// divide by 10 - Division, /// divide by 100 - Division, /// divide by 10000 - Division /// divide by 100000000 + Division, /// divide by 10 + Division, /// divide by 100 + Division, /// divide by 10000 + Division /// divide by 100000000 >::Result; template @@ -352,7 +352,7 @@ static inline char * writeUIntText(T x, char * p) static_assert(is_unsigned_v); int len = digits10(x); - auto pp = p + len; + auto * pp = p + len; while (x >= 100) { const auto i = x % 100; diff --git a/base/base/scope_guard.h b/base/base/scope_guard.h index 3d8f8ba0fe0..8524beac7ea 100644 --- a/base/base/scope_guard.h +++ b/base/base/scope_guard.h @@ -5,13 +5,13 @@ #include template -class [[nodiscard]] basic_scope_guard +class [[nodiscard]] BasicScopeGuard { public: - constexpr basic_scope_guard() = default; - constexpr basic_scope_guard(basic_scope_guard && src) : function{src.release()} {} + constexpr BasicScopeGuard() = default; + constexpr BasicScopeGuard(BasicScopeGuard && src) : function{src.release()} {} // NOLINT(hicpp-noexcept-move, performance-noexcept-move-constructor) - constexpr basic_scope_guard & operator=(basic_scope_guard && src) + constexpr BasicScopeGuard & operator=(BasicScopeGuard && src) // NOLINT(hicpp-noexcept-move, performance-noexcept-move-constructor) { if (this != &src) { @@ -23,11 +23,11 @@ public: template requires std::is_convertible_v - constexpr basic_scope_guard(basic_scope_guard && src) : function{src.release()} {} + constexpr BasicScopeGuard(BasicScopeGuard && src) : function{src.release()} {} // NOLINT(google-explicit-constructor) template requires std::is_convertible_v - constexpr basic_scope_guard & operator=(basic_scope_guard && src) + constexpr BasicScopeGuard & operator=(BasicScopeGuard && src) { if (this != &src) { @@ -39,13 +39,13 @@ public: template requires std::is_convertible_v - constexpr basic_scope_guard(const G & function_) : function{function_} {} + constexpr BasicScopeGuard(const G & function_) : function{function_} {} // NOLINT(google-explicit-constructor) template requires std::is_convertible_v - constexpr basic_scope_guard(G && function_) : function{std::move(function_)} {} + constexpr BasicScopeGuard(G && function_) : function{std::move(function_)} {} // NOLINT(google-explicit-constructor, bugprone-forwarding-reference-overload, bugprone-move-forwarding-reference) - ~basic_scope_guard() { invoke(); } + ~BasicScopeGuard() { invoke(); } static constexpr bool is_nullable = std::is_constructible_v; @@ -70,7 +70,7 @@ public: template requires std::is_convertible_v - basic_scope_guard & join(basic_scope_guard && other) + BasicScopeGuard & join(BasicScopeGuard && other) { if (other.function) { @@ -102,14 +102,13 @@ private: F function = F{}; }; -using scope_guard = basic_scope_guard>; +using scope_guard = BasicScopeGuard>; template -inline basic_scope_guard make_scope_guard(F && function_) { return std::forward(function_); } +inline BasicScopeGuard make_scope_guard(F && function_) { return std::forward(function_); } #define SCOPE_EXIT_CONCAT(n, ...) \ const auto scope_exit##n = make_scope_guard([&] { __VA_ARGS__; }) #define SCOPE_EXIT_FWD(n, ...) SCOPE_EXIT_CONCAT(n, __VA_ARGS__) #define SCOPE_EXIT(...) SCOPE_EXIT_FWD(__LINE__, __VA_ARGS__) - diff --git a/base/base/sort.h b/base/base/sort.h index 589469fffaa..912545979dc 100644 --- a/base/base/sort.h +++ b/base/base/sort.h @@ -14,7 +14,7 @@ template class DebugLessComparator { public: - constexpr DebugLessComparator(Comparator & cmp_) + constexpr DebugLessComparator(Comparator & cmp_) // NOLINT(google-explicit-constructor) : cmp(cmp_) {} diff --git a/base/base/strong_typedef.h b/base/base/strong_typedef.h index c9ea30b73fd..2ddea6412f5 100644 --- a/base/base/strong_typedef.h +++ b/base/base/strong_typedef.h @@ -34,8 +34,10 @@ public: template ::type> Self & operator=(T && rhs) { t = std::move(rhs); return *this;} + // NOLINTBEGIN(google-explicit-constructor) operator const T & () const { return t; } operator T & () { return t; } + // NOLINTEND(google-explicit-constructor) bool operator==(const Self & rhs) const { return t == rhs.t; } bool operator<(const Self & rhs) const { return t < rhs.t; } @@ -58,7 +60,10 @@ namespace std }; } +// NOLINTBEGIN(bugprone-macro-parentheses) + #define STRONG_TYPEDEF(T, D) \ struct D ## Tag {}; \ using D = StrongTypedef; \ +// NOLINTEND(bugprone-macro-parentheses) diff --git a/base/base/unit.h b/base/base/unit.h index 682b43512fc..1fb530be1f0 100644 --- a/base/base/unit.h +++ b/base/base/unit.h @@ -10,9 +10,11 @@ constexpr size_t GiB = 1024 * MiB; # pragma clang diagnostic ignored "-Wreserved-identifier" #endif +// NOLINTBEGIN(google-runtime-int) constexpr size_t operator"" _KiB(unsigned long long val) { return val * KiB; } constexpr size_t operator"" _MiB(unsigned long long val) { return val * MiB; } constexpr size_t operator"" _GiB(unsigned long long val) { return val * GiB; } +// NOLINTEND(google-runtime-int) #ifdef HAS_RESERVED_IDENTIFIER # pragma clang diagnostic pop diff --git a/base/base/wide_integer_to_string.h b/base/base/wide_integer_to_string.h index 8b794fe9bcb..160bf599516 100644 --- a/base/base/wide_integer_to_string.h +++ b/base/base/wide_integer_to_string.h @@ -51,8 +51,8 @@ struct fmt::formatter> { constexpr auto parse(format_parse_context & ctx) { - auto it = ctx.begin(); - auto end = ctx.end(); + const auto * it = ctx.begin(); + const auto * end = ctx.end(); /// Only support {}. if (it != end && *it != '}') diff --git a/base/glibc-compatibility/memcpy/memcpy.h b/base/glibc-compatibility/memcpy/memcpy.h index 9bee26a3722..0930dfb5c67 100644 --- a/base/glibc-compatibility/memcpy/memcpy.h +++ b/base/glibc-compatibility/memcpy/memcpy.h @@ -63,7 +63,7 @@ * Very large size of memcpy typically indicates suboptimal (not cache friendly) algorithms in code or unrealistic scenarios, * so we don't pay attention to using non-temporary stores. * - * On recent Intel CPUs, the presence of "erms" makes "rep movsb" the most benefitial, + * On recent Intel CPUs, the presence of "erms" makes "rep movsb" the most beneficial, * even comparing to non-temporary aligned unrolled stores even with the most wide registers. * * memcpy can be written in asm, C or C++. The latter can also use inline asm. @@ -214,4 +214,3 @@ tail: return ret; } - diff --git a/base/pcg-random/pcg_extras.hpp b/base/pcg-random/pcg_extras.hpp index f5ba4d48849..78ce726d48b 100644 --- a/base/pcg-random/pcg_extras.hpp +++ b/base/pcg-random/pcg_extras.hpp @@ -49,6 +49,8 @@ #include #endif +// NOLINTBEGIN(readability-identifier-naming, modernize-use-using, bugprone-macro-parentheses, google-explicit-constructor) + /* * Abstractions for compiler-specific directives */ @@ -90,8 +92,6 @@ #define PCG_EMULATED_128BIT_MATH 1 #endif -// NOLINTBEGIN(*) - namespace pcg_extras { /* @@ -553,6 +553,6 @@ std::ostream& operator<<(std::ostream& out, printable_typename) { } // namespace pcg_extras -// NOLINTEND(*) +// NOLINTEND(readability-identifier-naming, modernize-use-using, bugprone-macro-parentheses, google-explicit-constructor) #endif // PCG_EXTRAS_HPP_INCLUDED diff --git a/base/pcg-random/pcg_random.hpp b/base/pcg-random/pcg_random.hpp index 94e43e1007b..db7c3d7f66c 100644 --- a/base/pcg-random/pcg_random.hpp +++ b/base/pcg-random/pcg_random.hpp @@ -101,7 +101,7 @@ #endif /* - * The pcg_extras namespace contains some support code that is likley to + * The pcg_extras namespace contains some support code that is likely to * be useful for a variety of RNGs, including: * - 128-bit int support for platforms where it isn't available natively * - bit twiddling operations diff --git a/base/pcg-random/pcg_uint128.hpp b/base/pcg-random/pcg_uint128.hpp index 1a1f61b9366..3452ba6f1c4 100644 --- a/base/pcg-random/pcg_uint128.hpp +++ b/base/pcg-random/pcg_uint128.hpp @@ -22,7 +22,7 @@ /* * This code provides a a C++ class that can provide 128-bit (or higher) * integers. To produce 2K-bit integers, it uses two K-bit integers, - * placed in a union that allowes the code to also see them as four K/2 bit + * placed in a union that allows the code to also see them as four K/2 bit * integers (and access them either directly name, or by index). * * It may seem like we're reinventing the wheel here, because several diff --git a/contrib/NuRaft b/contrib/NuRaft index 362a8442dad..e7f834e87ec 160000 --- a/contrib/NuRaft +++ b/contrib/NuRaft @@ -1 +1 @@ -Subproject commit 362a8442dadcd230febfa7f19b86c41765c45575 +Subproject commit e7f834e87ec5c82bc111840bc1f934d5866c042d diff --git a/contrib/libcxx-cmake/CMakeLists.txt b/contrib/libcxx-cmake/CMakeLists.txt index a501c4df64f..6f42a479588 100644 --- a/contrib/libcxx-cmake/CMakeLists.txt +++ b/contrib/libcxx-cmake/CMakeLists.txt @@ -54,9 +54,8 @@ set(SRCS add_library(cxx ${SRCS}) set_target_properties(cxx PROPERTIES FOLDER "contrib/libcxx-cmake") -target_include_directories(cxx SYSTEM BEFORE PUBLIC - $ - $/src) +target_include_directories(cxx SYSTEM BEFORE PRIVATE $) +target_include_directories(cxx SYSTEM BEFORE PUBLIC $) target_compile_definitions(cxx PRIVATE -D_LIBCPP_BUILDING_LIBRARY -DLIBCXX_BUILDING_LIBCXXABI) # Enable capturing stack traces for all exceptions. diff --git a/docker/packager/binary/Dockerfile b/docker/packager/binary/Dockerfile index ba0a27c9801..b9b0c5c2c6c 100644 --- a/docker/packager/binary/Dockerfile +++ b/docker/packager/binary/Dockerfile @@ -83,5 +83,8 @@ RUN export CODENAME="$(lsb_release --codename --short | tr 'A-Z' 'a-z')" \ --yes --no-install-recommends \ && apt-get clean +# for external_symbolizer_path +RUN ln -s /usr/bin/llvm-symbolizer-15 /usr/bin/llvm-symbolizer + COPY build.sh / CMD ["bash", "-c", "/build.sh 2>&1"] diff --git a/docker/server/Dockerfile.ubuntu b/docker/server/Dockerfile.ubuntu index b9e8b89cd92..f4102a6ccaf 100644 --- a/docker/server/Dockerfile.ubuntu +++ b/docker/server/Dockerfile.ubuntu @@ -31,9 +31,6 @@ ARG deb_location_url="" # set non-empty single_binary_location_url to create docker image # from a single binary url (useful for non-standard builds - with sanitizers, for arm64). -# for example (run on aarch64 server): -# docker build . --network host --build-arg single_binary_location_url="https://builds.clickhouse.com/master/aarch64/clickhouse" -t altinity/clickhouse-server:master-testing-arm -# note: clickhouse-odbc-bridge is not supported there. ARG single_binary_location_url="" # user/group precreated explicitly with fixed uid/gid on purpose. diff --git a/docker/server/entrypoint.sh b/docker/server/entrypoint.sh index d4da5f0f38c..16372230d91 100755 --- a/docker/server/entrypoint.sh +++ b/docker/server/entrypoint.sh @@ -37,7 +37,6 @@ if [ -n "$ERROR_LOG_PATH" ]; then ERROR_LOG_DIR="$(dirname "$ERROR_LOG_PATH")"; FORMAT_SCHEMA_PATH="$(clickhouse extract-from-config --config-file "$CLICKHOUSE_CONFIG" --key=format_schema_path || true)" # There could be many disks declared in config -readarray -t FILESYSTEM_CACHE_PATHS < <(clickhouse extract-from-config --config-file "$CLICKHOUSE_CONFIG" --key='storage_configuration.disks.*.data_cache_path' || true) readarray -t DISKS_PATHS < <(clickhouse extract-from-config --config-file "$CLICKHOUSE_CONFIG" --key='storage_configuration.disks.*.path' || true) CLICKHOUSE_USER="${CLICKHOUSE_USER:-default}" @@ -51,7 +50,6 @@ for dir in "$DATA_DIR" \ "$TMP_DIR" \ "$USER_PATH" \ "$FORMAT_SCHEMA_PATH" \ - "${FILESYSTEM_CACHE_PATHS[@]}" \ "${DISKS_PATHS[@]}" do # check if variable not empty diff --git a/docker/test/style/Dockerfile b/docker/test/style/Dockerfile index 0ec3f09ab7f..683124feaa0 100644 --- a/docker/test/style/Dockerfile +++ b/docker/test/style/Dockerfile @@ -17,7 +17,7 @@ RUN apt-get update && env DEBIAN_FRONTEND=noninteractive apt-get install --yes \ python3-pip \ shellcheck \ yamllint \ - && pip3 install black boto3 codespell dohq-artifactory PyGithub unidiff pylint==2.6.2 \ + && pip3 install black==22.8.0 boto3 codespell==2.2.1 dohq-artifactory PyGithub unidiff pylint==2.6.2 \ && apt-get clean \ && rm -rf /root/.cache/pip diff --git a/docs/en/development/build-osx.md b/docs/en/development/build-osx.md index 97e4e4ddde1..12f74feb272 100644 --- a/docs/en/development/build-osx.md +++ b/docs/en/development/build-osx.md @@ -37,7 +37,7 @@ sudo xcode-select --install ``` bash brew update -brew install cmake ninja libtool gettext llvm gcc binutils grep findutils +brew install ccache cmake ninja libtool gettext llvm gcc binutils grep findutils ``` ## Checkout ClickHouse Sources {#checkout-clickhouse-sources} diff --git a/docs/en/engines/database-engines/replicated.md b/docs/en/engines/database-engines/replicated.md index 554345a3c15..f0ef1e981fe 100644 --- a/docs/en/engines/database-engines/replicated.md +++ b/docs/en/engines/database-engines/replicated.md @@ -12,7 +12,7 @@ One ClickHouse server can have multiple replicated databases running and updatin ## Creating a Database {#creating-a-database} ``` sql - CREATE DATABASE testdb ENGINE = Replicated('zoo_path', 'shard_name', 'replica_name') [SETTINGS ...] +CREATE DATABASE testdb ENGINE = Replicated('zoo_path', 'shard_name', 'replica_name') [SETTINGS ...] ``` **Engine Parameters** @@ -21,9 +21,7 @@ One ClickHouse server can have multiple replicated databases running and updatin - `shard_name` — Shard name. Database replicas are grouped into shards by `shard_name`. - `replica_name` — Replica name. Replica names must be different for all replicas of the same shard. -:::warning For [ReplicatedMergeTree](../table-engines/mergetree-family/replication.md#table_engines-replication) tables if no arguments provided, then default arguments are used: `/clickhouse/tables/{uuid}/{shard}` and `{replica}`. These can be changed in the server settings [default_replica_path](../../operations/server-configuration-parameters/settings.md#default_replica_path) and [default_replica_name](../../operations/server-configuration-parameters/settings.md#default_replica_name). Macro `{uuid}` is unfolded to table's uuid, `{shard}` and `{replica}` are unfolded to values from server config, not from database engine arguments. But in the future, it will be possible to use `shard_name` and `replica_name` of Replicated database. -::: ## Specifics and Recommendations {#specifics-and-recommendations} diff --git a/docs/en/engines/table-engines/special/generate.md b/docs/en/engines/table-engines/special/generate.md index d03d6dc9d13..e42429a1b10 100644 --- a/docs/en/engines/table-engines/special/generate.md +++ b/docs/en/engines/table-engines/special/generate.md @@ -15,7 +15,7 @@ Usage examples: ## Usage in ClickHouse Server {#usage-in-clickhouse-server} ``` sql -ENGINE = GenerateRandom(random_seed, max_string_length, max_array_length) +ENGINE = GenerateRandom([random_seed] [,max_string_length] [,max_array_length]) ``` The `max_array_length` and `max_string_length` parameters specify maximum length of all diff --git a/docs/en/getting-started/example-datasets/cell-towers.md b/docs/en/getting-started/example-datasets/cell-towers.md index e74849a76e0..e31ce3de5ce 100644 --- a/docs/en/getting-started/example-datasets/cell-towers.md +++ b/docs/en/getting-started/example-datasets/cell-towers.md @@ -13,7 +13,7 @@ OpenCelliD Project is licensed under a Creative Commons Attribution-ShareAlike 4 ## Get the Dataset {#get-the-dataset} -1. Download the snapshot of the dataset from February 2021: [https://datasets.clickhouse.com/cell_towers.csv.xz] (729 MB). +1. Download the snapshot of the dataset from February 2021: [cell_towers.csv.xz](https://datasets.clickhouse.com/cell_towers.csv.xz) (729 MB). 2. Validate the integrity (optional step): ``` diff --git a/docs/en/interfaces/http.md b/docs/en/interfaces/http.md index 036fcde6d7a..c980bc65152 100644 --- a/docs/en/interfaces/http.md +++ b/docs/en/interfaces/http.md @@ -175,6 +175,10 @@ You can also choose to use [HTTP compression](https://en.wikipedia.org/wiki/HTTP - `br` - `deflate` - `xz` +- `zstd` +- `lz4` +- `bz2` +- `snappy` To send a compressed `POST` request, append the request header `Content-Encoding: compression_method`. In order for ClickHouse to compress the response, enable compression with [enable_http_compression](../operations/settings/settings.md#settings-enable_http_compression) setting and append `Accept-Encoding: compression_method` header to the request. You can configure the data compression level in the [http_zlib_compression_level](../operations/settings/settings.md#settings-http_zlib_compression_level) setting for all compression methods. diff --git a/docs/en/operations/access-rights.md b/docs/en/operations/access-rights.md index 1919aa49ab9..fc39d8b6dc7 100644 --- a/docs/en/operations/access-rights.md +++ b/docs/en/operations/access-rights.md @@ -151,4 +151,3 @@ Management queries: By default, SQL-driven access control and account management is disabled for all users. You need to configure at least one user in the `users.xml` configuration file and set the value of the [access_management](../operations/settings/settings-users.md#access_management-user-setting) setting to 1. -[Original article](https://clickhouse.com/docs/en/operations/access_rights/) diff --git a/docs/en/operations/quotas.md b/docs/en/operations/quotas.md index f35bf44fcd0..05355e615fd 100644 --- a/docs/en/operations/quotas.md +++ b/docs/en/operations/quotas.md @@ -2,10 +2,9 @@ slug: /en/operations/quotas sidebar_position: 51 sidebar_label: Quotas +title: Quotas --- -# Quotas - Quotas allow you to limit resource usage over a period of time or track the use of resources. Quotas are set up in the user config, which is usually ‘users.xml’. @@ -118,4 +117,3 @@ For distributed query processing, the accumulated amounts are stored on the requ When the server is restarted, quotas are reset. -[Original article](https://clickhouse.com/docs/en/operations/quotas/) diff --git a/docs/en/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-lifetime.md b/docs/en/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-lifetime.md index 433300eefa4..6e4c8c4b94e 100644 --- a/docs/en/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-lifetime.md +++ b/docs/en/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-lifetime.md @@ -94,6 +94,21 @@ It is also possible for `Flat`, `Hashed`, `ComplexKeyHashed` dictionaries to onl - If the source is HTTP then `update_field` will be added as a query parameter with the last update time as the parameter value. - If the source is Executable then `update_field` will be added as an executable script argument with the last update time as the argument value. - If the source is ClickHouse, MySQL, PostgreSQL, ODBC there will be an additional part of `WHERE`, where `update_field` is compared as greater or equal with the last update time. + - Per default, this `WHERE`-condition is checked at the highest level of the SQL-Query. Alternatively, the condition can be checked in any other `WHERE`-clause within the query using the `{condition}`-keyword. Example: + ```sql + ... + SOURCE(CLICKHOUSE(... + update_field 'added_time' + QUERY ' + SELECT my_arr.1 AS x, my_arr.2 AS y, creation_time + FROM ( + SELECT arrayZip(x_arr, y_arr) AS my_arr, creation_time + FROM dictionary_source + WHERE {condition} + )' + )) + ... + ``` If `update_field` option is set, additional option `update_lag` can be set. Value of `update_lag` option is subtracted from previous update time before request updated data. diff --git a/docs/en/sql-reference/functions/date-time-functions.md b/docs/en/sql-reference/functions/date-time-functions.md index 52f9a06df72..3515e903adf 100644 --- a/docs/en/sql-reference/functions/date-time-functions.md +++ b/docs/en/sql-reference/functions/date-time-functions.md @@ -267,7 +267,7 @@ Result: └────────────────┘ ``` -:::Attention +:::note The return type of `toStartOf*`, `toLastDayOfMonth`, `toMonday` functions described below is `Date` or `DateTime`. Though these functions can take values of the extended types `Date32` and `DateTime64` as an argument, passing them a time outside the normal range (year 1970 to 2149 for `Date` / 2106 for `DateTime`) will produce wrong results. In case argument is out of normal range: diff --git a/docs/en/sql-reference/functions/tuple-map-functions.md b/docs/en/sql-reference/functions/tuple-map-functions.md index c555a838927..52023df4d72 100644 --- a/docs/en/sql-reference/functions/tuple-map-functions.md +++ b/docs/en/sql-reference/functions/tuple-map-functions.md @@ -430,5 +430,119 @@ Result: └────────────────────────────┘ ``` +## mapApply + +**Syntax** + +```sql +mapApply(func, map) +``` + +**Parameters** + +- `func` - [Lamda function](../../sql-reference/functions/index.md#higher-order-functions---operator-and-lambdaparams-expr-function). +- `map` — [Map](../../sql-reference/data-types/map.md). + +**Returned value** + +- Returns a map obtained from the original map by application of `func(map1[i], …, mapN[i])` for each element. + +**Example** + +Query: + +```sql +SELECT mapApply((k, v) -> (k, v * 10), _map) AS r +FROM +( + SELECT map('key1', number, 'key2', number * 2) AS _map + FROM numbers(3) +) +``` + +Result: + +```text +┌─r─────────────────────┐ +│ {'key1':0,'key2':0} │ +│ {'key1':10,'key2':20} │ +│ {'key1':20,'key2':40} │ +└───────────────────────┘ +``` + +## mapFilter + +**Syntax** + +```sql +mapFilter(func, map) +``` + +**Parameters** + +- `func` - [Lamda function](../../sql-reference/functions/index.md#higher-order-functions---operator-and-lambdaparams-expr-function). +- `map` — [Map](../../sql-reference/data-types/map.md). + +**Returned value** + +- Returns a map containing only the elements in `map` for which `func(map1[i], …, mapN[i])` returns something other than 0. + + +**Example** + +Query: + +```sql +SELECT mapFilter((k, v) -> ((v % 2) = 0), _map) AS r +FROM +( + SELECT map('key1', number, 'key2', number * 2) AS _map + FROM numbers(3) +) +``` + +Result: + +```text +┌─r───────────────────┐ +│ {'key1':0,'key2':0} │ +│ {'key2':2} │ +│ {'key1':2,'key2':4} │ +└─────────────────────┘ +``` + + +## mapUpdate + +**Syntax** + +```sql +mapUpdate(map1, map2) +``` + +**Parameters** + +- `map1` [Map](../../sql-reference/data-types/map.md). +- `map2` [Map](../../sql-reference/data-types/map.md). + +**Returned value** + +- Returns a map1 with values updated of values for the corresponding keys in map2. + +**Example** + +Query: + +```sql +SELECT mapUpdate(map('key1', 0, 'key3', 0), map('key1', 10, 'key2', 10)) AS map; +``` + +Result: + +```text +┌─map────────────────────────────┐ +│ {'key3':0,'key1':10,'key2':10} │ +└────────────────────────────────┘ +``` [Original article](https://clickhouse.com/docs/en/sql-reference/functions/tuple-map-functions/) diff --git a/docs/en/sql-reference/statements/alter/constraint.md b/docs/en/sql-reference/statements/alter/constraint.md index 15bd27e1a95..844b24d7374 100644 --- a/docs/en/sql-reference/statements/alter/constraint.md +++ b/docs/en/sql-reference/statements/alter/constraint.md @@ -9,8 +9,8 @@ sidebar_label: CONSTRAINT Constraints could be added or deleted using following syntax: ``` sql -ALTER TABLE [db].name ADD CONSTRAINT constraint_name CHECK expression; -ALTER TABLE [db].name DROP CONSTRAINT constraint_name; +ALTER TABLE [db].name [ON CLUSTER cluster] ADD CONSTRAINT constraint_name CHECK expression; +ALTER TABLE [db].name [ON CLUSTER cluster] DROP CONSTRAINT constraint_name; ``` See more on [constraints](../../../sql-reference/statements/create/table.md#constraints). diff --git a/docs/en/sql-reference/statements/alter/ttl.md b/docs/en/sql-reference/statements/alter/ttl.md index 3e9846ba1ab..a312e8cad91 100644 --- a/docs/en/sql-reference/statements/alter/ttl.md +++ b/docs/en/sql-reference/statements/alter/ttl.md @@ -11,7 +11,7 @@ sidebar_label: TTL You can change [table TTL](../../../engines/table-engines/mergetree-family/mergetree.md#mergetree-table-ttl) with a request of the following form: ``` sql -ALTER TABLE table_name MODIFY TTL ttl_expression; +ALTER TABLE [db.]table_name [ON CLUSTER cluster] MODIFY TTL ttl_expression; ``` ## REMOVE TTL @@ -19,7 +19,7 @@ ALTER TABLE table_name MODIFY TTL ttl_expression; TTL-property can be removed from table with the following query: ```sql -ALTER TABLE table_name REMOVE TTL +ALTER TABLE [db.]table_name [ON CLUSTER cluster] REMOVE TTL ``` **Example** diff --git a/docs/en/sql-reference/statements/show.md b/docs/en/sql-reference/statements/show.md index 0721f17e9e2..00347d9cb5b 100644 --- a/docs/en/sql-reference/statements/show.md +++ b/docs/en/sql-reference/statements/show.md @@ -303,7 +303,7 @@ SHOW USERS ## SHOW ROLES -Returns a list of [roles](../../operations/access-rights.md#role-management). To view another parameters, see system tables [system.roles](../../operations/system-tables/roles.md#system_tables-roles) and [system.role-grants](../../operations/system-tables/role-grants.md#system_tables-role_grants). +Returns a list of [roles](../../operations/access-rights.md#role-management). To view another parameters, see system tables [system.roles](../../operations/system-tables/roles.md#system_tables-roles) and [system.role_grants](../../operations/system-tables/role-grants.md#system_tables-role_grants). ### Syntax diff --git a/docs/ru/sql-reference/functions/date-time-functions.md b/docs/ru/sql-reference/functions/date-time-functions.md index 63386bb32b9..1c623cd1dab 100644 --- a/docs/ru/sql-reference/functions/date-time-functions.md +++ b/docs/ru/sql-reference/functions/date-time-functions.md @@ -267,7 +267,7 @@ SELECT toUnixTimestamp('2017-11-05 08:07:47', 'Asia/Tokyo') AS unix_timestamp; └────────────────┘ ``` -:::Attention +:::note Тип возвращаемого описанными далее функциями `toStartOf*`, `toMonday` значения - `Date` или `DateTime`. Хотя эти функции могут принимать значения типа `Date32` или `DateTime64` в качестве аргумента, при обработке аргумента вне нормального диапазона значений (`1970` - `2148` для `Date` и `1970-01-01 00:00:00`-`2106-02-07 08:28:15` для `DateTime`) будет получен некорректный результат. Возвращаемые значения для значений вне нормального диапазона: @@ -277,7 +277,7 @@ SELECT toUnixTimestamp('2017-11-05 08:07:47', 'Asia/Tokyo') AS unix_timestamp; * `2149-05-31` будет результатом функции `toLastDayOfMonth` при обработке аргумента больше `2149-05-31`. ::: -:::Attention +:::note Тип возвращаемого описанными далее функциями `toStartOf*`, `toLastDayOfMonth`, `toMonday` значения - `Date` или `DateTime`. Хотя эти функции могут принимать значения типа `Date32` или `DateTime64` в качестве аргумента, при обработке аргумента вне нормального диапазона значений (`1970` - `2148` для `Date` и `1970-01-01 00:00:00`-`2106-02-07 08:28:15` для `DateTime`) будет получен некорректный результат. Возвращаемые значения для значений вне нормального диапазона: diff --git a/docs/ru/sql-reference/statements/alter/constraint.md b/docs/ru/sql-reference/statements/alter/constraint.md index cb380bc6a60..bc65b0bbf27 100644 --- a/docs/ru/sql-reference/statements/alter/constraint.md +++ b/docs/ru/sql-reference/statements/alter/constraint.md @@ -11,8 +11,8 @@ sidebar_label: "Манипуляции с ограничениями" Добавить или удалить ограничение можно с помощью запросов ``` sql -ALTER TABLE [db].name ADD CONSTRAINT constraint_name CHECK expression; -ALTER TABLE [db].name DROP CONSTRAINT constraint_name; +ALTER TABLE [db].name [ON CLUSTER cluster] ADD CONSTRAINT constraint_name CHECK expression; +ALTER TABLE [db].name [ON CLUSTER cluster] DROP CONSTRAINT constraint_name; ``` Запросы выполняют добавление или удаление метаданных об ограничениях таблицы `[db].name`, поэтому выполняются мгновенно. diff --git a/docs/ru/sql-reference/statements/alter/ttl.md b/docs/ru/sql-reference/statements/alter/ttl.md index 855a651ffc2..2917e96fd34 100644 --- a/docs/ru/sql-reference/statements/alter/ttl.md +++ b/docs/ru/sql-reference/statements/alter/ttl.md @@ -11,7 +11,7 @@ sidebar_label: TTL Вы можете изменить [TTL для таблицы](../../../engines/table-engines/mergetree-family/mergetree.md#mergetree-column-ttl) запросом следующего вида: ``` sql -ALTER TABLE table-name MODIFY TTL ttl-expression +ALTER TABLE [db.]table-name [ON CLUSTER cluster] MODIFY TTL ttl-expression ``` ## REMOVE TTL {#remove-ttl} @@ -19,7 +19,7 @@ ALTER TABLE table-name MODIFY TTL ttl-expression Удалить табличный TTL можно запросом следующего вида: ```sql -ALTER TABLE table_name REMOVE TTL +ALTER TABLE [db.]table_name [ON CLUSTER cluster] REMOVE TTL ``` **Пример** @@ -83,4 +83,4 @@ SELECT * FROM table_with_ttl; ### Смотрите также - Подробнее о [свойстве TTL](../../../engines/table-engines/mergetree-family/mergetree.md#mergetree-column-ttl). -- Изменить столбец [с TTL](../../../sql-reference/statements/alter/column.md#alter_modify-column). \ No newline at end of file +- Изменить столбец [с TTL](../../../sql-reference/statements/alter/column.md#alter_modify-column). diff --git a/docs/ru/sql-reference/statements/show.md b/docs/ru/sql-reference/statements/show.md index 1d072c9d5de..59f33c691ae 100644 --- a/docs/ru/sql-reference/statements/show.md +++ b/docs/ru/sql-reference/statements/show.md @@ -305,7 +305,7 @@ SHOW USERS ## SHOW ROLES {#show-roles-statement} -Выводит список [ролей](../../operations/access-rights.md#role-management). Для просмотра параметров ролей, см. системные таблицы [system.roles](../../operations/system-tables/roles.md#system_tables-roles) и [system.role-grants](../../operations/system-tables/role-grants.md#system_tables-role_grants). +Выводит список [ролей](../../operations/access-rights.md#role-management). Для просмотра параметров ролей, см. системные таблицы [system.roles](../../operations/system-tables/roles.md#system_tables-roles) и [system.role_grants](../../operations/system-tables/role-grants.md#system_tables-role_grants). ### Синтаксис {#show-roles-syntax} diff --git a/docs/zh/development/tests.md b/docs/zh/development/tests.md index ca9300597c7..e6d5cf66de9 100644 --- a/docs/zh/development/tests.md +++ b/docs/zh/development/tests.md @@ -1,338 +1,297 @@ --- -slug: /zh/development/tests +slug: /en/development/tests +sidebar_position: 70 +sidebar_label: Testing +title: ClickHouse Testing +description: Most of ClickHouse features can be tested with functional tests and they are mandatory to use for every change in ClickHouse code that can be tested that way. --- -# ClickHouse 测试 {#clickhouse-testing} -## 功能测试 {#functional-tests} +## Functional Tests -功能测试使用起来最简单方便. 大多数 ClickHouse 特性都可以通过功能测试进行测试, 并且对于可以通过功能测试进行测试的 ClickHouse 代码的每一个更改, 都必须使用这些特性 +Functional tests are the most simple and convenient to use. Most of ClickHouse features can be tested with functional tests and they are mandatory to use for every change in ClickHouse code that can be tested that way. -每个功能测试都会向正在运行的 ClickHouse 服务器发送一个或多个查询, 并将结果与参考进行比较. +Each functional test sends one or multiple queries to the running ClickHouse server and compares the result with reference. -测试位于 `查询` 目录中. 有两个子目录: `无状态` 和 `有状态`. 无状态测试在没有任何预加载测试数据的情况下运行查询 - 它们通常在测试本身内即时创建小型合成数据集. 状态测试需要来自 Yandex.Metrica 的预加载测试数据, 它对公众开放. +Tests are located in `queries` directory. There are two subdirectories: `stateless` and `stateful`. Stateless tests run queries without any preloaded test data - they often create small synthetic datasets on the fly, within the test itself. Stateful tests require preloaded test data from ClickHouse and it is available to general public. -每个测试可以是两种类型之一: `.sql` 和 `.sh`. `.sql` 测试是简单的 SQL 脚本, 它通过管道传输到 `clickhouse-client --multiquery --testmode`. `.sh` 测试是一个自己运行的脚本. SQL 测试通常比 `.sh` 测试更可取. 仅当您必须测试某些无法从纯 SQL 中执行的功能时才应使用 `.sh` 测试, 例如将一些输入数据传送到 `clickhouse-client` 或测试 `clickhouse-local`. +Each test can be one of two types: `.sql` and `.sh`. `.sql` test is the simple SQL script that is piped to `clickhouse-client --multiquery`. `.sh` test is a script that is run by itself. SQL tests are generally preferable to `.sh` tests. You should use `.sh` tests only when you have to test some feature that cannot be exercised from pure SQL, such as piping some input data into `clickhouse-client` or testing `clickhouse-local`. -### 在本地运行测试 {#functional-test-locally} +### Running a Test Locally {#functional-test-locally} -在本地启动ClickHouse服务器, 监听默认端口(9000). 例如, 要运行测试 `01428_hash_set_nan_key`, 请切换到存储库文件夹并运行以下命令: +Start the ClickHouse server locally, listening on the default port (9000). To +run, for example, the test `01428_hash_set_nan_key`, change to the repository +folder and run the following command: ``` PATH=$PATH: tests/clickhouse-test 01428_hash_set_nan_key ``` -有关更多选项, 请参阅`tests/clickhouse-test --help`. 您可以简单地运行所有测试或运行由测试名称中的子字符串过滤的测试子集：`./clickhouse-test substring`. 还有并行或随机顺序运行测试的选项. +For more options, see `tests/clickhouse-test --help`. You can simply run all tests or run subset of tests filtered by substring in test name: `./clickhouse-test substring`. There are also options to run tests in parallel or in randomized order. -### 添加新测试 {#adding-new-test} +### Adding a New Test -添加新的测试, 在 `queries/0_stateless` 目录下创建 `.sql` 或 `.sh` 文件, 手动检查, 然后通过以下方式生成`.reference`文件：`clickhouse-client -n --testmode < 00000_test.sql > 00000_test.reference` 或 `./00000_test.sh > ./00000_test.reference`. +To add new test, create a `.sql` or `.sh` file in `queries/0_stateless` directory, check it manually and then generate `.reference` file in the following way: `clickhouse-client --multiquery < 00000_test.sql > 00000_test.reference` or `./00000_test.sh > ./00000_test.reference`. -测试应仅使用(创建、删除等)`test` 数据库中假定已预先创建的表; 测试也可以使用临时表. +Tests should use (create, drop, etc) only tables in `test` database that is assumed to be created beforehand; also tests can use temporary tables. -### 选择测试名称 {#choosing-test-name} +### Choosing the Test Name -测试名称以五位数前缀开头, 后跟描述性名称, 例如 `00422_hash_function_constexpr.sql`. 要选择前缀, 请找到目录中已存在的最大前缀, 并将其加一. 在此期间, 可能会添加一些具有相同数字前缀的其他测试, 但这没关系并且不会导致任何问题, 您以后不必更改它. +The name of the test starts with a five-digit prefix followed by a descriptive name, such as `00422_hash_function_constexpr.sql`. To choose the prefix, find the largest prefix already present in the directory, and increment it by one. In the meantime, some other tests might be added with the same numeric prefix, but this is OK and does not lead to any problems, you don't have to change it later. -一些测试的名称中标有 `zookeeper`、`shard` 或 `long` . `zookeeper` 用于使用 ZooKeeper 的测试. `shard` 用于需要服务器监听 `127.0.0.*` 的测试; `distributed` 或 `global` 具有相同的含义. `long` 用于运行时间稍长于一秒的测试. Yo你可以分别使用 `--no-zookeeper`、`--no-shard` 和 `--no-long` 选项禁用这些测试组. 如果需要 ZooKeeper 或分布式查询，请确保为您的测试名称添加适当的前缀. +Some tests are marked with `zookeeper`, `shard` or `long` in their names. `zookeeper` is for tests that are using ZooKeeper. `shard` is for tests that requires server to listen `127.0.0.*`; `distributed` or `global` have the same meaning. `long` is for tests that run slightly longer that one second. You can disable these groups of tests using `--no-zookeeper`, `--no-shard` and `--no-long` options, respectively. Make sure to add a proper prefix to your test name if it needs ZooKeeper or distributed queries. -### 检查必须发生的错误 {#checking-error-must-occur} +### Checking for an Error that Must Occur -有时您想测试是否因不正确的查询而发生服务器错误. 我们支持在 SQL 测试中对此进行特殊注释, 形式如下: +Sometimes you want to test that a server error occurs for an incorrect query. We support special annotations for this in SQL tests, in the following form: ``` select x; -- { serverError 49 } ``` -此测试确保服务器返回关于未知列“x”的错误代码为 49. 如果没有错误, 或者错误不同, 则测试失败. 如果您想确保错误发生在客户端, 请改用 `clientError` 注释. +This test ensures that the server returns an error with code 49 about unknown column `x`. If there is no error, or the error is different, the test will fail. If you want to ensure that an error occurs on the client side, use `clientError` annotation instead. -不要检查错误消息的特定措辞, 它将来可能会发生变化, 并且测试将不必要地中断. 只检查错误代码. 如果现有的错误代码不足以满足您的需求, 请考虑添加一个新的. +Do not check for a particular wording of error message, it may change in the future, and the test will needlessly break. Check only the error code. If the existing error code is not precise enough for your needs, consider adding a new one. -### 测试分布式查询 {#testing-distributed-query} +### Testing a Distributed Query -如果你想在功能测试中使用分布式查询, 你可以使用 `127.0.0.{1..2}` 的地址, 以便服务器查询自己; 或者您可以在服务器配置文件中使用预定义的测试集群, 例如`test_shard_localhost`. 请记住在测试名称中添加 `shard` 或 `distributed` 字样, 以便它以正确的配置在 CI 中运行, 其中服务器配置为支持分布式查询. +If you want to use distributed queries in functional tests, you can leverage `remote` table function with `127.0.0.{1..2}` addresses for the server to query itself; or you can use predefined test clusters in server configuration file like `test_shard_localhost`. Remember to add the words `shard` or `distributed` to the test name, so that it is run in CI in correct configurations, where the server is configured to support distributed queries. -## 已知错误 {#known-bugs} +## Known Bugs {#known-bugs} -如果我们知道一些可以通过功能测试轻松重现的错误, 我们将准备好的功能测试放在 `tests/queries/bugs` 目录中. 修复错误后, 这些测试将移至 `tests/queries/0_stateless` . +If we know some bugs that can be easily reproduced by functional tests, we place prepared functional tests in `tests/queries/bugs` directory. These tests will be moved to `tests/queries/0_stateless` when bugs are fixed. -## 集成测试 {#integration-tests} +## Integration Tests {#integration-tests} -集成测试允许在集群配置中测试 ClickHouse 以及 ClickHouse 与其他服务器(如 MySQL、Postgres、MongoDB)的交互. 它们可以用来模拟网络分裂、丢包等情况. 这些测试在Docker下运行, 并使用各种软件创建多个容器. +Integration tests allow testing ClickHouse in clustered configuration and ClickHouse interaction with other servers like MySQL, Postgres, MongoDB. They are useful to emulate network splits, packet drops, etc. These tests are run under Docker and create multiple containers with various software. -有关如何运行这些测试, 请参阅 `tests/integration/README.md` . +See `tests/integration/README.md` on how to run these tests. -注意, ClickHouse与第三方驱动程序的集成没有经过测试. 另外, 我们目前还没有JDBC和ODBC驱动程序的集成测试. +Note that integration of ClickHouse with third-party drivers is not tested. Also, we currently do not have integration tests with our JDBC and ODBC drivers. -## 单元测试 {#unit-tests} +## Unit Tests {#unit-tests} -当您想测试的不是 ClickHouse 整体, 而是单个独立库或类时，单元测试很有用. 您可以使用 `ENABLE_TESTS` CMake 选项启用或禁用测试构建. 单元测试(和其他测试程序)位于代码中的 `tests` 子目录中. 要运行单元测试, 请键入 `ninja test` 。有些测试使用 `gtest` , 但有些程序在测试失败时会返回非零退出码. +Unit tests are useful when you want to test not the ClickHouse as a whole, but a single isolated library or class. You can enable or disable build of tests with `ENABLE_TESTS` CMake option. Unit tests (and other test programs) are located in `tests` subdirectories across the code. To run unit tests, type `ninja test`. Some tests use `gtest`, but some are just programs that return non-zero exit code on test failure. -如果代码已经被功能测试覆盖了, 就没有必要进行单元测试(而且功能测试通常更易于使用). +It’s not necessary to have unit tests if the code is already covered by functional tests (and functional tests are usually much more simple to use). -例如, 您可以通过直接调用可执行文件来运行单独的 gtest 检查: +You can run individual gtest checks by calling the executable directly, for example: ```bash $ ./src/unit_tests_dbms --gtest_filter=LocalAddress* ``` -## 性能测试 {#performance-tests} +## Performance Tests {#performance-tests} -性能测试允许测量和比较 ClickHouse 的某些孤立部分在合成查询上的性能. 测试位于 `tests/performance`. 每个测试都由带有测试用例描述的 `.xml` 文件表示. 测试使用 `docker/tests/performance-comparison` 工具运行. 请参阅自述文件以进行调用. +Performance tests allow to measure and compare performance of some isolated part of ClickHouse on synthetic queries. Performance tests are located at `tests/performance/`. Each test is represented by an `.xml` file with a description of the test case. Tests are run with `docker/test/performance-comparison` tool . See the readme file for invocation. -每个测试在循环中运行一个或多个查询(可能带有参数组合). 一些测试可以包含预加载测试数据集的先决条件. +Each test run one or multiple queries (possibly with combinations of parameters) in a loop. -如果您希望在某些场景中提高ClickHouse的性能，并且如果可以在简单的查询中观察到改进，那么强烈建议编写性能测试。在测试期间使用 `perf top` 或其他perf工具总是有意义的. +If you want to improve performance of ClickHouse in some scenario, and if improvements can be observed on simple queries, it is highly recommended to write a performance test. Also, it is recommended to write performance tests when you add or modify SQL functions which are relatively isolated and not too obscure. It always makes sense to use `perf top` or other `perf` tools during your tests. -## 测试工具和脚本 {#test-tools-and-scripts} +## Test Tools and Scripts {#test-tools-and-scripts} - `tests` 目录中的一些程序不是准备好的测试，而是测试工具. 例如, 对于 `Lexer`, 有一个工具 `src/Parsers/tests/lexer` , 它只是对标准输入进行标记化并将着色结果写入标准输出. 您可以将这些类型的工具用作代码示例以及用于探索和手动测试. +Some programs in `tests` directory are not prepared tests, but are test tools. For example, for `Lexer` there is a tool `src/Parsers/tests/lexer` that just do tokenization of stdin and writes colorized result to stdout. You can use these kind of tools as a code examples and for exploration and manual testing. -## 其他测试 {#miscellaneous-tests} +## Miscellaneous Tests {#miscellaneous-tests} -在 `tests/external_models` 中有机器学习模型的测试. 这些测试不会更新, 必须转移到集成测试. +There are tests for machine learned models in `tests/external_models`. These tests are not updated and must be transferred to integration tests. -仲裁插入有单独的测试. 该测试在不同的服务器上运行 ClickHouse 集群并模拟各种故障情况：网络分裂、丢包(ClickHouse 节点之间、ClickHouse 和 ZooKeeper 之间、ClickHouse 服务器和客户端之间等)、`kill -9`、`kill -STOP` 和 `kill -CONT` , 比如 [Jepsen](https://aphyr.com/tags/Jepsen). 然后测试检查所有已确认的插入是否已写入并且所有被拒绝的插入均未写入. +There is separate test for quorum inserts. This test run ClickHouse cluster on separate servers and emulate various failure cases: network split, packet drop (between ClickHouse nodes, between ClickHouse and ZooKeeper, between ClickHouse server and client, etc.), `kill -9`, `kill -STOP` and `kill -CONT` , like [Jepsen](https://aphyr.com/tags/Jepsen). Then the test checks that all acknowledged inserts was written and all rejected inserts was not. -在 ClickHouse 开源之前, Quorum 测试是由单独的团队编写的. 这个团队不再与ClickHouse合作. 测试碰巧是用Java编写的. 由于这些原因, 必须重写仲裁测试并将其转移到集成测试. +Quorum test was written by separate team before ClickHouse was open-sourced. This team no longer work with ClickHouse. Test was accidentally written in Java. For these reasons, quorum test must be rewritten and moved to integration tests. -## 手动测试 {#manual-testing} +## Manual Testing {#manual-testing} -当您开发一个新特性时, 手动测试它也是合理的. 您可以按照以下步骤进行操作: +When you develop a new feature, it is reasonable to also test it manually. You can do it with the following steps: -构建 ClickHouse. 从终端运行 ClickHouse：将目录更改为 `programs/clickhouse-server` 并使用 `./clickhouse-server` 运行它. 默认情况下, 它将使用当前目录中的配置(`config.xml`、`users.xml` 和`config.d` 和`users.d` 目录中的文件). 要连接到 ClickHouse 服务器, 请运行 `programs/clickhouse-client/clickhouse-client` . +Build ClickHouse. Run ClickHouse from the terminal: change directory to `programs/clickhouse-server` and run it with `./clickhouse-server`. It will use configuration (`config.xml`, `users.xml` and files within `config.d` and `users.d` directories) from the current directory by default. To connect to ClickHouse server, run `programs/clickhouse-client/clickhouse-client`. -请注意, 所有 clickhouse 工具(服务器、客户端等)都只是指向名为 `clickhouse` 的单个二进制文件的符号链接. 你可以在 `programs/clickhouse` 找到这个二进制文件. 所有工具也可以作为 `clickhouse tool` 而不是 `clickhouse-tool` 调用. +Note that all clickhouse tools (server, client, etc) are just symlinks to a single binary named `clickhouse`. You can find this binary at `programs/clickhouse`. All tools can also be invoked as `clickhouse tool` instead of `clickhouse-tool`. -或者, 您可以安装 ClickHouse 包: 从 Yandex 存储库稳定发布, 或者您可以在 ClickHouse 源根目录中使用 `./release` 为自己构建包. 然后使用 `sudo service clickhouse-server start` 启动服务器(或停止以停止服务器). 在 `/etc/clickhouse-server/clickhouse-server.log` 中查找日志. +Alternatively you can install ClickHouse package: either stable release from ClickHouse repository or you can build package for yourself with `./release` in ClickHouse sources root. Then start the server with `sudo clickhouse start` (or stop to stop the server). Look for logs at `/etc/clickhouse-server/clickhouse-server.log`. -当您的系统上已经安装了 ClickHouse 时，您可以构建一个新的 `clickhouse` 二进制文件并替换现有的二进制文件: +When ClickHouse is already installed on your system, you can build a new `clickhouse` binary and replace the existing binary: ``` bash -$ sudo service clickhouse-server stop +$ sudo clickhouse stop $ sudo cp ./clickhouse /usr/bin/ -$ sudo service clickhouse-server start +$ sudo clickhouse start ``` -您也可以停止系统 clickhouse-server 并使用相同的配置运行您自己的服务器, 但登录到终端: +Also you can stop system clickhouse-server and run your own with the same configuration but with logging to terminal: ``` bash -$ sudo service clickhouse-server stop +$ sudo clickhouse stop $ sudo -u clickhouse /usr/bin/clickhouse server --config-file /etc/clickhouse-server/config.xml ``` -使用 gdb 的示例: +Example with gdb: ``` bash $ sudo -u clickhouse gdb --args /usr/bin/clickhouse server --config-file /etc/clickhouse-server/config.xml ``` -如果系统 clickhouse-server 已经在运行并且你不想停止它, 你可以在你的 `config.xml` 中更改端口号(或在 `config.d` 目录中的文件中覆盖它们), 提供适当的数据路径, 并运行它. +If the system clickhouse-server is already running and you do not want to stop it, you can change port numbers in your `config.xml` (or override them in a file in `config.d` directory), provide appropriate data path, and run it. -`clickhouse` 二进制文件几乎没有依赖关系, 可以在广泛的 Linux 发行版中使用. 要在服务器上快速而肮脏地测试您的更改, 您可以简单地将新构建的 `clickhouse` 二进制文件 `scp` 到您的服务器, 然后按照上面的示例运行它. +`clickhouse` binary has almost no dependencies and works across wide range of Linux distributions. To quick and dirty test your changes on a server, you can simply `scp` your fresh built `clickhouse` binary to your server and then run it as in examples above. -## 测试环境 {#testing-environment} +## Build Tests {#build-tests} -在发布稳定版之前, 我们将其部署在测试环境中.测试环境是一个集群，处理 [Yandex.Metrica](https://metrica.yandex.com/) 数据的 1/39 部分. 我们与 Yandex.Metrica 团队共享我们的测试环境. ClickHouse无需在现有数据上停机即可升级. 我们首先看到的是, 数据被成功地处理了, 没有滞后于实时, 复制继续工作, Yandex.Metrica 团队没有发现任何问题. 第一次检查可以通过以下方式进行: +Build tests allow to check that build is not broken on various alternative configurations and on some foreign systems. These tests are automated as well. -``` sql -SELECT hostName() AS h, any(version()), any(uptime()), max(UTCEventTime), count() FROM remote('example01-01-{1..3}t', merge, hits) WHERE EventDate >= today() - 2 GROUP BY h ORDER BY h; -``` +Examples: +- cross-compile for Darwin x86_64 (Mac OS X) +- cross-compile for FreeBSD x86_64 +- cross-compile for Linux AArch64 +- build on Ubuntu with libraries from system packages (discouraged) +- build with shared linking of libraries (discouraged) -在某些情况下, 我们还会部署到 Yandex 中我们朋友团队的测试环境：Market、Cloud 等. 此外, 我们还有一些用于开发目的的硬件服务器. +For example, build with system packages is bad practice, because we cannot guarantee what exact version of packages a system will have. But this is really needed by Debian maintainers. For this reason we at least have to support this variant of build. Another example: shared linking is a common source of trouble, but it is needed for some enthusiasts. -## 负载测试 {#load-testing} +Though we cannot run all tests on all variant of builds, we want to check at least that various build variants are not broken. For this purpose we use build tests. -部署到测试环境后, 我们使用来自生产集群的查询运行负载测试. 这是手动完成的. +We also test that there are no translation units that are too long to compile or require too much RAM. -确保您在生产集群上启用了 `query_log`. +We also test that there are no too large stack frames. -收集一天或更长时间的查询日志: +## Testing for Protocol Compatibility {#testing-for-protocol-compatibility} -``` bash -$ clickhouse-client --query="SELECT DISTINCT query FROM system.query_log WHERE event_date = today() AND query LIKE '%ym:%' AND query NOT LIKE '%system.query_log%' AND type = 2 AND is_initial_query" > queries.tsv -``` +When we extend ClickHouse network protocol, we test manually that old clickhouse-client works with new clickhouse-server and new clickhouse-client works with old clickhouse-server (simply by running binaries from corresponding packages). -这是一个复杂的例子. `type = 2` 将过滤成功执行的查询. `query LIKE '%ym:%'` 是从 Yandex.Metrica 中选择相关查询. `is_initial_query` 是只选择客户端发起的查询, 而不是 ClickHouse 本身(作为分布式查询处理的一部分). +We also test some cases automatically with integrational tests: +- if data written by old version of ClickHouse can be successfully read by the new version; +- do distributed queries work in a cluster with different ClickHouse versions. -`scp` 将此日志记录到您的测试集群并按如下方式运行它: +## Help from the Compiler {#help-from-the-compiler} -``` bash -$ clickhouse benchmark --concurrency 16 < queries.tsv -``` +Main ClickHouse code (that is located in `dbms` directory) is built with `-Wall -Wextra -Werror` and with some additional enabled warnings. Although these options are not enabled for third-party libraries. -(可能你还想指定一个 `--user`) +Clang has even more useful warnings - you can look for them with `-Weverything` and pick something to default build. -然后把它留到晚上或周末, 去休息一下. +For production builds, clang is used, but we also test make gcc builds. For development, clang is usually more convenient to use. You can build on your own machine with debug mode (to save battery of your laptop), but please note that compiler is able to generate more warnings with `-O3` due to better control flow and inter-procedure analysis. When building with clang in debug mode, debug version of `libc++` is used that allows to catch more errors at runtime. -您应该检查 `clickhouse-server` 没有崩溃, 内存占用是有限的, 且性能不会随着时间的推移而降低. +## Sanitizers {#sanitizers} -由于查询和环境的高度可变性, 没有记录和比较精确的查询执行时间. +### Address sanitizer +We run functional, integration, stress and unit tests under ASan on per-commit basis. -## 构建测试 {#build-tests} +### Thread sanitizer +We run functional, integration, stress and unit tests under TSan on per-commit basis. -构建测试允许检查在各种可选配置和一些外部系统上的构建是否被破坏. 这些测试也是自动化的. +### Memory sanitizer +We run functional, integration, stress and unit tests under MSan on per-commit basis. -示例: -- Darwin x86_64 (Mac OS X) 交叉编译 -- FreeBSD x86_64 交叉编译 -- Linux AArch64 交叉编译 -- 使用系统包中的库在 Ubuntu 上构建（不鼓励） -- 使用库的共享链接构建（不鼓励） - -例如, 使用系统包构建是不好的做法, 因为我们无法保证系统将拥有哪个确切版本的包. 但这确实是 Debian 维护者所需要的. 出于这个原因, 我们至少必须支持这种构建变体. 另一个例子: 共享链接是一个常见的麻烦来源, 但对于一些爱好者来说是需要的. - -虽然我们无法对所有构建变体运行所有测试, 但我们希望至少检查各种构建变体没有被破坏. 为此, 我们使用构建测试. - -我们还测试了那些太长而无法编译或需要太多RAM的没有翻译单元. - -我们还测试没有太大的堆栈帧. - -## 协议兼容性测试 {#testing-for-protocol-compatibility} - -当我们扩展 ClickHouse 网络协议时, 我们手动测试旧的 clickhouse-client 与新的 clickhouse-server 一起工作, 而新的 clickhouse-client 与旧的 clickhouse-server 一起工作(只需从相应的包中运行二进制文件). - -我们还使用集成测试自动测试一些案例: -- 旧版本ClickHouse写入的数据是否可以被新版本成功读取; -- 在具有不同 ClickHouse 版本的集群中执行分布式查询. - -## 编译器的帮助 {#help-from-the-compiler} - -主要的 ClickHouse 代码(位于 `dbms` 目录中)是用 `-Wall -Wextra -Werror` 和一些额外的启用警告构建的. 虽然没有为第三方库启用这些选项. - -Clang 有更多有用的警告 - 你可以用 `-Weverything` 寻找它们并选择一些东西来默认构建. - -对于生产构建, 使用 clang, 但我们也测试 make gcc 构建. 对于开发, clang 通常使用起来更方便. 您可以使用调试模式在自己的机器上构建(以节省笔记本电脑的电池), 但请注意, 由于更好的控制流和过程间分析, 编译器能够使用 `-O3` 生成更多警告. 在调试模式下使用 clang 构建时, 使用调试版本的 `libc++` 允许在运行时捕获更多错误. - -## 地址清理器 {#sanitizers} - -### 地址清理器 -我们在ASan上运行功能测试、集成测试、压力测试和单元测试. - -### 线程清理器 -我们在TSan下运行功能测试、集成测试、压力测试和单元测试. - -### 内存清理器 -我们在MSan上运行功能测试、集成测试、压力测试和单元测试. - -### 未定义的行为清理器 -我们在UBSan下运行功能测试、集成测试、压力测试和单元测试. 某些第三方库的代码未针对 UB 进行清理. +### Undefined behaviour sanitizer +We run functional, integration, stress and unit tests under UBSan on per-commit basis. The code of some third-party libraries is not sanitized for UB. ### Valgrind (Memcheck) -我们曾经在 Valgrind 下通宵运行功能测试, 但不再这样做了. 这需要几个小时. 目前在`re2`库中有一个已知的误报, 见[这篇文章](https://research.swtch.com/sparse). +We used to run functional tests under Valgrind overnight, but don't do it anymore. It takes multiple hours. Currently there is one known false positive in `re2` library, see [this article](https://research.swtch.com/sparse). -## 模糊测试 {#fuzzing} +## Fuzzing {#fuzzing} -ClickHouse 模糊测试是使用 [libFuzzer](https://llvm.org/docs/LibFuzzer.html) 和随机 SQL 查询实现的. 所有模糊测试都应使用sanitizers(地址和未定义)进行. +ClickHouse fuzzing is implemented both using [libFuzzer](https://llvm.org/docs/LibFuzzer.html) and random SQL queries. +All the fuzz testing should be performed with sanitizers (Address and Undefined). -LibFuzzer 用于库代码的隔离模糊测试. Fuzzer 作为测试代码的一部分实现, 并具有 `_fuzzer` 名称后缀. -Fuzzer 示例可以在 `src/Parsers/tests/lexer_fuzzer.cpp` 中找到. LibFuzzer 特定的配置、字典和语料库存储在 `tests/fuzz` 中. -我们鼓励您为处理用户输入的每个功能编写模糊测试. +LibFuzzer is used for isolated fuzz testing of library code. Fuzzers are implemented as part of test code and have “_fuzzer” name postfixes. +Fuzzer example can be found at `src/Parsers/fuzzers/lexer_fuzzer.cpp`. LibFuzzer-specific configs, dictionaries and corpus are stored at `tests/fuzz`. +We encourage you to write fuzz tests for every functionality that handles user input. -默认情况下不构建模糊器. 要构建模糊器, 应设置` -DENABLE_FUZZING=1` 和 `-DENABLE_TESTS=1` 选项. -我们建议在构建模糊器时禁用 Jemalloc. 用于将 ClickHouse fuzzing 集成到 Google OSS-Fuzz 的配置可以在 `docker/fuzz` 中找到. +Fuzzers are not built by default. To build fuzzers both `-DENABLE_FUZZING=1` and `-DENABLE_TESTS=1` options should be set. +We recommend to disable Jemalloc while building fuzzers. Configuration used to integrate ClickHouse fuzzing to +Google OSS-Fuzz can be found at `docker/fuzz`. -我们还使用简单的模糊测试来生成随机SQL查询, 并检查服务器在执行这些查询时是否会死亡. -你可以在 `00746_sql_fuzzy.pl` 中找到它. 这个测试应该连续运行(通宵或更长时间). +We also use simple fuzz test to generate random SQL queries and to check that the server does not die executing them. +You can find it in `00746_sql_fuzzy.pl`. This test should be run continuously (overnight and longer). -我们还使用复杂的基于 AST 的查询模糊器, 它能够找到大量的极端情况. 它在查询 AST 中进行随机排列和替换. 它会记住先前测试中的 AST 节点, 以使用它们对后续测试进行模糊测试, 同时以随机顺序处理它们. 您可以在 [这篇博客文章](https://clickhouse.com/blog/en/2021/fuzzing-clickhouse/) 中了解有关此模糊器的更多信息. +We also use sophisticated AST-based query fuzzer that is able to find huge amount of corner cases. It does random permutations and substitutions in queries AST. It remembers AST nodes from previous tests to use them for fuzzing of subsequent tests while processing them in random order. You can learn more about this fuzzer in [this blog article](https://clickhouse.com/blog/en/2021/fuzzing-clickhouse/). -## 压力测试 {#stress-test} +## Stress test -压力测试是另一种模糊测试. 它使用单个服务器以随机顺序并行运行所有功能测试. 不检查测试结果. +Stress tests are another case of fuzzing. It runs all functional tests in parallel in random order with a single server. Results of the tests are not checked. -经检查: -- 服务器不会崩溃，不会触发调试或清理程序陷阱; -- 没有死锁; -- 数据库结构一致; -- 服务器可以在测试后成功停止并重新启动，没有异常; +It is checked that: +- server does not crash, no debug or sanitizer traps are triggered; +- there are no deadlocks; +- the database structure is consistent; +- server can successfully stop after the test and start again without exceptions. -有五种变体 (Debug, ASan, TSan, MSan, UBSan). +There are five variants (Debug, ASan, TSan, MSan, UBSan). -## 线程模糊器 {#thread-fuzzer} +## Thread Fuzzer -Thread Fuzzer(请不要与 Thread Sanitizer 混淆)是另一种允许随机化线程执行顺序的模糊测试. 它有助于找到更多特殊情况. +Thread Fuzzer (please don't mix up with Thread Sanitizer) is another kind of fuzzing that allows to randomize thread order of execution. It helps to find even more special cases. -## 安全审计 {#security-audit} +## Security Audit -Yandex安全团队的人员从安全的角度对ClickHouse的功能做了一些基本的概述. +Our Security Team did some basic overview of ClickHouse capabilities from the security standpoint. -## 静态分析仪 {#static-analyzers} +## Static Analyzers {#static-analyzers} -我们在每次提交的基础上运行 `clang-tidy`. `clang-static-analyzer` 检查也被启用. `clang-tidy` 也用于一些样式检查. +We run `clang-tidy` on per-commit basis. `clang-static-analyzer` checks are also enabled. `clang-tidy` is also used for some style checks. -我们已经评估了 `clang-tidy`、`Coverity`、`cppcheck`、`PVS-Studio`、`tscancode`、`CodeQL`. 您将在 `tests/instructions/` 目录中找到使用说明. 你也可以阅读[俄文文章](https://habr.com/company/yandex/blog/342018/). +We have evaluated `clang-tidy`, `Coverity`, `cppcheck`, `PVS-Studio`, `tscancode`, `CodeQL`. You will find instructions for usage in `tests/instructions/` directory. -如果你使用 `CLion` 作为 IDE, 你可以利用一些开箱即用的 `clang-tidy` 检查 +If you use `CLion` as an IDE, you can leverage some `clang-tidy` checks out of the box. -我们还使用 `shellcheck` 对shell脚本进行静态分析. +We also use `shellcheck` for static analysis of shell scripts. -## 硬化 {#hardening} +## Hardening {#hardening} -在调试版本中, 我们使用自定义分配器执行用户级分配的 ASLR. +In debug build we are using custom allocator that does ASLR of user-level allocations. -我们还手动保护在分配后预期为只读的内存区域. +We also manually protect memory regions that are expected to be readonly after allocation. -在调试构建中, 我们还需要对libc进行自定义, 以确保不会调用 "有害的" (过时的、不安全的、非线程安全的)函数. +In debug build we also involve a customization of libc that ensures that no "harmful" (obsolete, insecure, not thread-safe) functions are called. -Debug 断言被广泛使用. +Debug assertions are used extensively. -在调试版本中，如果抛出带有 "逻辑错误" 代码(暗示错误)的异常, 则程序会过早终止. 它允许在发布版本中使用异常, 但在调试版本中使其成为断言. +In debug build, if exception with "logical error" code (implies a bug) is being thrown, the program is terminated prematurely. It allows to use exceptions in release build but make it an assertion in debug build. -jemalloc 的调试版本用于调试版本. -libc++ 的调试版本用于调试版本. +Debug version of jemalloc is used for debug builds. +Debug version of libc++ is used for debug builds. -## 运行时完整性检查 +## Runtime Integrity Checks -对存储在磁盘上的数据是校验和. MergeTree 表中的数据同时以三种方式进行校验和*(压缩数据块、未压缩数据块、跨块的总校验和). 客户端和服务器之间或服务器之间通过网络传输的数据也会进行校验和. 复制确保副本上的数据位相同. +Data stored on disk is checksummed. Data in MergeTree tables is checksummed in three ways simultaneously* (compressed data blocks, uncompressed data blocks, the total checksum across blocks). Data transferred over network between client and server or between servers is also checksummed. Replication ensures bit-identical data on replicas. -需要防止硬件故障(存储介质上的位腐烂、服务器上 RAM 中的位翻转、网络控制器 RAM 中的位翻转、网络交换机 RAM 中的位翻转、客户端 RAM 中的位翻转、线路上的位翻转). 请注意，比特位操作很常见, 即使对于 ECC RAM 和 TCP 校验和(如果您每天设法运行数千台处理 PB 数据的服务器, 也可能发生比特位操作. [观看视频(俄语)](https://www.youtube.com/watch?v=ooBAQIe0KlQ). +It is required to protect from faulty hardware (bit rot on storage media, bit flips in RAM on server, bit flips in RAM of network controller, bit flips in RAM of network switch, bit flips in RAM of client, bit flips on the wire). Note that bit flips are common and likely to occur even for ECC RAM and in presence of TCP checksums (if you manage to run thousands of servers processing petabytes of data each day). [See the video (russian)](https://www.youtube.com/watch?v=ooBAQIe0KlQ). -ClickHouse 提供诊断功能, 可帮助运维工程师找到故障硬件. +ClickHouse provides diagnostics that will help ops engineers to find faulty hardware. -\* 它并不慢. +\* and it is not slow. -## 代码风格 {#code-style} +## Code Style {#code-style} -[此处](style.md)描述了代码样式规则. +Code style rules are described [here](style.md). -要检查一些常见的样式违规，您可以使用 `utils/check-style` 脚本. +To check for some common style violations, you can use `utils/check-style` script. -要强制使用正确的代码样式, 您可以使用 `clang-format`. 文件 `.clang-format` 位于源根目录. 它大多与我们的实际代码风格相对应. 但是不建议将 `clang-format` 应用于现有文件, 因为它会使格式变得更糟. 您可以使用可以在 clang 源代码库中找到的 `clang-format-diff` 工具. +To force proper style of your code, you can use `clang-format`. File `.clang-format` is located at the sources root. It mostly corresponding with our actual code style. But it’s not recommended to apply `clang-format` to existing files because it makes formatting worse. You can use `clang-format-diff` tool that you can find in clang source repository. -或者, 您可以尝试使用 `uncrustify` 工具来重新格式化您的代码. 配置位于源根目录中的 `uncrustify.cfg` 中. 它比 `clang-format` 测试更少. +Alternatively you can try `uncrustify` tool to reformat your code. Configuration is in `uncrustify.cfg` in the sources root. It is less tested than `clang-format`. -`CLion` 有自己的代码格式化程序, 必须根据我们的代码风格进行调整. +`CLion` has its own code formatter that has to be tuned for our code style. -我们还使用 `codespell` 来查找代码中的拼写错误.它也是自动化的. +We also use `codespell` to find typos in code. It is automated as well. -## Metrica B2B 测试 {#metrica-b2b-tests} +## Test Coverage {#test-coverage} -每个 ClickHouse 版本都使用 Yandex Metrica 和 AppMetrica 引擎进行测试. ClickHouse 的测试版和稳定版部署在 VM 上, 并使用 Metrica 引擎的小副本运行, 该引擎处理输入数据的固定样本. 然后将两个 Metrica 引擎实例的结果放在一起比较. - -这些测试由单独的团队自动化. 由于移动部件数量众多, 测试在大多数情况下都因完全不相关的原因而失败, 这些原因很难弄清楚. 这些测试很可能对我们有负面价值. 尽管如此, 这些测试在数百次中被证明是有用的. - -## 测试覆盖率 {#test-coverage} - -我们还跟踪测试覆盖率, 但仅针对功能测试和 clickhouse-server. 它每天进行. +We also track test coverage but only for functional tests and only for clickhouse-server. It is performed on daily basis. ## Tests for Tests -有自动检测薄片测试. 它运行所有新测试100次(用于功能测试)或10次(用于集成测试). 如果至少有一次测试失败，它就被认为是脆弱的. +There is automated check for flaky tests. It runs all new tests 100 times (for functional tests) or 10 times (for integration tests). If at least single time the test failed, it is considered flaky. ## Testflows -[Testflows](https://testflows.com/) 是一个企业级的测试框架. Altinity 使用它进行一些测试, 我们在 CI 中运行这些测试. +[Testflows](https://testflows.com/) is an enterprise-grade open-source testing framework, which is used to test a subset of ClickHouse. -## Yandex 检查 (only for Yandex employees) +## Test Automation {#test-automation} -这些检查将ClickHouse代码导入到Yandex内部的单一存储库中, 所以ClickHouse代码库可以被Yandex的其他产品(YT和YDB)用作库. 请注意, clickhouse-server本身并不是由内部回购构建的, Yandex应用程序使用的是未经修改的开源构建的. +We run tests with [GitHub Actions](https://github.com/features/actions). -## 测试自动化 {#test-automation} +Build jobs and tests are run in Sandbox on per commit basis. Resulting packages and test results are published in GitHub and can be downloaded by direct links. Artifacts are stored for several months. When you send a pull request on GitHub, we tag it as “can be tested” and our CI system will build ClickHouse packages (release, debug, with address sanitizer, etc) for you. -我们使用 Yandex 内部 CI 和名为 "Sandbox" 的作业自动化系统运行测试. +We do not use Travis CI due to the limit on time and computational power. +We do not use Jenkins. It was used before and now we are happy we are not using Jenkins. -在每次提交的基础上, 构建作业和测试都在沙箱中运行. 生成的包和测试结果发布在GitHub上, 可以通过直接链接下载. 产物要保存几个月. 当你在GitHub上发送一个pull请求时, 我们会把它标记为 "可以测试" , 我们的CI系统会为你构建ClickHouse包(发布、调试、使用地址清理器等). - -由于时间和计算能力的限制, 我们不使用 Travis CI. -我们不用Jenkins. 以前用过, 现在我们很高兴不用Jenkins了. - -[原始文章](https://clickhouse.com/docs/en/development/tests/) +[Original article](https://clickhouse.com/docs/en/development/tests/) diff --git a/docs/zh/sql-reference/data-types/lowcardinality.md b/docs/zh/sql-reference/data-types/lowcardinality.md index e089a7f9d41..717c3c979a4 100644 --- a/docs/zh/sql-reference/data-types/lowcardinality.md +++ b/docs/zh/sql-reference/data-types/lowcardinality.md @@ -55,6 +55,5 @@ ORDER BY id ## 参考 -- [高效低基数类型](https://www.altinity.com/blog/2019/3/27/low-cardinality). - [使用低基数类型减少ClickHouse的存储成本 – 来自Instana工程师的分享](https://www.instana.com/blog/reducing-clickhouse-storage-cost-with-the-low-cardinality-type-lessons-from-an-instana-engineer/). -- [字符优化 (俄语视频分享)](https://youtu.be/rqf-ILRgBdY?list=PL0Z2YDlm0b3iwXCpEFiOOYmwXzVmjJfEt). [英语分享](https://github.com/ClickHouse/clickhouse-presentations/raw/master/meetup19/string_optimization.pdf). \ No newline at end of file +- [字符优化 (俄语视频分享)](https://youtu.be/rqf-ILRgBdY?list=PL0Z2YDlm0b3iwXCpEFiOOYmwXzVmjJfEt). [英语分享](https://github.com/ClickHouse/clickhouse-presentations/raw/master/meetup19/string_optimization.pdf). diff --git a/docs/zh/sql-reference/statements/create.md b/docs/zh/sql-reference/statements/create.md index aef21a704b5..af77f4750b5 100644 --- a/docs/zh/sql-reference/statements/create.md +++ b/docs/zh/sql-reference/statements/create.md @@ -121,8 +121,6 @@ ENGINE = ... ``` -如果指定了编解ec，则默认编解码器不适用。 编解码器可以组合在一个流水线中，例如, `CODEC(Delta, ZSTD)`. 要为您的项目选择最佳的编解码器组合，请通过类似于Altinity中描述的基准测试 [新编码提高ClickHouse效率](https://www.altinity.com/blog/2019/7/new-encodings-to-improve-clickhouse) 文章. - !!! warning "警告" 您无法使用外部实用程序解压缩ClickHouse数据库文件，如 `lz4`. 相反，使用特殊的 [ﾂ环板compressorｮﾂ嘉ｯﾂ偲](https://github.com/ClickHouse/ClickHouse/tree/master/programs/compressor) 实用程序。 diff --git a/programs/client/Client.cpp b/programs/client/Client.cpp index e27845de184..6506c23428a 100644 --- a/programs/client/Client.cpp +++ b/programs/client/Client.cpp @@ -723,7 +723,7 @@ bool Client::processWithFuzzing(const String & full_query) // queries, for lack of a better solution. // There is also a problem that fuzzer substitutes positive Int64 // literals or Decimal literals, which are then parsed back as - // UInt64, and suddenly duplicate alias substitition starts or stops + // UInt64, and suddenly duplicate alias substitution starts or stops // working (ASTWithAlias::formatImpl) or something like that. // So we compare not even the first and second formatting of the // query, but second and third. diff --git a/programs/disks/CommandMkDir.cpp b/programs/disks/CommandMkDir.cpp new file mode 100644 index 00000000000..11a940028a3 --- /dev/null +++ b/programs/disks/CommandMkDir.cpp @@ -0,0 +1,67 @@ +#pragma once + +#include "ICommand.h" +#include + +namespace DB +{ + +namespace ErrorCodes +{ + extern const int BAD_ARGUMENTS; +} + +class CommandMkDir : public ICommand +{ +public: + CommandMkDir() + { + command_name = "mkdir"; + command_option_description.emplace(createOptionsDescription("Allowed options", getTerminalWidth())); + description = "Create directory or directories recursively"; + usage = "mkdir [OPTION]... "; + command_option_description->add_options() + ("recursive", "recursively create directories") + ; + } + + void processOptions( + Poco::Util::LayeredConfiguration & config, + po::variables_map & options) const override + { + if (options.count("recursive")) + config.setBool("recursive", true); + } + + void execute( + const std::vector & command_arguments, + DB::ContextMutablePtr & global_context, + Poco::Util::LayeredConfiguration & config) override + { + if (command_arguments.size() != 1) + { + printHelpMessage(); + throw DB::Exception("Bad Arguments", DB::ErrorCodes::BAD_ARGUMENTS); + } + + String disk_name = config.getString("disk", "default"); + + String path = command_arguments[0]; + + DiskPtr disk = global_context->getDisk(disk_name); + + String full_path = fullPathWithValidate(disk, path); + bool recursive = config.getBool("recursive", false); + + if (recursive) + disk->createDirectories(full_path); + else + disk->createDirectory(full_path); + } +}; +} + +std::unique_ptr makeCommandMkDir() +{ + return std::make_unique(); +} diff --git a/programs/disks/DisksApp.cpp b/programs/disks/DisksApp.cpp index 08768386808..b662921a3b1 100644 --- a/programs/disks/DisksApp.cpp +++ b/programs/disks/DisksApp.cpp @@ -63,7 +63,7 @@ void DisksApp::addOptions( positional_options_description.add("command_name", 1); - supported_commands = {"list-disks", "list", "move", "remove", "link", "copy", "write", "read"}; + supported_commands = {"list-disks", "list", "move", "remove", "link", "copy", "write", "read", "mkdir"}; command_descriptions.emplace("list-disks", makeCommandListDisks()); command_descriptions.emplace("list", makeCommandList()); @@ -73,6 +73,7 @@ void DisksApp::addOptions( command_descriptions.emplace("copy", makeCommandCopy()); command_descriptions.emplace("write", makeCommandWrite()); command_descriptions.emplace("read", makeCommandRead()); + command_descriptions.emplace("mkdir", makeCommandMkDir()); } void DisksApp::processOptions() diff --git a/programs/disks/DisksApp.h b/programs/disks/DisksApp.h index cbb3a7dfcc9..24fa9c3d9eb 100644 --- a/programs/disks/DisksApp.h +++ b/programs/disks/DisksApp.h @@ -4,6 +4,7 @@ #include "CommandLink.cpp" #include "CommandList.cpp" #include "CommandListDisks.cpp" +#include "CommandMkDir.cpp" #include "CommandMove.cpp" #include "CommandRead.cpp" #include "CommandRemove.cpp" diff --git a/programs/disks/ICommand.h b/programs/disks/ICommand.h index 9cde55dbb6b..f57f74a880e 100644 --- a/programs/disks/ICommand.h +++ b/programs/disks/ICommand.h @@ -65,3 +65,4 @@ std::unique_ptr makeCommandMove(); std::unique_ptr makeCommandRead(); std::unique_ptr makeCommandRemove(); std::unique_ptr makeCommandWrite(); +std::unique_ptr makeCommandMkDir(); diff --git a/programs/git-import/git-import.cpp b/programs/git-import/git-import.cpp index 54f5d7f75ea..030ddd263fa 100644 --- a/programs/git-import/git-import.cpp +++ b/programs/git-import/git-import.cpp @@ -67,7 +67,7 @@ Run this tool inside your git repository. It will create .tsv files that can be The tool can process large enough repositories in a reasonable time. It has been tested on: - ClickHouse: 31 seconds; 3 million rows; -- LLVM: 8 minues; 62 million rows; +- LLVM: 8 minutes; 62 million rows; - Linux - 12 minutes; 85 million rows; - Chromium - 67 minutes; 343 million rows; (the numbers as of Sep 2020) diff --git a/programs/server/Server.cpp b/programs/server/Server.cpp index b621a897035..d788270ecf9 100644 --- a/programs/server/Server.cpp +++ b/programs/server/Server.cpp @@ -736,7 +736,9 @@ int Server::main(const std::vector & /*args*/) std::vector servers_to_start_before_tables; /// This object will periodically calculate some metrics. AsynchronousMetrics async_metrics( - global_context, config().getUInt("asynchronous_metrics_update_period_s", 1), + global_context, + config().getUInt("asynchronous_metrics_update_period_s", 1), + config().getUInt("asynchronous_heavy_metrics_update_period_s", 120), [&]() -> std::vector { std::vector metrics; diff --git a/src/Access/AccessControl.cpp b/src/Access/AccessControl.cpp index 4f62c4ca203..89292fe9272 100644 --- a/src/Access/AccessControl.cpp +++ b/src/Access/AccessControl.cpp @@ -79,7 +79,7 @@ public: /// No user, probably the user has been dropped while it was in the cache. cache.remove(params); } - auto res = ContextAccess::make(access_control, params); + auto res = std::make_shared(access_control, params); res->initialize(); cache.add(params, res); return res; diff --git a/src/Access/ContextAccess.cpp b/src/Access/ContextAccess.cpp index 59c0a692a84..4e409946666 100644 --- a/src/Access/ContextAccess.cpp +++ b/src/Access/ContextAccess.cpp @@ -410,7 +410,7 @@ std::shared_ptr ContextAccess::getFullAccess() { static const std::shared_ptr res = [] { - auto full_access = ContextAccess::make(); + auto full_access = std::make_shared(); full_access->is_full_access = true; full_access->access = std::make_shared(AccessRights::getFullAccess()); full_access->access_with_implicit = full_access->access; diff --git a/src/Access/ContextAccess.h b/src/Access/ContextAccess.h index 331b449125e..ce1ea2d1220 100644 --- a/src/Access/ContextAccess.h +++ b/src/Access/ContextAccess.h @@ -166,12 +166,6 @@ public: /// without any limitations. This is used for the global context. static std::shared_ptr getFullAccess(); - template - static std::shared_ptr make(Args &&... args) - { - return std::make_shared(std::forward(args)...); - } - ~ContextAccess(); private: diff --git a/src/Backups/BackupImpl.cpp b/src/Backups/BackupImpl.cpp index f6442545f48..ffd20e02dd3 100644 --- a/src/Backups/BackupImpl.cpp +++ b/src/Backups/BackupImpl.cpp @@ -537,7 +537,7 @@ SizeAndChecksum BackupImpl::getFileSizeAndChecksum(const String & file_name) con if (!info) throw Exception( ErrorCodes::BACKUP_ENTRY_NOT_FOUND, "Backup {}: Entry {} not found in the backup", backup_name, quoteString(file_name)); - return std::pair(info->size, info->checksum); + return {info->size, info->checksum}; } BackupEntryPtr BackupImpl::readFile(const String & file_name) const @@ -625,7 +625,7 @@ CheckBackupResult checkBaseBackupForFile(const SizeAndChecksum & base_backup_inf { /// We cannot reuse base backup because our file is smaller /// than file stored in previous backup - if (new_entry_info.size > base_backup_info.first) + if (new_entry_info.size < base_backup_info.first) return CheckBackupResult::HasNothing; if (base_backup_info.first == new_entry_info.size) @@ -682,8 +682,6 @@ ChecksumsForNewEntry calculateNewEntryChecksumsIfNeeded(BackupEntryPtr entry, si void BackupImpl::writeFile(const String & file_name, BackupEntryPtr entry) { - - std::lock_guard lock{mutex}; if (open_mode != OpenMode::WRITE) throw Exception("Backup is not opened for writing", ErrorCodes::LOGICAL_ERROR); @@ -802,7 +800,12 @@ void BackupImpl::writeFile(const String & file_name, BackupEntryPtr entry) /// or have only prefix of it in previous backup. Let's go long path. info.data_file_name = info.file_name; - info.archive_suffix = current_archive_suffix; + + if (use_archives) + { + std::lock_guard lock{mutex}; + info.archive_suffix = current_archive_suffix; + } bool is_data_file_required; coordination->addFileInfo(info, is_data_file_required); @@ -818,9 +821,11 @@ void BackupImpl::writeFile(const String & file_name, BackupEntryPtr entry) /// if source and destination are compatible if (!use_archives && info.base_size == 0 && writer->supportNativeCopy(reader_description)) { - + /// Should be much faster than writing data through server. LOG_TRACE(log, "Will copy file {} using native copy", adjusted_path); - /// Should be much faster than writing data through server + + /// NOTE: `mutex` must be unlocked here otherwise writing will be in one thread maximum and hence slow. + writer->copyFileNative(entry->tryGetDiskIfExists(), entry->getFilePath(), info.data_file_name); } else @@ -838,6 +843,11 @@ void BackupImpl::writeFile(const String & file_name, BackupEntryPtr entry) if (use_archives) { LOG_TRACE(log, "Adding file {} to archive", adjusted_path); + + /// An archive must be written strictly in one thread, so it's correct to lock the mutex for all the time we're writing the file + /// to the archive. + std::lock_guard lock{mutex}; + String archive_suffix = current_archive_suffix; bool next_suffix = false; if (current_archive_suffix.empty() && is_internal_backup) @@ -859,6 +869,7 @@ void BackupImpl::writeFile(const String & file_name, BackupEntryPtr entry) } else { + /// NOTE: `mutex` must be unlocked here otherwise writing will be in one thread maximum and hence slow. writer->copyFileThroughBuffer(std::move(read_buffer), info.data_file_name); } } diff --git a/src/Backups/BackupImpl.h b/src/Backups/BackupImpl.h index 525aec2fcd6..e539239d3ef 100644 --- a/src/Backups/BackupImpl.h +++ b/src/Backups/BackupImpl.h @@ -130,7 +130,7 @@ private: std::pair> archive_writers[2]; String current_archive_suffix; String lock_file_name; - size_t num_files_written = 0; + std::atomic num_files_written = 0; bool writing_finalized = false; const Poco::Logger * log; }; diff --git a/src/CMakeLists.txt b/src/CMakeLists.txt index 3ece5fd410b..fd8771c1529 100644 --- a/src/CMakeLists.txt +++ b/src/CMakeLists.txt @@ -247,6 +247,7 @@ add_object_library(clickhouse_databases Databases) add_object_library(clickhouse_databases_mysql Databases/MySQL) add_object_library(clickhouse_disks Disks) add_object_library(clickhouse_interpreters Interpreters) +add_object_library(clickhouse_interpreters_cache Interpreters/Cache) add_object_library(clickhouse_interpreters_access Interpreters/Access) add_object_library(clickhouse_interpreters_mysql Interpreters/MySQL) add_object_library(clickhouse_interpreters_clusterproxy Interpreters/ClusterProxy) diff --git a/src/Client/Connection.cpp b/src/Client/Connection.cpp index 5047f958a10..7a663195655 100644 --- a/src/Client/Connection.cpp +++ b/src/Client/Connection.cpp @@ -24,6 +24,7 @@ #include #include "Core/Block.h" #include +#include #include #include #include @@ -483,6 +484,22 @@ void Connection::sendQuery( bool with_pending_data, std::function) { + OpenTelemetry::SpanHolder span("Connection::sendQuery()"); + span.addAttribute("clickhouse.query_id", query_id_); + span.addAttribute("clickhouse.query", query); + span.addAttribute("target", [this] () { return this->getHost() + ":" + std::to_string(this->getPort()); }); + + ClientInfo new_client_info; + const auto ¤t_trace_context = OpenTelemetry::CurrentContext(); + if (client_info && current_trace_context.isTraceEnabled()) + { + // use current span as the parent of remote span + new_client_info = *client_info; + new_client_info.client_trace_context = current_trace_context; + + client_info = &new_client_info; + } + if (!connected) connect(timeouts); @@ -540,7 +557,7 @@ void Connection::sendQuery( /// Send correct hash only for !INITIAL_QUERY, due to: /// - this will avoid extra protocol complexity for simplest cases /// - there is no need in hash for the INITIAL_QUERY anyway - /// (since there is no secure/unsecure changes) + /// (since there is no secure/non-secure changes) if (client_info && !cluster_secret.empty() && client_info->query_kind != ClientInfo::QueryKind::INITIAL_QUERY) { #if USE_SSL diff --git a/src/Client/HedgedConnectionsFactory.cpp b/src/Client/HedgedConnectionsFactory.cpp index a1b816deecb..81067f51d29 100644 --- a/src/Client/HedgedConnectionsFactory.cpp +++ b/src/Client/HedgedConnectionsFactory.cpp @@ -41,7 +41,7 @@ HedgedConnectionsFactory::HedgedConnectionsFactory( HedgedConnectionsFactory::~HedgedConnectionsFactory() { /// Stop anything that maybe in progress, - /// to avoid interfer with the subsequent connections. + /// to avoid interference with the subsequent connections. /// /// I.e. some replcas may be in the establishing state, /// this means that hedged connection is waiting for TablesStatusResponse, diff --git a/src/Columns/ColumnObject.cpp b/src/Columns/ColumnObject.cpp index 1680f7050c6..5b72b838b99 100644 --- a/src/Columns/ColumnObject.cpp +++ b/src/Columns/ColumnObject.cpp @@ -12,6 +12,7 @@ #include #include #include +#include namespace DB { @@ -823,6 +824,44 @@ MutableColumnPtr ColumnObject::cloneResized(size_t new_size) const return applyForSubcolumns([&](const auto & subcolumn) { return subcolumn.cloneResized(new_size); }); } +void ColumnObject::getPermutation(PermutationSortDirection, PermutationSortStability, size_t, int, Permutation & res) const +{ + res.resize(num_rows); + std::iota(res.begin(), res.end(), 0); +} + +void ColumnObject::compareColumn(const IColumn & rhs, size_t rhs_row_num, + PaddedPODArray * row_indexes, PaddedPODArray & compare_results, + int direction, int nan_direction_hint) const +{ + return doCompareColumn(assert_cast(rhs), rhs_row_num, row_indexes, + compare_results, direction, nan_direction_hint); +} + +void ColumnObject::getExtremes(Field & min, Field & max) const +{ + if (num_rows == 0) + { + min = Object(); + max = Object(); + } + else + { + get(0, min); + get(0, max); + } +} + +MutableColumns ColumnObject::scatter(ColumnIndex num_columns, const Selector & selector) const +{ + return scatterImpl(num_columns, selector); +} + +void ColumnObject::gather(ColumnGathererStream & gatherer) +{ + gatherer.gather(*this); +} + const ColumnObject::Subcolumn & ColumnObject::getSubcolumn(const PathInData & key) const { if (const auto * node = subcolumns.findLeaf(key)) diff --git a/src/Columns/ColumnObject.h b/src/Columns/ColumnObject.h index 594c6e047c4..07099307258 100644 --- a/src/Columns/ColumnObject.h +++ b/src/Columns/ColumnObject.h @@ -15,7 +15,7 @@ namespace DB namespace ErrorCodes { - extern const int LOGICAL_ERROR; + extern const int NOT_IMPLEMENTED; } /// Info that represents a scalar or array field in a decomposed view. @@ -224,6 +224,19 @@ public: ColumnPtr replicate(const Offsets & offsets) const override; MutableColumnPtr cloneResized(size_t new_size) const override; + /// Order of rows in ColumnObject is undefined. + void getPermutation(PermutationSortDirection, PermutationSortStability, size_t, int, Permutation & res) const override; + void compareColumn(const IColumn & rhs, size_t rhs_row_num, + PaddedPODArray * row_indexes, PaddedPODArray & compare_results, + int direction, int nan_direction_hint) const override; + + void updatePermutation(PermutationSortDirection, PermutationSortStability, size_t, int, Permutation &, EqualRanges &) const override {} + int compareAt(size_t, size_t, const IColumn &, int) const override { return 0; } + void getExtremes(Field & min, Field & max) const override; + + MutableColumns scatter(ColumnIndex num_columns, const Selector & selector) const override; + void gather(ColumnGathererStream & gatherer) override; + /// All other methods throw exception. StringRef getDataAt(size_t) const override { throwMustBeConcrete(); } @@ -236,14 +249,7 @@ public: void updateWeakHash32(WeakHash32 &) const override { throwMustBeConcrete(); } void updateHashFast(SipHash &) const override { throwMustBeConcrete(); } void expand(const Filter &, bool) override { throwMustBeConcrete(); } - int compareAt(size_t, size_t, const IColumn &, int) const override { throwMustBeConcrete(); } - void compareColumn(const IColumn &, size_t, PaddedPODArray *, PaddedPODArray &, int, int) const override { throwMustBeConcrete(); } bool hasEqualValues() const override { throwMustBeConcrete(); } - void getPermutation(PermutationSortDirection, PermutationSortStability, size_t, int, Permutation &) const override { throwMustBeConcrete(); } - void updatePermutation(PermutationSortDirection, PermutationSortStability, size_t, int, Permutation &, EqualRanges &) const override { throwMustBeConcrete(); } - MutableColumns scatter(ColumnIndex, const Selector &) const override { throwMustBeConcrete(); } - void gather(ColumnGathererStream &) override { throwMustBeConcrete(); } - void getExtremes(Field &, Field &) const override { throwMustBeConcrete(); } size_t byteSizeAt(size_t) const override { throwMustBeConcrete(); } double getRatioOfDefaultRows(double) const override { throwMustBeConcrete(); } void getIndicesOfNonDefaultRows(Offsets &, size_t, size_t) const override { throwMustBeConcrete(); } @@ -251,7 +257,7 @@ public: private: [[noreturn]] static void throwMustBeConcrete() { - throw Exception("ColumnObject must be converted to ColumnTuple before use", ErrorCodes::LOGICAL_ERROR); + throw Exception("ColumnObject must be converted to ColumnTuple before use", ErrorCodes::NOT_IMPLEMENTED); } template diff --git a/src/Common/CaresPTRResolver.cpp b/src/Common/CaresPTRResolver.cpp index e5d48b864c8..a02909309b6 100644 --- a/src/Common/CaresPTRResolver.cpp +++ b/src/Common/CaresPTRResolver.cpp @@ -15,8 +15,8 @@ namespace DB static void callback(void * arg, int status, int, struct hostent * host) { - auto * ptr_records = reinterpret_cast*>(arg); - if (status == ARES_SUCCESS && host->h_aliases) + auto * ptr_records = static_cast*>(arg); + if (ptr_records && status == ARES_SUCCESS) { /* * In some cases (e.g /etc/hosts), hostent::h_name is filled and hostent::h_aliases is empty. @@ -28,11 +28,14 @@ namespace DB ptr_records->insert(ptr_record); } - int i = 0; - while (auto * ptr_record = host->h_aliases[i]) + if (host->h_aliases) { - ptr_records->insert(ptr_record); - i++; + int i = 0; + while (auto * ptr_record = host->h_aliases[i]) + { + ptr_records->insert(ptr_record); + i++; + } } } } diff --git a/src/Common/CurrentMemoryTracker.cpp b/src/Common/CurrentMemoryTracker.cpp index 921c244da21..720df07efb9 100644 --- a/src/Common/CurrentMemoryTracker.cpp +++ b/src/Common/CurrentMemoryTracker.cpp @@ -52,15 +52,10 @@ void CurrentMemoryTracker::allocImpl(Int64 size, bool throw_if_memory_exceeded) if (current_thread) { Int64 will_be = current_thread->untracked_memory + size; - Int64 limit = current_thread->untracked_memory_limit + current_thread->untracked_memory_limit_increase; - if (will_be > limit) + if (will_be > current_thread->untracked_memory_limit) { - /// Increase limit before track. If tracker throws out-of-limit we would be able to alloc up to untracked_memory_limit bytes - /// more. It could be useful to enlarge Exception message in rethrow logic. - current_thread->untracked_memory_limit_increase = current_thread->untracked_memory_limit; memory_tracker->allocImpl(will_be, throw_if_memory_exceeded); - current_thread->untracked_memory_limit_increase = 0; current_thread->untracked_memory = 0; } else diff --git a/src/Common/IntervalKind.h b/src/Common/IntervalKind.h index 59e8d32f3e7..b46805655b1 100644 --- a/src/Common/IntervalKind.h +++ b/src/Common/IntervalKind.h @@ -64,7 +64,7 @@ struct IntervalKind const char * toNameOfFunctionExtractTimePart() const; /// Converts the string representation of an interval kind to its IntervalKind equivalent. - /// Returns false if the conversion unsucceeded. + /// Returns false if the conversion did not succeed. /// For example, `IntervalKind::tryParseString('second', result)` returns `result` equals `IntervalKind::Kind::Second`. static bool tryParseString(const std::string & kind, IntervalKind::Kind & result); }; diff --git a/src/Common/MemoryTracker.cpp b/src/Common/MemoryTracker.cpp index da3ce3cc3d2..29dbcdd28b0 100644 --- a/src/Common/MemoryTracker.cpp +++ b/src/Common/MemoryTracker.cpp @@ -166,27 +166,7 @@ void MemoryTracker::allocImpl(Int64 size, bool throw_if_memory_exceeded, MemoryT } } - std::bernoulli_distribution fault(fault_probability); - if (unlikely(fault_probability && fault(thread_local_rng)) && memoryTrackerCanThrow(level, true) && throw_if_memory_exceeded) - { - /// Revert - amount.fetch_sub(size, std::memory_order_relaxed); - - /// Prevent recursion. Exception::ctor -> std::string -> new[] -> MemoryTracker::alloc - MemoryTrackerBlockerInThread untrack_lock(VariableContext::Global); - - ProfileEvents::increment(ProfileEvents::QueryMemoryLimitExceeded); - const auto * description = description_ptr.load(std::memory_order_relaxed); - throw DB::Exception( - DB::ErrorCodes::MEMORY_LIMIT_EXCEEDED, - "Memory tracker{}{}: fault injected. Would use {} (attempt to allocate chunk of {} bytes), maximum: {}", - description ? " " : "", - description ? description : "", - formatReadableSizeWithBinarySuffix(will_be), - size, - formatReadableSizeWithBinarySuffix(current_hard_limit)); - } - + bool memory_limit_exceeded_ignored = false; bool allocation_traced = false; if (unlikely(current_profiler_limit && will_be > current_profiler_limit)) @@ -205,54 +185,92 @@ void MemoryTracker::allocImpl(Int64 size, bool throw_if_memory_exceeded, MemoryT allocation_traced = true; } - if (unlikely(current_hard_limit && will_be > current_hard_limit) && memoryTrackerCanThrow(level, false) && throw_if_memory_exceeded) + std::bernoulli_distribution fault(fault_probability); + if (unlikely(fault_probability && fault(thread_local_rng))) { - OvercommitResult overcommit_result = OvercommitResult::NONE; - if (auto * overcommit_tracker_ptr = overcommit_tracker.load(std::memory_order_relaxed); overcommit_tracker_ptr != nullptr && query_tracker != nullptr) - overcommit_result = overcommit_tracker_ptr->needToStopQuery(query_tracker, size); - - if (overcommit_result != OvercommitResult::MEMORY_FREED) + if (memoryTrackerCanThrow(level, true) && throw_if_memory_exceeded) { /// Revert amount.fetch_sub(size, std::memory_order_relaxed); /// Prevent recursion. Exception::ctor -> std::string -> new[] -> MemoryTracker::alloc MemoryTrackerBlockerInThread untrack_lock(VariableContext::Global); + ProfileEvents::increment(ProfileEvents::QueryMemoryLimitExceeded); const auto * description = description_ptr.load(std::memory_order_relaxed); throw DB::Exception( DB::ErrorCodes::MEMORY_LIMIT_EXCEEDED, - "Memory limit{}{} exceeded: would use {} (attempt to allocate chunk of {} bytes), maximum: {}. OvercommitTracker decision: {}.", + "Memory tracker{}{}: fault injected. Would use {} (attempt to allocate chunk of {} bytes), maximum: {}", description ? " " : "", description ? description : "", formatReadableSizeWithBinarySuffix(will_be), size, - formatReadableSizeWithBinarySuffix(current_hard_limit), - toDescription(overcommit_result)); + formatReadableSizeWithBinarySuffix(current_hard_limit)); + } + else + memory_limit_exceeded_ignored = true; + } + + + if (unlikely(current_hard_limit && will_be > current_hard_limit)) + { + if (memoryTrackerCanThrow(level, false) && throw_if_memory_exceeded) + { + OvercommitResult overcommit_result = OvercommitResult::NONE; + if (auto * overcommit_tracker_ptr = overcommit_tracker.load(std::memory_order_relaxed); overcommit_tracker_ptr != nullptr && query_tracker != nullptr) + overcommit_result = overcommit_tracker_ptr->needToStopQuery(query_tracker, size); + + if (overcommit_result != OvercommitResult::MEMORY_FREED) + { + /// Revert + amount.fetch_sub(size, std::memory_order_relaxed); + + /// Prevent recursion. Exception::ctor -> std::string -> new[] -> MemoryTracker::alloc + MemoryTrackerBlockerInThread untrack_lock(VariableContext::Global); + ProfileEvents::increment(ProfileEvents::QueryMemoryLimitExceeded); + const auto * description = description_ptr.load(std::memory_order_relaxed); + throw DB::Exception( + DB::ErrorCodes::MEMORY_LIMIT_EXCEEDED, + "Memory limit{}{} exceeded: would use {} (attempt to allocate chunk of {} bytes), maximum: {}. OvercommitTracker decision: {}.", + description ? " " : "", + description ? description : "", + formatReadableSizeWithBinarySuffix(will_be), + size, + formatReadableSizeWithBinarySuffix(current_hard_limit), + toDescription(overcommit_result)); + } + else + { + // If OvercommitTracker::needToStopQuery returned false, it guarantees that enough memory is freed. + // This memory is already counted in variable `amount` in the moment of `will_be` initialization. + // Now we just need to update value stored in `will_be`, because it should have changed. + will_be = amount.load(std::memory_order_relaxed); + } + } + else + memory_limit_exceeded_ignored = true; + } + + bool peak_updated = false; + /// In case of MEMORY_LIMIT_EXCEEDED was ignored, will_be may include + /// memory of other allocations, that may fail but not reverted yet, and so + /// updating peak will be inaccurate. + if (!memory_limit_exceeded_ignored) + { + if (throw_if_memory_exceeded) + { + /// Prevent recursion. Exception::ctor -> std::string -> new[] -> MemoryTracker::alloc + MemoryTrackerBlockerInThread untrack_lock(VariableContext::Global); + bool log_memory_usage = true; + peak_updated = updatePeak(will_be, log_memory_usage); } else { - // If OvercommitTracker::needToStopQuery returned false, it guarantees that enough memory is freed. - // This memory is already counted in variable `amount` in the moment of `will_be` initialization. - // Now we just need to update value stored in `will_be`, because it should have changed. - will_be = amount.load(std::memory_order_relaxed); + bool log_memory_usage = false; + peak_updated = updatePeak(will_be, log_memory_usage); } } - bool peak_updated; - if (throw_if_memory_exceeded) - { - /// Prevent recursion. Exception::ctor -> std::string -> new[] -> MemoryTracker::alloc - MemoryTrackerBlockerInThread untrack_lock(VariableContext::Global); - bool log_memory_usage = true; - peak_updated = updatePeak(will_be, log_memory_usage); - } - else - { - bool log_memory_usage = false; - peak_updated = updatePeak(will_be, log_memory_usage); - } - if (peak_updated && allocation_traced) { MemoryTrackerBlockerInThread untrack_lock(VariableContext::Global); diff --git a/src/Common/OpenTelemetryTraceContext.cpp b/src/Common/OpenTelemetryTraceContext.cpp new file mode 100644 index 00000000000..c05d3385bc8 --- /dev/null +++ b/src/Common/OpenTelemetryTraceContext.cpp @@ -0,0 +1,334 @@ +#include "Interpreters/OpenTelemetrySpanLog.h" + +#include +#include +#include +#include +#include +#include + +namespace DB +{ +namespace OpenTelemetry +{ + +thread_local TracingContextOnThread current_thread_trace_context; + +void Span::addAttribute(std::string_view name, UInt64 value) +{ + if (!this->isTraceEnabled() || name.empty()) + return; + + this->attributes.push_back(Tuple{name, toString(value)}); +} + +void Span::addAttributeIfNotZero(std::string_view name, UInt64 value) +{ + if (value != 0) + addAttribute(name, value); +} + +void Span::addAttribute(std::string_view name, std::string_view value) +{ + if (!this->isTraceEnabled() || name.empty()) + return; + + this->attributes.push_back(Tuple{name, value}); +} + +void Span::addAttributeIfNotEmpty(std::string_view name, std::string_view value) +{ + if (!this->isTraceEnabled() || name.empty() || value.empty()) + return; + + this->attributes.push_back(Tuple{name, value}); +} + +void Span::addAttribute(std::string_view name, std::function value_supplier) +{ + if (!this->isTraceEnabled() || !value_supplier) + return; + + String value = value_supplier(); + if (value.empty()) + return; + + this->attributes.push_back(Tuple{name, value}); +} + +void Span::addAttribute(const Exception & e) noexcept +{ + if (!this->isTraceEnabled()) + return; + + try + { + this->attributes.push_back(Tuple{"clickhouse.exception", getExceptionMessage(e, false)}); + } + catch (...) + { + /// Ignore exceptions + } +} + +void Span::addAttribute(std::exception_ptr e) noexcept +{ + if (!this->isTraceEnabled() || e == nullptr) + return; + + try + { + this->attributes.push_back(Tuple{"clickhouse.exception", getExceptionMessage(e, false)}); + } + catch (...) + { + /// Ignore exceptions + } +} + +SpanHolder::SpanHolder(std::string_view _operation_name) +{ + if (current_thread_trace_context.isTraceEnabled()) + { + this->trace_id = current_thread_trace_context.trace_id; + this->parent_span_id = current_thread_trace_context.span_id; + this->span_id = thread_local_rng(); // create a new id for this span + this->operation_name = _operation_name; + this->start_time_us + = std::chrono::duration_cast(std::chrono::system_clock::now().time_since_epoch()).count(); + + // set current span id to this + current_thread_trace_context.span_id = this->span_id; + } +} + +void SpanHolder::finish() noexcept +{ + if (!this->isTraceEnabled()) + return; + + // First of all, restore old value of current span. + assert(current_thread_trace_context.span_id == span_id); + current_thread_trace_context.span_id = parent_span_id; + + try + { + auto log = current_thread_trace_context.span_log.lock(); + if (!log) + { + // The log might be disabled. + return; + } + + this->finish_time_us + = std::chrono::duration_cast(std::chrono::system_clock::now().time_since_epoch()).count(); + + log->add(OpenTelemetrySpanLogElement(*this)); + } + catch (...) + { + tryLogCurrentException(__FUNCTION__); + } + + trace_id = UUID(); +} + +SpanHolder::~SpanHolder() +{ + finish(); +} + +bool TracingContext::parseTraceparentHeader(std::string_view traceparent, String & error) +{ + trace_id = 0; + + // Version 00, which is the only one we can parse, is fixed width. Use this + // fact for an additional sanity check. + const int expected_length = strlen("xx-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx-xxxxxxxxxxxxxxxx-xx"); + if (traceparent.length() != expected_length) + { + error = fmt::format("unexpected length {}, expected {}", traceparent.length(), expected_length); + return false; + } + + const char * data = traceparent.data(); + + uint8_t version = unhex2(data); + data += 2; + + if (version != 0) + { + error = fmt::format("unexpected version {}, expected 00", version); + return false; + } + + if (*data != '-') + { + error = fmt::format("Malformed traceparant header: {}", traceparent); + return false; + } + + ++data; + UInt64 trace_id_higher_64 = unhexUInt(data); + UInt64 trace_id_lower_64 = unhexUInt(data + 16); + data += 32; + + if (*data != '-') + { + error = fmt::format("Malformed traceparant header: {}", traceparent); + return false; + } + + ++data; + UInt64 span_id_64 = unhexUInt(data); + data += 16; + + if (*data != '-') + { + error = fmt::format("Malformed traceparant header: {}", traceparent); + return false; + } + + ++data; + this->trace_flags = unhex2(data); + this->trace_id.toUnderType().items[0] = trace_id_higher_64; + this->trace_id.toUnderType().items[1] = trace_id_lower_64; + this->span_id = span_id_64; + return true; +} + +String TracingContext::composeTraceparentHeader() const +{ + // This span is a parent for its children, so we specify this span_id as a + // parent id. + return fmt::format( + "00-{:016x}{:016x}-{:016x}-{:02x}", + trace_id.toUnderType().items[0], + trace_id.toUnderType().items[1], + span_id, + // This cast is needed because fmt is being weird and complaining that + // "mixing character types is not allowed". + static_cast(trace_flags)); +} + +const TracingContextOnThread & CurrentContext() +{ + return current_thread_trace_context; +} + +void TracingContextOnThread::reset() +{ + this->trace_id = UUID(); + this->span_id = 0; + this->trace_flags = TRACE_FLAG_NONE; + this->tracestate = ""; + this->span_log.reset(); +} + +TracingContextHolder::TracingContextHolder( + std::string_view _operation_name, + TracingContext _parent_trace_context, + const Settings * settings_ptr, + const std::weak_ptr & _span_log) +{ + if (current_thread_trace_context.isTraceEnabled()) + { + /// + /// This is not the normal case, + /// it means that construction of current object is not at the start of current thread. + /// Usually this is due to: + /// 1. bad design + /// 2. right design but code changes so that original point where this object is constructing is not the new start execution of current thread + /// + /// In such case, we should use current context as parent of this new constructing object, + /// So this branch ensures this class can be instantiated multiple times on one same thread safely. + /// + this->is_context_owner = false; + this->root_span.trace_id = current_thread_trace_context.trace_id; + this->root_span.parent_span_id = current_thread_trace_context.span_id; + this->root_span.span_id = thread_local_rng(); + this->root_span.operation_name = _operation_name; + this->root_span.start_time_us + = std::chrono::duration_cast(std::chrono::system_clock::now().time_since_epoch()).count(); + + current_thread_trace_context.span_id = this->root_span.span_id; + return; + } + + if (!_parent_trace_context.isTraceEnabled()) + { + if (settings_ptr == nullptr) + /// skip tracing context initialization on current thread + return; + + // start the trace ourselves, with some configurable probability. + std::bernoulli_distribution should_start_trace{settings_ptr->opentelemetry_start_trace_probability}; + if (!should_start_trace(thread_local_rng)) + /// skip tracing context initialization on current thread + return; + + while (_parent_trace_context.trace_id == UUID()) + { + // make sure the random generated trace_id is not 0 which is an invalid id + _parent_trace_context.trace_id.toUnderType().items[0] = thread_local_rng(); //-V656 + _parent_trace_context.trace_id.toUnderType().items[1] = thread_local_rng(); //-V656 + } + _parent_trace_context.span_id = 0; + } + + this->root_span.trace_id = _parent_trace_context.trace_id; + this->root_span.parent_span_id = _parent_trace_context.span_id; + this->root_span.span_id = thread_local_rng(); + this->root_span.operation_name = _operation_name; + this->root_span.start_time_us + = std::chrono::duration_cast(std::chrono::system_clock::now().time_since_epoch()).count(); + + /// This object is created to initialize tracing context on a new thread, + /// it's helpful to record the thread_id so that we know the thread switching from the span log + this->root_span.addAttribute("clickhouse.thread_id", getThreadId()); + + /// set up trace context on current thread + current_thread_trace_context = _parent_trace_context; + current_thread_trace_context.span_id = this->root_span.span_id; + current_thread_trace_context.trace_flags = TRACE_FLAG_SAMPLED; + current_thread_trace_context.span_log = _span_log; +} + +TracingContextHolder::~TracingContextHolder() +{ + if (!this->root_span.isTraceEnabled()) + { + return; + } + + try + { + auto shared_span_log = current_thread_trace_context.span_log.lock(); + if (shared_span_log) + { + this->root_span.finish_time_us + = std::chrono::duration_cast(std::chrono::system_clock::now().time_since_epoch()).count(); + + shared_span_log->add(OpenTelemetrySpanLogElement(this->root_span)); + } + } + catch (...) + { + tryLogCurrentException(__FUNCTION__); + } + + this->root_span.trace_id = UUID(); + + if (this->is_context_owner) + { + /// Clear the context on current thread + current_thread_trace_context.reset(); + } + else + { + current_thread_trace_context.span_id = this->root_span.parent_span_id; + } +} + +} +} diff --git a/src/Common/OpenTelemetryTraceContext.h b/src/Common/OpenTelemetryTraceContext.h index 4d2fc656100..3964b5030fc 100644 --- a/src/Common/OpenTelemetryTraceContext.h +++ b/src/Common/OpenTelemetryTraceContext.h @@ -1,24 +1,161 @@ #pragma once -#include -#include +#include namespace DB { -// The runtime info we need to create new OpenTelemetry spans. -struct OpenTelemetryTraceContext +struct Settings; +class OpenTelemetrySpanLog; + +namespace OpenTelemetry +{ + +struct Span +{ + UUID trace_id{}; + UInt64 span_id = 0; + UInt64 parent_span_id = 0; + String operation_name; + UInt64 start_time_us = 0; + UInt64 finish_time_us = 0; + Map attributes; + + void addAttribute(std::string_view name, UInt64 value); + void addAttributeIfNotZero(std::string_view name, UInt64 value); + void addAttribute(std::string_view name, std::string_view value); + void addAttributeIfNotEmpty(std::string_view name, std::string_view value); + void addAttribute(std::string_view name, std::function value_supplier); + + /// Following two methods are declared as noexcept to make sure they're exception safe + /// This is because they're usually called in exception handler + void addAttribute(const Exception & e) noexcept; + void addAttribute(std::exception_ptr e) noexcept; + + bool isTraceEnabled() const + { + return trace_id != UUID(); + } +}; + +/// See https://www.w3.org/TR/trace-context/ for trace_flags definition +enum TraceFlags : UInt8 +{ + TRACE_FLAG_NONE = 0, + TRACE_FLAG_SAMPLED = 1, +}; + +/// The runtime info we need to create new OpenTelemetry spans. +struct TracingContext { UUID trace_id{}; UInt64 span_id = 0; // The incoming tracestate header and the trace flags, we just pass them // downstream. See https://www.w3.org/TR/trace-context/ String tracestate; - UInt8 trace_flags = 0; + UInt8 trace_flags = TRACE_FLAG_NONE; // Parse/compose OpenTelemetry traceparent header. - bool parseTraceparentHeader(const std::string & traceparent, std::string & error); - std::string composeTraceparentHeader() const; + bool parseTraceparentHeader(std::string_view traceparent, String & error); + String composeTraceparentHeader() const; + + bool isTraceEnabled() const + { + return trace_id != UUID(); + } +}; + +/// Tracing context kept on each thread +struct TracingContextOnThread : TracingContext +{ + TracingContextOnThread& operator =(const TracingContext& context) + { + *(static_cast(this)) = context; + return *this; + } + + void reset(); + + /// Use weak_ptr instead of shared_ptr to hold a reference to the underlying system.opentelemetry_span_log table + /// Since this object is kept on threads and passed across threads, a weak_ptr is more safe to prevent potential leak + std::weak_ptr span_log; +}; + +/// Get tracing context on current thread +const TracingContextOnThread& CurrentContext(); + +/// Holder of tracing context. +/// It should be initialized at the beginning of each thread execution. +/// And once it's destructed, it clears the context automatically. +/// +/// It's also the root of all spans on current thread execution. +/// +/// Although it's SAFE to construct this object multiple times on one same thread, it should be created at the beginning of one thread execution. +struct TracingContextHolder +{ + /// Forbidden copy ctor and assignment to make the destructor safe + TracingContextHolder(const TracingContextHolder& scope) = delete; + TracingContextHolder& operator =(const TracingContextHolder& scope) = delete; + + TracingContextHolder(std::string_view _operation_name, + const TracingContext& _parent_trace_context, + const std::weak_ptr& _log) + : TracingContextHolder(_operation_name, + _parent_trace_context, + nullptr, + _log) + { + } + + /// Initialize a tracing context on a child thread based on the context from the parent thread + TracingContextHolder(std::string_view _operation_name, const TracingContextOnThread & _parent_thread_trace_context) + : TracingContextHolder(_operation_name, + _parent_thread_trace_context, + nullptr, + _parent_thread_trace_context.span_log) + { + } + + /// For servers like HTTP/TCP/GRPC to initialize tracing context on thread that process requests from clients + TracingContextHolder(std::string_view _operation_name, + TracingContext _parent_trace_context, + const Settings & _settings, + const std::weak_ptr & _log) + : TracingContextHolder(_operation_name, + _parent_trace_context, + &_settings, + _log) + { + } + + TracingContextHolder(std::string_view _operation_name, + TracingContext _parent_trace_context, + const Settings* settings_ptr, + const std::weak_ptr & _log); + + ~TracingContextHolder(); + + Span root_span; + +private: + bool is_context_owner = true; +}; + +using TracingContextHolderPtr = std::unique_ptr; + +/// A span holder that creates span automatically in a (function) scope if tracing is enabled. +/// Once it's created or destructed, it automatically maitains the tracing context on the thread that it lives. +struct SpanHolder : public Span +{ + SpanHolder(std::string_view); + ~SpanHolder(); + + /// Finish a span explicitly if needed. + /// It's safe to call it multiple times + void finish() noexcept; }; } + +} + diff --git a/src/Common/SLRUCachePolicy.h b/src/Common/SLRUCachePolicy.h index 10b043ebaca..8d4709c66a7 100644 --- a/src/Common/SLRUCachePolicy.h +++ b/src/Common/SLRUCachePolicy.h @@ -33,7 +33,7 @@ public: * max_protected_size shows how many of the most frequently used entries will not be evicted after a sequential scan. * max_protected_size == 0 means that the default protected size is equal to half of the total max size. */ - /// TODO: construct from special struct with cache policy parametrs (also with max_protected_size). + /// TODO: construct from special struct with cache policy parameters (also with max_protected_size). SLRUCachePolicy(size_t max_size_, size_t max_elements_size_ = 0, double size_ratio = 0.5, OnWeightLossFunction on_weight_loss_function_ = {}) : max_protected_size(max_size_ * std::min(1.0, size_ratio)) , max_size(max_size_) diff --git a/src/Common/Stopwatch.h b/src/Common/Stopwatch.h index 2b60bbde196..cabc6d8ba1e 100644 --- a/src/Common/Stopwatch.h +++ b/src/Common/Stopwatch.h @@ -31,7 +31,7 @@ inline UInt64 clock_gettime_ns_adjusted(UInt64 prev_time, clockid_t clock_type = } /** Differs from Poco::Stopwatch only by using 'clock_gettime' instead of 'gettimeofday', - * returns nanoseconds instead of microseconds, and also by other minor differencies. + * returns nanoseconds instead of microseconds, and also by other minor differences. */ class Stopwatch { @@ -152,4 +152,3 @@ private: /// Most significant bit is a lock. When it is set, compareAndRestartDeferred method will return false. UInt64 nanoseconds(UInt64 prev_time) const { return clock_gettime_ns_adjusted(prev_time, clock_type) & 0x7FFFFFFFFFFFFFFFULL; } }; - diff --git a/src/Common/ThreadPool.cpp b/src/Common/ThreadPool.cpp index 3f5091af0c9..0b89139fa53 100644 --- a/src/Common/ThreadPool.cpp +++ b/src/Common/ThreadPool.cpp @@ -2,6 +2,7 @@ #include #include #include +#include #include #include @@ -86,7 +87,7 @@ void ThreadPoolImpl::setQueueSize(size_t value) template template -ReturnType ThreadPoolImpl::scheduleImpl(Job job, int priority, std::optional wait_microseconds) +ReturnType ThreadPoolImpl::scheduleImpl(Job job, int priority, std::optional wait_microseconds, bool propagate_opentelemetry_tracing_context) { auto on_error = [&](const std::string & reason) { @@ -149,7 +150,11 @@ ReturnType ThreadPoolImpl::scheduleImpl(Job job, int priority, std::opti } } - jobs.emplace(std::move(job), priority); + jobs.emplace(std::move(job), + priority, + /// Tracing context on this thread is used as parent context for the sub-thread that runs the job + propagate_opentelemetry_tracing_context ? DB::OpenTelemetry::CurrentContext() : DB::OpenTelemetry::TracingContextOnThread()); + ++scheduled_jobs; new_job_or_shutdown.notify_one(); } @@ -170,9 +175,9 @@ bool ThreadPoolImpl::trySchedule(Job job, int priority, uint64_t wait_mi } template -void ThreadPoolImpl::scheduleOrThrow(Job job, int priority, uint64_t wait_microseconds) +void ThreadPoolImpl::scheduleOrThrow(Job job, int priority, uint64_t wait_microseconds, bool propagate_opentelemetry_tracing_context) { - scheduleImpl(std::move(job), priority, wait_microseconds); + scheduleImpl(std::move(job), priority, wait_microseconds, propagate_opentelemetry_tracing_context); } template @@ -250,6 +255,9 @@ void ThreadPoolImpl::worker(typename std::list::iterator thread_ Job job; bool need_shutdown = false; + /// A copy of parent trace context + DB::OpenTelemetry::TracingContextOnThread parent_thead_trace_context; + { std::unique_lock lock(mutex); new_job_or_shutdown.wait(lock, [this] { return shutdown || !jobs.empty(); }); @@ -260,6 +268,7 @@ void ThreadPoolImpl::worker(typename std::list::iterator thread_ /// boost::priority_queue does not provide interface for getting non-const reference to an element /// to prevent us from modifying its priority. We have to use const_cast to force move semantics on JobWithPriority::job. job = std::move(const_cast(jobs.top().job)); + parent_thead_trace_context = std::move(const_cast(jobs.top().thread_trace_context)); jobs.pop(); } else @@ -272,22 +281,40 @@ void ThreadPoolImpl::worker(typename std::list::iterator thread_ if (!need_shutdown) { + ALLOW_ALLOCATIONS_IN_SCOPE; + + /// Set up tracing context for this thread by its parent context + DB::OpenTelemetry::TracingContextHolder thread_trace_context("ThreadPool::worker()", parent_thead_trace_context); + try { - ALLOW_ALLOCATIONS_IN_SCOPE; CurrentMetrics::Increment metric_active_threads( std::is_same_v ? CurrentMetrics::GlobalThreadActive : CurrentMetrics::LocalThreadActive); job(); + + if (thread_trace_context.root_span.isTraceEnabled()) + { + /// Use the thread name as operation name so that the tracing log will be more clear. + /// The thread name is usually set in the jobs, we can only get the name after the job finishes + std::string thread_name = getThreadName(); + if (!thread_name.empty()) + thread_trace_context.root_span.operation_name = thread_name; + } + /// job should be reset before decrementing scheduled_jobs to /// ensure that the Job destroyed before wait() returns. job = {}; + parent_thead_trace_context.reset(); } catch (...) { + thread_trace_context.root_span.addAttribute(std::current_exception()); + /// job should be reset before decrementing scheduled_jobs to /// ensure that the Job destroyed before wait() returns. job = {}; + parent_thead_trace_context.reset(); { std::lock_guard lock(mutex); @@ -323,7 +350,8 @@ void ThreadPoolImpl::worker(typename std::list::iterator thread_ template class ThreadPoolImpl; -template class ThreadPoolImpl; +template class ThreadPoolImpl>; +template class ThreadFromGlobalPoolImpl; std::unique_ptr GlobalThreadPool::the_instance; diff --git a/src/Common/ThreadPool.h b/src/Common/ThreadPool.h index eb3f631b92a..fc5377b3783 100644 --- a/src/Common/ThreadPool.h +++ b/src/Common/ThreadPool.h @@ -14,6 +14,7 @@ #include #include +#include #include /** Very simple thread pool similar to boost::threadpool. @@ -55,7 +56,7 @@ public: bool trySchedule(Job job, int priority = 0, uint64_t wait_microseconds = 0) noexcept; /// Similar to scheduleOrThrowOnError(...). Wait for specified amount of time and schedule a job or throw an exception. - void scheduleOrThrow(Job job, int priority = 0, uint64_t wait_microseconds = 0); + void scheduleOrThrow(Job job, int priority = 0, uint64_t wait_microseconds = 0, bool propagate_opentelemetry_tracing_context = true); /// Wait for all currently active jobs to be done. /// You may call schedule and wait many times in arbitrary order. @@ -96,9 +97,10 @@ private: { Job job; int priority; + DB::OpenTelemetry::TracingContextOnThread thread_trace_context; - JobWithPriority(Job job_, int priority_) - : job(job_), priority(priority_) {} + JobWithPriority(Job job_, int priority_, const DB::OpenTelemetry::TracingContextOnThread& thread_trace_context_) + : job(job_), priority(priority_), thread_trace_context(thread_trace_context_) {} bool operator< (const JobWithPriority & rhs) const { @@ -111,7 +113,7 @@ private: std::exception_ptr first_exception; template - ReturnType scheduleImpl(Job job, int priority, std::optional wait_microseconds); + ReturnType scheduleImpl(Job job, int priority, std::optional wait_microseconds, bool propagate_opentelemetry_tracing_context = true); void worker(typename std::list::iterator thread_it); @@ -154,14 +156,18 @@ public: /** Looks like std::thread but allocates threads in GlobalThreadPool. * Also holds ThreadStatus for ClickHouse. + * + * NOTE: User code should use 'ThreadFromGlobalPool' declared below instead of directly using this class. + * */ -class ThreadFromGlobalPool : boost::noncopyable +template +class ThreadFromGlobalPoolImpl : boost::noncopyable { public: - ThreadFromGlobalPool() = default; + ThreadFromGlobalPoolImpl() = default; template - explicit ThreadFromGlobalPool(Function && func, Args &&... args) + explicit ThreadFromGlobalPoolImpl(Function && func, Args &&... args) : state(std::make_shared()) { /// NOTE: @@ -185,15 +191,19 @@ public: /// before sending signal that permits to join this thread. DB::ThreadStatus thread_status; std::apply(function, arguments); - }); + }, + 0, // default priority + 0, // default wait_microseconds + propagate_opentelemetry_context + ); } - ThreadFromGlobalPool(ThreadFromGlobalPool && rhs) noexcept + ThreadFromGlobalPoolImpl(ThreadFromGlobalPoolImpl && rhs) noexcept { *this = std::move(rhs); } - ThreadFromGlobalPool & operator=(ThreadFromGlobalPool && rhs) noexcept + ThreadFromGlobalPoolImpl & operator=(ThreadFromGlobalPoolImpl && rhs) noexcept { if (initialized()) abort(); @@ -201,7 +211,7 @@ public: return *this; } - ~ThreadFromGlobalPool() + ~ThreadFromGlobalPoolImpl() { if (initialized()) abort(); @@ -233,7 +243,7 @@ public: return true; } -private: +protected: struct State { /// Should be atomic() because of possible concurrent access between @@ -254,6 +264,19 @@ private: } }; - /// Recommended thread pool for the case when multiple thread pools are created and destroyed. -using ThreadPool = ThreadPoolImpl; +/// +/// The template parameter of ThreadFromGlobalPool is set to false to disable tracing context propagation to underlying worker. +/// Because ThreadFromGlobalPool schedules a job upon GlobalThreadPool, this means there will be two workers to schedule a job in 'ThreadPool', +/// one is at GlobalThreadPool level, the other is at ThreadPool level, so tracing context will be initialized on the same thread twice. +/// +/// Once the worker on ThreadPool gains the control of execution, it won't return until it's shutdown, +/// which means the tracing context initialized at underlying worker level won't be delete for a very long time. +/// This would cause wrong context for further jobs scheduled in ThreadPool. +/// +/// To make sure the tracing context are correctly propagated, we explicitly disable context propagation(including initialization and de-initialization) at underlying worker level. +/// +using ThreadPool = ThreadPoolImpl>; + +/// An alias for user code to execute a job in the global thread pool +using ThreadFromGlobalPool = ThreadFromGlobalPoolImpl; diff --git a/src/Common/ThreadStatus.cpp b/src/Common/ThreadStatus.cpp index 0cbc6f4ce0f..b62a7af6c71 100644 --- a/src/Common/ThreadStatus.cpp +++ b/src/Common/ThreadStatus.cpp @@ -3,7 +3,6 @@ #include #include #include -#include #include #include diff --git a/src/Common/ThreadStatus.h b/src/Common/ThreadStatus.h index 2a4ffd229f2..b414a9bccf5 100644 --- a/src/Common/ThreadStatus.h +++ b/src/Common/ThreadStatus.h @@ -4,7 +4,6 @@ #include #include #include -#include #include #include #include @@ -33,7 +32,6 @@ class ThreadStatus; class QueryProfilerReal; class QueryProfilerCPU; class QueryThreadLog; -struct OpenTelemetrySpanHolder; class TasksStatsCounters; struct RUsageCounters; struct PerfEventsCounters; @@ -135,8 +133,6 @@ public: Int64 untracked_memory = 0; /// Each thread could new/delete memory in range of (-untracked_memory_limit, untracked_memory_limit) without access to common counters. Int64 untracked_memory_limit = 4 * 1024 * 1024; - /// Increase limit in case of exception. - Int64 untracked_memory_limit_increase = 0; /// Statistics of read and write rows/bytes Progress progress_in; @@ -145,12 +141,6 @@ public: using Deleter = std::function; Deleter deleter; - // This is the current most-derived OpenTelemetry span for this thread. It - // can be changed throughout the query execution, whenever we enter a new - // span or exit it. See OpenTelemetrySpanHolder that is normally responsible - // for these changes. - OpenTelemetryTraceContext thread_trace_context; - protected: ThreadGroupStatusPtr thread_group; diff --git a/src/Common/Volnitsky.h b/src/Common/Volnitsky.h index d7ca7d35277..6f5948b6564 100644 --- a/src/Common/Volnitsky.h +++ b/src/Common/Volnitsky.h @@ -497,7 +497,7 @@ private: /// last index of offsets that was not processed size_t last; - /// limit for adding to hashtable. In worst case with case insentive search, the table will be filled at most as half + /// limit for adding to hashtable. In worst case with case insensitive search, the table will be filled at most as half static constexpr size_t small_limit = VolnitskyTraits::hash_size / 8; public: diff --git a/src/Common/tests/gtest_lru_file_cache.cpp b/src/Common/tests/gtest_lru_file_cache.cpp deleted file mode 100644 index bb454cc41eb..00000000000 --- a/src/Common/tests/gtest_lru_file_cache.cpp +++ /dev/null @@ -1,514 +0,0 @@ -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include - -namespace fs = std::filesystem; - -fs::path caches_dir = fs::current_path() / "lru_cache_test"; -String cache_base_path = caches_dir / "cache1" / ""; - -void assertRange( - [[maybe_unused]] size_t assert_n, DB::FileSegmentPtr file_segment, - const DB::FileSegment::Range & expected_range, DB::FileSegment::State expected_state) -{ - auto range = file_segment->range(); - - std::cerr << fmt::format("\nAssert #{} : {} == {} (state: {} == {})\n", assert_n, - range.toString(), expected_range.toString(), - toString(file_segment->state()), toString(expected_state)); - - ASSERT_EQ(range.left, expected_range.left); - ASSERT_EQ(range.right, expected_range.right); - ASSERT_EQ(file_segment->state(), expected_state); -} - -void printRanges(const auto & segments) -{ - std::cerr << "\nHaving file segments: "; - for (const auto & segment : segments) - std::cerr << '\n' << segment->range().toString() << " (state: " + DB::FileSegment::stateToString(segment->state()) + ")" << "\n"; -} - -std::vector fromHolder(const DB::FileSegmentsHolder & holder) -{ - return std::vector(holder.file_segments.begin(), holder.file_segments.end()); -} - -String getFileSegmentPath(const String & base_path, const DB::FileCache::Key & key, size_t offset) -{ - auto key_str = key.toString(); - return fs::path(base_path) / key_str.substr(0, 3) / key_str / DB::toString(offset); -} - -void download(DB::FileSegmentPtr file_segment) -{ - const auto & key = file_segment->key(); - size_t size = file_segment->range().size(); - - auto key_str = key.toString(); - auto subdir = fs::path(cache_base_path) / key_str.substr(0, 3) / key_str; - if (!fs::exists(subdir)) - fs::create_directories(subdir); - - std::string data(size, '0'); - file_segment->write(data.data(), size, file_segment->getDownloadOffset()); -} - -void prepareAndDownload(DB::FileSegmentPtr file_segment) -{ - // std::cerr << "Reserving: " << file_segment->range().size() << " for: " << file_segment->range().toString() << "\n"; - ASSERT_TRUE(file_segment->reserve(file_segment->range().size())); - download(file_segment); -} - -void complete(const DB::FileSegmentsHolder & holder) -{ - for (const auto & file_segment : holder.file_segments) - { - ASSERT_TRUE(file_segment->getOrSetDownloader() == DB::FileSegment::getCallerId()); - prepareAndDownload(file_segment); - file_segment->completeWithState(DB::FileSegment::State::DOWNLOADED); - } -} - - -TEST(FileCache, get) -{ - if (fs::exists(cache_base_path)) - fs::remove_all(cache_base_path); - fs::create_directories(cache_base_path); - - DB::ThreadStatus thread_status; - - /// To work with cache need query_id and query context. - std::string query_id = "query_id"; - auto query_context = DB::Context::createCopy(getContext().context); - query_context->makeQueryContext(); - query_context->setCurrentQueryId(query_id); - DB::CurrentThread::QueryScope query_scope_holder(query_context); - - DB::FileCacheSettings settings; - settings.max_size = 30; - settings.max_elements = 5; - auto cache = DB::FileCache(cache_base_path, settings); - cache.initialize(); - auto key = cache.hash("key1"); - - { - auto holder = cache.getOrSet(key, 0, 10, false); /// Add range [0, 9] - auto segments = fromHolder(holder); - /// Range was not present in cache. It should be added in cache as one while file segment. - ASSERT_EQ(segments.size(), 1); - - assertRange(1, segments[0], DB::FileSegment::Range(0, 9), DB::FileSegment::State::EMPTY); - - /// Exception because space not reserved. - /// EXPECT_THROW(download(segments[0]), DB::Exception); - /// Exception because space can be reserved only by downloader - /// EXPECT_THROW(segments[0]->reserve(segments[0]->range().size()), DB::Exception); - - ASSERT_TRUE(segments[0]->getOrSetDownloader() == DB::FileSegment::getCallerId()); - ASSERT_TRUE(segments[0]->reserve(segments[0]->range().size())); - assertRange(2, segments[0], DB::FileSegment::Range(0, 9), DB::FileSegment::State::DOWNLOADING); - - download(segments[0]); - segments[0]->completeWithState(DB::FileSegment::State::DOWNLOADED); - assertRange(3, segments[0], DB::FileSegment::Range(0, 9), DB::FileSegment::State::DOWNLOADED); - } - - /// Current cache: [__________] - /// ^ ^ - /// 0 9 - ASSERT_EQ(cache.getFileSegmentsNum(), 1); - ASSERT_EQ(cache.getUsedCacheSize(), 10); - - { - /// Want range [5, 14], but [0, 9] already in cache, so only [10, 14] will be put in cache. - auto holder = cache.getOrSet(key, 5, 10, false); - auto segments = fromHolder(holder); - ASSERT_EQ(segments.size(), 2); - - assertRange(4, segments[0], DB::FileSegment::Range(0, 9), DB::FileSegment::State::DOWNLOADED); - assertRange(5, segments[1], DB::FileSegment::Range(10, 14), DB::FileSegment::State::EMPTY); - - ASSERT_TRUE(segments[1]->getOrSetDownloader() == DB::FileSegment::getCallerId()); - prepareAndDownload(segments[1]); - segments[1]->completeWithState(DB::FileSegment::State::DOWNLOADED); - assertRange(6, segments[1], DB::FileSegment::Range(10, 14), DB::FileSegment::State::DOWNLOADED); - } - - /// Current cache: [__________][_____] - /// ^ ^^ ^ - /// 0 910 14 - ASSERT_EQ(cache.getFileSegmentsNum(), 2); - ASSERT_EQ(cache.getUsedCacheSize(), 15); - - { - auto holder = cache.getOrSet(key, 9, 1, false); /// Get [9, 9] - auto segments = fromHolder(holder); - ASSERT_EQ(segments.size(), 1); - assertRange(7, segments[0], DB::FileSegment::Range(0, 9), DB::FileSegment::State::DOWNLOADED); - } - - { - auto holder = cache.getOrSet(key, 9, 2, false); /// Get [9, 10] - auto segments = fromHolder(holder); - ASSERT_EQ(segments.size(), 2); - assertRange(8, segments[0], DB::FileSegment::Range(0, 9), DB::FileSegment::State::DOWNLOADED); - assertRange(9, segments[1], DB::FileSegment::Range(10, 14), DB::FileSegment::State::DOWNLOADED); - } - - { - auto holder = cache.getOrSet(key, 10, 1, false); /// Get [10, 10] - auto segments = fromHolder(holder); - ASSERT_EQ(segments.size(), 1); - assertRange(10, segments[0], DB::FileSegment::Range(10, 14), DB::FileSegment::State::DOWNLOADED); - } - - complete(cache.getOrSet(key, 17, 4, false)); /// Get [17, 20] - complete(cache.getOrSet(key, 24, 3, false)); /// Get [24, 26] - /// complete(cache.getOrSet(key, 27, 1, false)); /// Get [27, 27] - - /// Current cache: [__________][_____] [____] [___][] - /// ^ ^^ ^ ^ ^ ^ ^^^ - /// 0 910 14 17 20 24 2627 - /// - ASSERT_EQ(cache.getFileSegmentsNum(), 4); - ASSERT_EQ(cache.getUsedCacheSize(), 22); - - { - auto holder = cache.getOrSet(key, 0, 26, false); /// Get [0, 25] - auto segments = fromHolder(holder); - ASSERT_EQ(segments.size(), 6); - - assertRange(11, segments[0], DB::FileSegment::Range(0, 9), DB::FileSegment::State::DOWNLOADED); - assertRange(12, segments[1], DB::FileSegment::Range(10, 14), DB::FileSegment::State::DOWNLOADED); - - /// Missing [15, 16] should be added in cache. - assertRange(13, segments[2], DB::FileSegment::Range(15, 16), DB::FileSegment::State::EMPTY); - - ASSERT_TRUE(segments[2]->getOrSetDownloader() == DB::FileSegment::getCallerId()); - prepareAndDownload(segments[2]); - - segments[2]->completeWithState(DB::FileSegment::State::DOWNLOADED); - - assertRange(14, segments[3], DB::FileSegment::Range(17, 20), DB::FileSegment::State::DOWNLOADED); - - /// New [21, 23], but will not be added in cache because of elements limit (5) - assertRange(15, segments[4], DB::FileSegment::Range(21, 23), DB::FileSegment::State::EMPTY); - ASSERT_TRUE(segments[4]->getOrSetDownloader() == DB::FileSegment::getCallerId()); - ASSERT_FALSE(segments[4]->reserve(1)); - - assertRange(16, segments[5], DB::FileSegment::Range(24, 26), DB::FileSegment::State::DOWNLOADED); - - /// Current cache: [__________][_____][ ][____] [___] - /// ^ ^ ^ - /// 0 20 24 - /// - - /// Range [27, 27] must be evicted in previous getOrSet [0, 25]. - /// Let's not invalidate pointers to returned segments from range [0, 25] and - /// as max elements size is reached, next attempt to put something in cache should fail. - /// This will also check that [27, 27] was indeed evicted. - - auto holder1 = cache.getOrSet(key, 27, 1, false); - auto segments_1 = fromHolder(holder1); /// Get [27, 27] - ASSERT_EQ(segments_1.size(), 1); - assertRange(17, segments_1[0], DB::FileSegment::Range(27, 27), DB::FileSegment::State::EMPTY); - } - - { - auto holder = cache.getOrSet(key, 12, 10, false); /// Get [12, 21] - auto segments = fromHolder(holder); - ASSERT_EQ(segments.size(), 4); - - assertRange(18, segments[0], DB::FileSegment::Range(10, 14), DB::FileSegment::State::DOWNLOADED); - assertRange(19, segments[1], DB::FileSegment::Range(15, 16), DB::FileSegment::State::DOWNLOADED); - assertRange(20, segments[2], DB::FileSegment::Range(17, 20), DB::FileSegment::State::DOWNLOADED); - - assertRange(21, segments[3], DB::FileSegment::Range(21, 21), DB::FileSegment::State::EMPTY); - - ASSERT_TRUE(segments[3]->getOrSetDownloader() == DB::FileSegment::getCallerId()); - prepareAndDownload(segments[3]); - - segments[3]->completeWithState(DB::FileSegment::State::DOWNLOADED); - ASSERT_TRUE(segments[3]->state() == DB::FileSegment::State::DOWNLOADED); - } - - /// Current cache: [_____][__][____][_] [___] - /// ^ ^ ^ ^ ^ - /// 10 17 21 24 26 - - ASSERT_EQ(cache.getFileSegmentsNum(), 5); - - { - auto holder = cache.getOrSet(key, 23, 5, false); /// Get [23, 28] - auto segments = fromHolder(holder); - ASSERT_EQ(segments.size(), 3); - - assertRange(22, segments[0], DB::FileSegment::Range(23, 23), DB::FileSegment::State::EMPTY); - assertRange(23, segments[1], DB::FileSegment::Range(24, 26), DB::FileSegment::State::DOWNLOADED); - assertRange(24, segments[2], DB::FileSegment::Range(27, 27), DB::FileSegment::State::EMPTY); - - ASSERT_TRUE(segments[0]->getOrSetDownloader() == DB::FileSegment::getCallerId()); - ASSERT_TRUE(segments[2]->getOrSetDownloader() == DB::FileSegment::getCallerId()); - prepareAndDownload(segments[0]); - prepareAndDownload(segments[2]); - segments[0]->completeWithState(DB::FileSegment::State::DOWNLOADED); - segments[2]->completeWithState(DB::FileSegment::State::DOWNLOADED); - } - - /// Current cache: [____][_] [][___][__] - /// ^ ^ ^^^ ^^ ^ - /// 17 21 2324 26 28 - - { - auto holder5 = cache.getOrSet(key, 2, 3,false); /// Get [2, 4] - auto s5 = fromHolder(holder5); - ASSERT_EQ(s5.size(), 1); - assertRange(25, s5[0], DB::FileSegment::Range(2, 4), DB::FileSegment::State::EMPTY); - - auto holder1 = cache.getOrSet(key, 30, 2, false); /// Get [30, 31] - auto s1 = fromHolder(holder1); - ASSERT_EQ(s1.size(), 1); - assertRange(26, s1[0], DB::FileSegment::Range(30, 31), DB::FileSegment::State::EMPTY); - - ASSERT_TRUE(s5[0]->getOrSetDownloader() == DB::FileSegment::getCallerId()); - ASSERT_TRUE(s1[0]->getOrSetDownloader() == DB::FileSegment::getCallerId()); - prepareAndDownload(s5[0]); - prepareAndDownload(s1[0]); - s5[0]->completeWithState(DB::FileSegment::State::DOWNLOADED); - s1[0]->completeWithState(DB::FileSegment::State::DOWNLOADED); - - /// Current cache: [___] [_][___][_] [__] - /// ^ ^ ^ ^ ^ ^ ^ ^ - /// 2 4 23 24 26 27 30 31 - - auto holder2 = cache.getOrSet(key, 23, 1, false); /// Get [23, 23] - auto s2 = fromHolder(holder2); - ASSERT_EQ(s2.size(), 1); - - auto holder3 = cache.getOrSet(key, 24, 3, false); /// Get [24, 26] - auto s3 = fromHolder(holder3); - ASSERT_EQ(s3.size(), 1); - - auto holder4 = cache.getOrSet(key, 27, 1, false); /// Get [27, 27] - auto s4 = fromHolder(holder4); - ASSERT_EQ(s4.size(), 1); - - /// All cache is now unreleasable because pointers are still hold - auto holder6 = cache.getOrSet(key, 0, 40, false); - auto f = fromHolder(holder6); - ASSERT_EQ(f.size(), 9); - - assertRange(27, f[0], DB::FileSegment::Range(0, 1), DB::FileSegment::State::EMPTY); - assertRange(28, f[2], DB::FileSegment::Range(5, 22), DB::FileSegment::State::EMPTY); - assertRange(29, f[6], DB::FileSegment::Range(28, 29), DB::FileSegment::State::EMPTY); - assertRange(30, f[8], DB::FileSegment::Range(32, 39), DB::FileSegment::State::EMPTY); - - ASSERT_TRUE(f[0]->getOrSetDownloader() == DB::FileSegment::getCallerId()); - ASSERT_TRUE(f[2]->getOrSetDownloader() == DB::FileSegment::getCallerId()); - ASSERT_TRUE(f[6]->getOrSetDownloader() == DB::FileSegment::getCallerId()); - ASSERT_TRUE(f[8]->getOrSetDownloader() == DB::FileSegment::getCallerId()); - - ASSERT_FALSE(f[0]->reserve(1)); - ASSERT_FALSE(f[2]->reserve(1)); - ASSERT_FALSE(f[6]->reserve(1)); - ASSERT_FALSE(f[8]->reserve(1)); - } - - { - auto holder = cache.getOrSet(key, 2, 3, false); /// Get [2, 4] - auto segments = fromHolder(holder); - ASSERT_EQ(segments.size(), 1); - assertRange(31, segments[0], DB::FileSegment::Range(2, 4), DB::FileSegment::State::DOWNLOADED); - } - - /// Current cache: [___] [_][___][_] [__] - /// ^ ^ ^ ^ ^ ^ ^ ^ - /// 2 4 23 24 26 27 30 31 - - { - auto holder = cache.getOrSet(key, 25, 5, false); /// Get [25, 29] - auto segments = fromHolder(holder); - ASSERT_EQ(segments.size(), 3); - - assertRange(32, segments[0], DB::FileSegment::Range(24, 26), DB::FileSegment::State::DOWNLOADED); - assertRange(33, segments[1], DB::FileSegment::Range(27, 27), DB::FileSegment::State::DOWNLOADED); - - assertRange(34, segments[2], DB::FileSegment::Range(28, 29), DB::FileSegment::State::EMPTY); - ASSERT_TRUE(segments[2]->getOrSetDownloader() == DB::FileSegment::getCallerId()); - ASSERT_TRUE(segments[2]->state() == DB::FileSegment::State::DOWNLOADING); - - bool lets_start_download = false; - std::mutex mutex; - std::condition_variable cv; - - std::thread other_1([&] - { - DB::ThreadStatus thread_status_1; - auto query_context_1 = DB::Context::createCopy(getContext().context); - query_context_1->makeQueryContext(); - query_context_1->setCurrentQueryId("query_id_1"); - DB::CurrentThread::QueryScope query_scope_holder_1(query_context_1); - thread_status_1.attachQueryContext(query_context_1); - - auto holder_2 = cache.getOrSet(key, 25, 5, false); /// Get [25, 29] once again. - auto segments_2 = fromHolder(holder_2); - ASSERT_EQ(segments.size(), 3); - - assertRange(35, segments_2[0], DB::FileSegment::Range(24, 26), DB::FileSegment::State::DOWNLOADED); - assertRange(36, segments_2[1], DB::FileSegment::Range(27, 27), DB::FileSegment::State::DOWNLOADED); - assertRange(37, segments_2[2], DB::FileSegment::Range(28, 29), DB::FileSegment::State::DOWNLOADING); - - ASSERT_TRUE(segments[2]->getOrSetDownloader() != DB::FileSegment::getCallerId()); - ASSERT_TRUE(segments[2]->state() == DB::FileSegment::State::DOWNLOADING); - - { - std::lock_guard lock(mutex); - lets_start_download = true; - } - cv.notify_one(); - - segments_2[2]->wait(); - ASSERT_TRUE(segments_2[2]->state() == DB::FileSegment::State::DOWNLOADED); - }); - - { - std::unique_lock lock(mutex); - cv.wait(lock, [&]{ return lets_start_download; }); - } - - prepareAndDownload(segments[2]); - segments[2]->completeWithState(DB::FileSegment::State::DOWNLOADED); - ASSERT_TRUE(segments[2]->state() == DB::FileSegment::State::DOWNLOADED); - - other_1.join(); - } - - /// Current cache: [___] [___][_][__][__] - /// ^ ^ ^ ^ ^^ ^^ ^ - /// 2 4 24 26 27 2930 31 - - { - /// Now let's check the similar case but getting ERROR state after segment->wait(), when - /// state is changed not manually via segment->complete(state) but from destructor of holder - /// and notify_all() is also called from destructor of holder. - - std::optional holder; - holder.emplace(cache.getOrSet(key, 3, 23, false)); /// Get [3, 25] - - auto segments = fromHolder(*holder); - ASSERT_EQ(segments.size(), 3); - - assertRange(38, segments[0], DB::FileSegment::Range(2, 4), DB::FileSegment::State::DOWNLOADED); - - assertRange(39, segments[1], DB::FileSegment::Range(5, 23), DB::FileSegment::State::EMPTY); - ASSERT_TRUE(segments[1]->getOrSetDownloader() == DB::FileSegment::getCallerId()); - ASSERT_TRUE(segments[1]->state() == DB::FileSegment::State::DOWNLOADING); - - assertRange(40, segments[2], DB::FileSegment::Range(24, 26), DB::FileSegment::State::DOWNLOADED); - - bool lets_start_download = false; - std::mutex mutex; - std::condition_variable cv; - - std::thread other_1([&] - { - DB::ThreadStatus thread_status_1; - auto query_context_1 = DB::Context::createCopy(getContext().context); - query_context_1->makeQueryContext(); - query_context_1->setCurrentQueryId("query_id_1"); - DB::CurrentThread::QueryScope query_scope_holder_1(query_context_1); - thread_status_1.attachQueryContext(query_context_1); - - auto holder_2 = cache.getOrSet(key, 3, 23, false); /// Get [3, 25] once again - auto segments_2 = fromHolder(*holder); - ASSERT_EQ(segments_2.size(), 3); - - assertRange(41, segments_2[0], DB::FileSegment::Range(2, 4), DB::FileSegment::State::DOWNLOADED); - assertRange(42, segments_2[1], DB::FileSegment::Range(5, 23), DB::FileSegment::State::DOWNLOADING); - assertRange(43, segments_2[2], DB::FileSegment::Range(24, 26), DB::FileSegment::State::DOWNLOADED); - - ASSERT_TRUE(segments_2[1]->getDownloader() != DB::FileSegment::getCallerId()); - ASSERT_TRUE(segments_2[1]->state() == DB::FileSegment::State::DOWNLOADING); - - { - std::lock_guard lock(mutex); - lets_start_download = true; - } - cv.notify_one(); - - segments_2[1]->wait(); - printRanges(segments_2); - ASSERT_TRUE(segments_2[1]->state() == DB::FileSegment::State::PARTIALLY_DOWNLOADED); - - ASSERT_TRUE(segments_2[1]->getOrSetDownloader() == DB::FileSegment::getCallerId()); - prepareAndDownload(segments_2[1]); - segments_2[1]->completeWithState(DB::FileSegment::State::DOWNLOADED); - }); - - { - std::unique_lock lock(mutex); - cv.wait(lock, [&]{ return lets_start_download; }); - } - - holder.reset(); - other_1.join(); - printRanges(segments); - ASSERT_TRUE(segments[1]->state() == DB::FileSegment::State::DOWNLOADED); - } - - /// Current cache: [___][ ][___][_][__] - /// ^ ^^ ^ ^^ ^ ^ - /// 2 45 24 2627 28 29 - - { - /// Test LRUCache::restore(). - - auto cache2 = DB::FileCache(cache_base_path, settings); - cache2.initialize(); - - auto holder1 = cache2.getOrSet(key, 2, 28, false); /// Get [2, 29] - - auto segments1 = fromHolder(holder1); - ASSERT_EQ(segments1.size(), 5); - - assertRange(44, segments1[0], DB::FileSegment::Range(2, 4), DB::FileSegment::State::DOWNLOADED); - assertRange(45, segments1[1], DB::FileSegment::Range(5, 23), DB::FileSegment::State::DOWNLOADED); - assertRange(45, segments1[2], DB::FileSegment::Range(24, 26), DB::FileSegment::State::DOWNLOADED); - assertRange(46, segments1[3], DB::FileSegment::Range(27, 27), DB::FileSegment::State::DOWNLOADED); - assertRange(47, segments1[4], DB::FileSegment::Range(28, 29), DB::FileSegment::State::DOWNLOADED); - } - - { - /// Test max file segment size - - auto settings2 = settings; - settings2.max_file_segment_size = 10; - auto cache2 = DB::FileCache(caches_dir / "cache2", settings2); - cache2.initialize(); - - auto holder1 = cache2.getOrSet(key, 0, 25, false); /// Get [0, 24] - auto segments1 = fromHolder(holder1); - - ASSERT_EQ(segments1.size(), 3); - assertRange(48, segments1[0], DB::FileSegment::Range(0, 9), DB::FileSegment::State::EMPTY); - assertRange(49, segments1[1], DB::FileSegment::Range(10, 19), DB::FileSegment::State::EMPTY); - assertRange(50, segments1[2], DB::FileSegment::Range(20, 24), DB::FileSegment::State::EMPTY); - } - -} diff --git a/src/Compression/fuzzers/encrypted_decompress_fuzzer.cpp b/src/Compression/fuzzers/encrypted_decompress_fuzzer.cpp index 9e4d4a3241f..3e3d0e164fe 100644 --- a/src/Compression/fuzzers/encrypted_decompress_fuzzer.cpp +++ b/src/Compression/fuzzers/encrypted_decompress_fuzzer.cpp @@ -58,7 +58,7 @@ Fuzzing data consists of: else: read_key() if (7): - read_nonce (simillar to read_key) + read_nonce (similar to read_key) if (8): set current_key diff --git a/src/Coordination/KeeperSnapshotManager.h b/src/Coordination/KeeperSnapshotManager.h index 4984e54f15f..c00ce9421e7 100644 --- a/src/Coordination/KeeperSnapshotManager.h +++ b/src/Coordination/KeeperSnapshotManager.h @@ -27,7 +27,7 @@ enum SnapshotVersion : uint8_t static constexpr auto CURRENT_SNAPSHOT_VERSION = SnapshotVersion::V5; -/// What is stored in binary shapsnot +/// What is stored in binary snapshot struct SnapshotDeserializationResult { /// Storage diff --git a/src/Coordination/KeeperStorage.cpp b/src/Coordination/KeeperStorage.cpp index 397cd2c0c71..711f3874868 100644 --- a/src/Coordination/KeeperStorage.cpp +++ b/src/Coordination/KeeperStorage.cpp @@ -2192,7 +2192,7 @@ void KeeperStorage::rollbackRequest(int64_t rollback_zxid, bool allow_missing) } catch (...) { - LOG_FATAL(&Poco::Logger::get("KeeperStorage"), "Failed to rollback log. Terminating to avoid incosistencies"); + LOG_FATAL(&Poco::Logger::get("KeeperStorage"), "Failed to rollback log. Terminating to avoid inconsistencies"); std::terminate(); } } diff --git a/src/Coordination/SessionExpiryQueue.h b/src/Coordination/SessionExpiryQueue.h index 8581800834d..862ec35e2f6 100644 --- a/src/Coordination/SessionExpiryQueue.h +++ b/src/Coordination/SessionExpiryQueue.h @@ -53,7 +53,7 @@ public: /// Session was actually removed bool remove(int64_t session_id); - /// Update session expiry time (must be called on hearbeats) + /// Update session expiry time (must be called on heartbeats) void addNewSessionOrUpdate(int64_t session_id, int64_t timeout_ms); /// Get all expired sessions diff --git a/src/Coordination/tests/gtest_coordination.cpp b/src/Coordination/tests/gtest_coordination.cpp index 5474ddefbdb..77a5552fca1 100644 --- a/src/Coordination/tests/gtest_coordination.cpp +++ b/src/Coordination/tests/gtest_coordination.cpp @@ -1340,7 +1340,7 @@ void testLogAndStateMachine(Coordination::CoordinationSettingsPtr settings, uint nuraft::async_result::handler_type when_done = [&snapshot_created] (bool & ret, nuraft::ptr &/*exception*/) { snapshot_created = ret; - std::cerr << "Snapshot finised\n"; + std::cerr << "Snapshot finished\n"; }; state_machine->create_snapshot(s, when_done); diff --git a/src/Core/Settings.h b/src/Core/Settings.h index af32c15a867..1a2b9e42a25 100644 --- a/src/Core/Settings.h +++ b/src/Core/Settings.h @@ -149,7 +149,7 @@ static constexpr UInt64 operator""_GiB(unsigned long long value) \ M(UInt64, parallel_distributed_insert_select, 0, "Process distributed INSERT SELECT query in the same cluster on local tables on every shard; if set to 1 - SELECT is executed on each shard; if set to 2 - SELECT and INSERT are executed on each shard", 0) \ M(UInt64, distributed_group_by_no_merge, 0, "If 1, Do not merge aggregation states from different servers for distributed queries (shards will process query up to the Complete stage, initiator just proxies the data from the shards). If 2 the initiator will apply ORDER BY and LIMIT stages (it is not in case when shard process query up to the Complete stage)", 0) \ - M(UInt64, distributed_push_down_limit, 1, "If 1, LIMIT will be applied on each shard separatelly. Usually you don't need to use it, since this will be done automatically if it is possible, i.e. for simple query SELECT FROM LIMIT.", 0) \ + M(UInt64, distributed_push_down_limit, 1, "If 1, LIMIT will be applied on each shard separately. Usually you don't need to use it, since this will be done automatically if it is possible, i.e. for simple query SELECT FROM LIMIT.", 0) \ M(Bool, optimize_distributed_group_by_sharding_key, true, "Optimize GROUP BY sharding_key queries (by avoiding costly aggregation on the initiator server).", 0) \ M(UInt64, optimize_skip_unused_shards_limit, 1000, "Limit for number of sharding key values, turns off optimize_skip_unused_shards if the limit is reached", 0) \ M(Bool, optimize_skip_unused_shards, false, "Assumes that data is distributed by sharding_key. Optimization to skip unused shards if SELECT query filters by sharding_key.", 0) \ @@ -346,7 +346,7 @@ static constexpr UInt64 operator""_GiB(unsigned long long value) M(UInt64, max_temporary_non_const_columns, 0, "Similar to the 'max_temporary_columns' setting but applies only to non-constant columns. This makes sense, because constant columns are cheap and it is reasonable to allow more of them.", 0) \ \ M(UInt64, max_subquery_depth, 100, "If a query has more than specified number of nested subqueries, throw an exception. This allows you to have a sanity check to protect the users of your cluster from going insane with their queries.", 0) \ - M(UInt64, max_pipeline_depth, 1000, "If a query has more than specified stages in the query pipeline, throw an exception. Pipeline has stages for every relational operator. This allows to limit the complexity of the queries.", 0) \ + M(UInt64, max_analyze_depth, 5000, "Maximum number of analyses performed by interpreter.", 0) \ M(UInt64, max_ast_depth, 1000, "Maximum depth of query syntax tree. Checked after parsing.", 0) \ M(UInt64, max_ast_elements, 50000, "Maximum size of query syntax tree in number of nodes. Checked after parsing.", 0) \ M(UInt64, max_expanded_ast_elements, 500000, "Maximum size of query syntax tree in number of nodes after expansion of aliases and the asterisk.", 0) \ @@ -366,6 +366,8 @@ static constexpr UInt64 operator""_GiB(unsigned long long value) M(UInt64, partial_merge_join_left_table_buffer_bytes, 0, "If not 0 group left table blocks in bigger ones for left-side table in partial merge join. It uses up to 2x of specified memory per joining thread.", 0) \ M(UInt64, partial_merge_join_rows_in_right_blocks, 65536, "Split right-hand joining data in blocks of specified size. It's a portion of data indexed by min-max values and possibly unloaded on disk.", 0) \ M(UInt64, join_on_disk_max_files_to_merge, 64, "For MergeJoin on disk set how much files it's allowed to sort simultaneously. Then this value bigger then more memory used and then less disk I/O needed. Minimum is 2.", 0) \ + M(UInt64, max_rows_in_set_to_optimize_join, 100'000, "Maximal size of the set to filter joined tables by each other row sets before joining. 0 - disable.", 0) \ + \ M(Bool, compatibility_ignore_collation_in_create_table, true, "Compatibility ignore collation in create table", 0) \ \ M(String, temporary_files_codec, "LZ4", "Set compression codec for temporary files (sort and join on disk). I.e. LZ4, NONE.", 0) \ @@ -554,7 +556,9 @@ static constexpr UInt64 operator""_GiB(unsigned long long value) M(UInt64, external_storage_connect_timeout_sec, DBMS_DEFAULT_CONNECT_TIMEOUT_SEC, "Connect timeout in seconds. Now supported only for MySQL", 0) \ M(UInt64, external_storage_rw_timeout_sec, DBMS_DEFAULT_RECEIVE_TIMEOUT_SEC, "Read/write timeout in seconds. Now supported only for MySQL", 0) \ \ - M(UnionMode, union_default_mode, UnionMode::Unspecified, "Set default Union Mode in SelectWithUnion query. Possible values: empty string, 'ALL', 'DISTINCT'. If empty, query without Union Mode will throw exception.", 0) \ + M(SetOperationMode, union_default_mode, SetOperationMode::Unspecified, "Set default mode in UNION query. Possible values: empty string, 'ALL', 'DISTINCT'. If empty, query without mode will throw exception.", 0) \ + M(SetOperationMode, intersect_default_mode, SetOperationMode::ALL, "Set default mode in INTERSECT query. Possible values: empty string, 'ALL', 'DISTINCT'. If empty, query without mode will throw exception.", 0) \ + M(SetOperationMode, except_default_mode, SetOperationMode::ALL, "Set default mode in EXCEPT query. Possible values: empty string, 'ALL', 'DISTINCT'. If empty, query without mode will throw exception.", 0) \ M(Bool, optimize_aggregators_of_group_by_keys, true, "Eliminates min/max/any/anyLast aggregators of GROUP BY keys in SELECT section", 0) \ M(Bool, optimize_group_by_function_keys, true, "Eliminates functions of other keys in GROUP BY section", 0) \ M(Bool, legacy_column_name_of_tuple_literal, false, "List all names of element of large tuple literals in their column names instead of hash. This settings exists only for compatibility reasons. It makes sense to set to 'true', while doing rolling update of cluster from version lower than 21.7 to higher.", 0) \ @@ -672,6 +676,8 @@ static constexpr UInt64 operator""_GiB(unsigned long long value) MAKE_OBSOLETE(M, UInt64, background_message_broker_schedule_pool_size, 16) \ MAKE_OBSOLETE(M, UInt64, background_distributed_schedule_pool_size, 16) \ MAKE_OBSOLETE(M, DefaultDatabaseEngine, default_database_engine, DefaultDatabaseEngine::Atomic) \ + MAKE_OBSOLETE(M, UInt64, max_pipeline_depth, 0) \ + /** The section above is for obsolete settings. Do not add anything there. */ diff --git a/src/Core/SettingsChangesHistory.h b/src/Core/SettingsChangesHistory.h index 8d0e69f4b29..be2def2c01a 100644 --- a/src/Core/SettingsChangesHistory.h +++ b/src/Core/SettingsChangesHistory.h @@ -89,7 +89,7 @@ static std::map sett {"22.3", {{"cast_ipv4_ipv6_default_on_conversion_error", true, false, "Make functions cast(value, 'IPv4') and cast(value, 'IPv6') behave same as toIPv4 and toIPv6 functions"}}}, {"21.12", {{"stream_like_engine_allow_direct_select", true, false, "Do not allow direct select for Kafka/RabbitMQ/FileLog by default"}}}, {"21.9", {{"output_format_decimal_trailing_zeros", true, false, "Do not output trailing zeros in text representation of Decimal types by default for better looking output"}, - {"use_hedged_requests", false, true, "Enable Hedged Requests feature bu default"}}}, + {"use_hedged_requests", false, true, "Enable Hedged Requests feature by default"}}}, {"21.7", {{"legacy_column_name_of_tuple_literal", true, false, "Add this setting only for compatibility reasons. It makes sense to set to 'true', while doing rolling update of cluster from version lower than 21.7 to higher"}}}, {"21.5", {{"async_socket_for_remote", false, true, "Fix all problems and turn on asynchronous reads from socket for remote queries by default again"}}}, {"21.3", {{"async_socket_for_remote", true, false, "Turn off asynchronous reads from socket for remote queries because of some problems"}, diff --git a/src/Core/SettingsEnums.cpp b/src/Core/SettingsEnums.cpp index 616026520db..daa678c0141 100644 --- a/src/Core/SettingsEnums.cpp +++ b/src/Core/SettingsEnums.cpp @@ -114,10 +114,10 @@ IMPLEMENT_SETTING_MULTI_ENUM(MySQLDataTypesSupport, ErrorCodes::UNKNOWN_MYSQL_DA {"date2Date32", MySQLDataTypesSupport::DATE2DATE32}, {"date2String", MySQLDataTypesSupport::DATE2STRING}}) -IMPLEMENT_SETTING_ENUM(UnionMode, ErrorCodes::UNKNOWN_UNION, - {{"", UnionMode::Unspecified}, - {"ALL", UnionMode::ALL}, - {"DISTINCT", UnionMode::DISTINCT}}) +IMPLEMENT_SETTING_ENUM(SetOperationMode, ErrorCodes::UNKNOWN_UNION, + {{"", SetOperationMode::Unspecified}, + {"ALL", SetOperationMode::ALL}, + {"DISTINCT", SetOperationMode::DISTINCT}}) IMPLEMENT_SETTING_ENUM(DistributedDDLOutputMode, ErrorCodes::BAD_ARGUMENTS, {{"none", DistributedDDLOutputMode::NONE}, diff --git a/src/Core/SettingsEnums.h b/src/Core/SettingsEnums.h index 308d53ff690..b5e908defc7 100644 --- a/src/Core/SettingsEnums.h +++ b/src/Core/SettingsEnums.h @@ -130,14 +130,14 @@ enum class MySQLDataTypesSupport DECLARE_SETTING_MULTI_ENUM(MySQLDataTypesSupport) -enum class UnionMode +enum class SetOperationMode { - Unspecified = 0, // Query UNION without UnionMode will throw exception - ALL, // Query UNION without UnionMode -> SELECT ... UNION ALL SELECT ... - DISTINCT // Query UNION without UnionMode -> SELECT ... UNION DISTINCT SELECT ... + Unspecified = 0, // Query UNION / EXCEPT / INTERSECT without SetOperationMode will throw exception + ALL, // Query UNION / EXCEPT / INTERSECT without SetOperationMode -> SELECT ... UNION / EXCEPT / INTERSECT ALL SELECT ... + DISTINCT // Query UNION / EXCEPT / INTERSECT without SetOperationMode -> SELECT ... UNION / EXCEPT / INTERSECT DISTINCT SELECT ... }; -DECLARE_SETTING_ENUM(UnionMode) +DECLARE_SETTING_ENUM(SetOperationMode) enum class DistributedDDLOutputMode { @@ -153,7 +153,7 @@ enum class HandleKafkaErrorMode { DEFAULT = 0, // Ignore errors with threshold. STREAM, // Put errors to stream in the virtual column named ``_error. - /*FIXED_SYSTEM_TABLE, Put errors to in a fixed system table likey system.kafka_errors. This is not implemented now. */ + /*FIXED_SYSTEM_TABLE, Put errors to in a fixed system table likely system.kafka_errors. This is not implemented now. */ /*CUSTOM_SYSTEM_TABLE, Put errors to in a custom system table. This is not implemented now. */ }; diff --git a/src/DataTypes/Serializations/SerializationObject.cpp b/src/DataTypes/Serializations/SerializationObject.cpp index 7ad7706ae1e..b893407e7a5 100644 --- a/src/DataTypes/Serializations/SerializationObject.cpp +++ b/src/DataTypes/Serializations/SerializationObject.cpp @@ -260,7 +260,12 @@ void SerializationObject::serializeBinaryBulkWithMultipleStreams( auto * state_object = checkAndGetState(state); if (!column_object.isFinalized()) - throw Exception(ErrorCodes::LOGICAL_ERROR, "Cannot write non-finalized ColumnObject"); + { + auto finalized_object = column_object.clone(); + assert_cast(*finalized_object).finalize(); + serializeBinaryBulkWithMultipleStreams(*finalized_object, offset, limit, settings, state); + return; + } auto [tuple_column, tuple_type] = unflattenObjectToTuple(column_object); diff --git a/src/Dictionaries/HierarchyDictionariesUtils.h b/src/Dictionaries/HierarchyDictionariesUtils.h index 621290f40f9..c7508ddd220 100644 --- a/src/Dictionaries/HierarchyDictionariesUtils.h +++ b/src/Dictionaries/HierarchyDictionariesUtils.h @@ -180,7 +180,7 @@ namespace detail /** Returns array with UInt8 represent if key from in_keys array is in hierarchy of key from keys column. * If value in result array is 1 that means key from in_keys array is in hierarchy of key from - * keys array with same index, 0 therwise. + * keys array with same index, 0 otherwise. * For getting hierarchy implementation uses getKeysHierarchy function. * * Not: keys size must be equal to in_keys_size. diff --git a/src/Dictionaries/MongoDBDictionarySource.cpp b/src/Dictionaries/MongoDBDictionarySource.cpp index 1ede0ec5045..a735f426ec7 100644 --- a/src/Dictionaries/MongoDBDictionarySource.cpp +++ b/src/Dictionaries/MongoDBDictionarySource.cpp @@ -118,7 +118,7 @@ MongoDBDictionarySource::MongoDBDictionarySource( Poco::URI poco_uri(uri); // Parse database from URI. This is required for correctness -- the - // cursor is created using database name and colleciton name, so we have + // cursor is created using database name and collection name, so we have // to specify them properly. db = poco_uri.getPath(); // getPath() may return a leading slash, remove it. diff --git a/src/Dictionaries/getDictionaryConfigurationFromAST.cpp b/src/Dictionaries/getDictionaryConfigurationFromAST.cpp index edc3c34fe81..7d8253c47ce 100644 --- a/src/Dictionaries/getDictionaryConfigurationFromAST.cpp +++ b/src/Dictionaries/getDictionaryConfigurationFromAST.cpp @@ -244,7 +244,7 @@ void buildAttributeExpressionIfNeeded( root->appendChild(expression_element); } -/** Transofrms single dictionary attribute to configuration +/** Transforms single dictionary attribute to configuration * third_column UInt8 DEFAULT 2 EXPRESSION rand() % 100 * 77 * to * diff --git a/src/Disks/IO/CachedOnDiskReadBufferFromFile.h b/src/Disks/IO/CachedOnDiskReadBufferFromFile.h index 365496c6d06..ed623272c12 100644 --- a/src/Disks/IO/CachedOnDiskReadBufferFromFile.h +++ b/src/Disks/IO/CachedOnDiskReadBufferFromFile.h @@ -1,13 +1,13 @@ #pragma once -#include +#include #include #include #include #include #include #include -#include +#include namespace CurrentMetrics diff --git a/src/Disks/IO/CachedOnDiskWriteBufferFromFile.cpp b/src/Disks/IO/CachedOnDiskWriteBufferFromFile.cpp index 21c120fd4c8..56a21307c36 100644 --- a/src/Disks/IO/CachedOnDiskWriteBufferFromFile.cpp +++ b/src/Disks/IO/CachedOnDiskWriteBufferFromFile.cpp @@ -1,7 +1,7 @@ #include "CachedOnDiskWriteBufferFromFile.h" -#include -#include +#include +#include #include #include #include diff --git a/src/Disks/IO/CachedOnDiskWriteBufferFromFile.h b/src/Disks/IO/CachedOnDiskWriteBufferFromFile.h index abc50078874..fa861fea14b 100644 --- a/src/Disks/IO/CachedOnDiskWriteBufferFromFile.h +++ b/src/Disks/IO/CachedOnDiskWriteBufferFromFile.h @@ -2,7 +2,7 @@ #include #include -#include +#include #include namespace Poco diff --git a/src/Disks/ObjectStorages/Cached/CachedObjectStorage.cpp b/src/Disks/ObjectStorages/Cached/CachedObjectStorage.cpp index a68f9fdaf2d..7e829847846 100644 --- a/src/Disks/ObjectStorages/Cached/CachedObjectStorage.cpp +++ b/src/Disks/ObjectStorages/Cached/CachedObjectStorage.cpp @@ -4,8 +4,8 @@ #include #include #include -#include -#include +#include +#include #include #include #include diff --git a/src/Disks/ObjectStorages/Cached/CachedObjectStorage.h b/src/Disks/ObjectStorages/Cached/CachedObjectStorage.h index 258b70ed579..68ded61a9f1 100644 --- a/src/Disks/ObjectStorages/Cached/CachedObjectStorage.h +++ b/src/Disks/ObjectStorages/Cached/CachedObjectStorage.h @@ -1,8 +1,8 @@ #pragma once #include -#include -#include +#include +#include namespace Poco { diff --git a/src/Disks/ObjectStorages/Cached/registerDiskCache.cpp b/src/Disks/ObjectStorages/Cached/registerDiskCache.cpp index 84b95bd3d16..902ebd0fcc8 100644 --- a/src/Disks/ObjectStorages/Cached/registerDiskCache.cpp +++ b/src/Disks/ObjectStorages/Cached/registerDiskCache.cpp @@ -1,6 +1,6 @@ -#include -#include -#include +#include +#include +#include #include #include #include diff --git a/src/Disks/ObjectStorages/DiskObjectStorage.cpp b/src/Disks/ObjectStorages/DiskObjectStorage.cpp index bac0d8e4c12..00ef01645cd 100644 --- a/src/Disks/ObjectStorages/DiskObjectStorage.cpp +++ b/src/Disks/ObjectStorages/DiskObjectStorage.cpp @@ -10,7 +10,6 @@ #include #include #include -#include #include #include #include diff --git a/src/Disks/ObjectStorages/DiskObjectStorage.h b/src/Disks/ObjectStorages/DiskObjectStorage.h index 6e2d719e3ad..34056f17b3c 100644 --- a/src/Disks/ObjectStorages/DiskObjectStorage.h +++ b/src/Disks/ObjectStorages/DiskObjectStorage.h @@ -2,7 +2,6 @@ #include #include -#include #include #include #include diff --git a/src/Disks/ObjectStorages/DiskObjectStorageTransaction.cpp b/src/Disks/ObjectStorages/DiskObjectStorageTransaction.cpp index 54d5a2f2368..6aabf300cb8 100644 --- a/src/Disks/ObjectStorages/DiskObjectStorageTransaction.cpp +++ b/src/Disks/ObjectStorages/DiskObjectStorageTransaction.cpp @@ -3,6 +3,7 @@ #include #include #include +#include namespace DB @@ -633,9 +634,11 @@ void DiskObjectStorageTransaction::commit() { operations_to_execute[i]->execute(metadata_transaction); } - catch (Exception & ex) + catch (...) { - ex.addMessage(fmt::format("While executing operation #{} ({})", i, operations_to_execute[i]->getInfoForLog())); + tryLogCurrentException( + &Poco::Logger::get("DiskObjectStorageTransaction"), + fmt::format("An error occurred while executing transaction's operation #{} ({})", i, operations_to_execute[i]->getInfoForLog())); for (int64_t j = i; j >= 0; --j) { @@ -643,9 +646,12 @@ void DiskObjectStorageTransaction::commit() { operations_to_execute[j]->undo(); } - catch (Exception & rollback_ex) + catch (...) { - rollback_ex.addMessage(fmt::format("While undoing operation #{}", i)); + tryLogCurrentException( + &Poco::Logger::get("DiskObjectStorageTransaction"), + fmt::format("An error occurred while undoing transaction's operation #{}", i)); + throw; } } diff --git a/src/Disks/ObjectStorages/IMetadataStorage.h b/src/Disks/ObjectStorages/IMetadataStorage.h index 300d8ec59b9..3d6c772157d 100644 --- a/src/Disks/ObjectStorages/IMetadataStorage.h +++ b/src/Disks/ObjectStorages/IMetadataStorage.h @@ -124,7 +124,7 @@ public: virtual ~IMetadataStorage() = default; - /// ==== More specefic methods. Previous were almost general purpose. ==== + /// ==== More specific methods. Previous were almost general purpose. ==== /// Read multiple metadata files into strings and return mapping from file_path -> metadata virtual std::unordered_map getSerializedMetadata(const std::vector & file_paths) const = 0; diff --git a/src/Disks/ObjectStorages/IObjectStorage.h b/src/Disks/ObjectStorages/IObjectStorage.h index 89d0e0d6d03..dc6683cfe95 100644 --- a/src/Disks/ObjectStorages/IObjectStorage.h +++ b/src/Disks/ObjectStorages/IObjectStorage.h @@ -17,7 +17,6 @@ #include #include #include -#include #include diff --git a/src/Disks/ObjectStorages/LocalObjectStorage.cpp b/src/Disks/ObjectStorages/LocalObjectStorage.cpp index 64512a07919..dbb3a7c2aba 100644 --- a/src/Disks/ObjectStorages/LocalObjectStorage.cpp +++ b/src/Disks/ObjectStorages/LocalObjectStorage.cpp @@ -1,8 +1,6 @@ #include #include -#include -#include #include #include #include diff --git a/src/Disks/ObjectStorages/S3/diskSettings.cpp b/src/Disks/ObjectStorages/S3/diskSettings.cpp index 4cdb867e48c..b34eddf63f0 100644 --- a/src/Disks/ObjectStorages/S3/diskSettings.cpp +++ b/src/Disks/ObjectStorages/S3/diskSettings.cpp @@ -22,8 +22,6 @@ #include #include -#include - namespace DB { diff --git a/src/Disks/ObjectStorages/S3/parseConfig.h b/src/Disks/ObjectStorages/S3/parseConfig.h index 725c86ce462..97fafd620f9 100644 --- a/src/Disks/ObjectStorages/S3/parseConfig.h +++ b/src/Disks/ObjectStorages/S3/parseConfig.h @@ -6,7 +6,6 @@ #include #include -#include #include #include #include @@ -14,7 +13,6 @@ #include #include #include -#include namespace DB diff --git a/src/Formats/JSONUtils.cpp b/src/Formats/JSONUtils.cpp index 895833abf4d..9aa0645f580 100644 --- a/src/Formats/JSONUtils.cpp +++ b/src/Formats/JSONUtils.cpp @@ -218,7 +218,6 @@ namespace JSONUtils { auto object = field.getObject(); DataTypes value_types; - bool have_object_value = false; for (const auto key_value_pair : object) { auto type = getDataTypeFromFieldImpl(key_value_pair.second, settings, numbers_parsed_from_json_strings); @@ -226,10 +225,7 @@ namespace JSONUtils continue; if (isObject(type)) - { - have_object_value = true; - break; - } + return std::make_shared("json", true); value_types.push_back(type); } @@ -242,7 +238,7 @@ namespace JSONUtils for (size_t i = 1; i < value_types.size(); ++i) are_types_equal &= value_types[i]->equals(*value_types[0]); - if (have_object_value || !are_types_equal) + if (!are_types_equal) return std::make_shared("json", true); return std::make_shared(std::make_shared(), value_types[0]); diff --git a/src/Functions/FunctionHelpers.h b/src/Functions/FunctionHelpers.h index 8d33c820185..18a4e584080 100644 --- a/src/Functions/FunctionHelpers.h +++ b/src/Functions/FunctionHelpers.h @@ -134,7 +134,7 @@ using FunctionArgumentDescriptors = std::vector; * (e.g. depending on result type or other trait). * First, checks that number of arguments is as expected (including optional arguments). * Second, checks that mandatory args present and have valid type. - * Third, checks optional arguents types, skipping ones that are missing. + * Third, checks optional arguments types, skipping ones that are missing. * * Please note that if you have several optional arguments, like f([a, b, c]), * only these calls are considered valid: diff --git a/src/IO/ConcatSeekableReadBuffer.cpp b/src/IO/ConcatSeekableReadBuffer.cpp index c5d48376e2f..0943d1eac45 100644 --- a/src/IO/ConcatSeekableReadBuffer.cpp +++ b/src/IO/ConcatSeekableReadBuffer.cpp @@ -9,6 +9,11 @@ namespace ErrorCodes extern const int ARGUMENT_OUT_OF_BOUND; } +ConcatSeekableReadBuffer::BufferInfo::BufferInfo(BufferInfo && src) noexcept + : in(std::exchange(src.in, nullptr)), own_in(std::exchange(src.own_in, false)), size(std::exchange(src.size, 0)) +{ +} + ConcatSeekableReadBuffer::BufferInfo::~BufferInfo() { if (own_in) diff --git a/src/IO/ConcatSeekableReadBuffer.h b/src/IO/ConcatSeekableReadBuffer.h index 5d7dca82524..c8c16c5d887 100644 --- a/src/IO/ConcatSeekableReadBuffer.h +++ b/src/IO/ConcatSeekableReadBuffer.h @@ -30,7 +30,7 @@ private: struct BufferInfo { BufferInfo() = default; - BufferInfo(BufferInfo &&) = default; + BufferInfo(BufferInfo && src) noexcept; ~BufferInfo(); SeekableReadBuffer * in = nullptr; bool own_in = false; diff --git a/src/IO/HashingReadBuffer.h b/src/IO/HashingReadBuffer.h index 5d42c64478c..a0a029e6f85 100644 --- a/src/IO/HashingReadBuffer.h +++ b/src/IO/HashingReadBuffer.h @@ -18,29 +18,38 @@ public: { working_buffer = in.buffer(); pos = in.position(); + hashing_begin = pos; + } - /// calculate hash from the data already read - if (!working_buffer.empty()) + uint128 getHash() + { + if (pos > hashing_begin) { - calculateHash(pos, working_buffer.end() - pos); + calculateHash(hashing_begin, pos - hashing_begin); + hashing_begin = pos; } + return IHashingBuffer::getHash(); } private: bool nextImpl() override { + if (pos > hashing_begin) + calculateHash(hashing_begin, pos - hashing_begin); + in.position() = pos; bool res = in.next(); working_buffer = in.buffer(); - pos = in.position(); // `pos` may be different from working_buffer.begin() when using sophisticated ReadBuffers. - calculateHash(pos, working_buffer.end() - pos); + pos = in.position(); + hashing_begin = pos; return res; } ReadBuffer & in; + BufferBase::Position hashing_begin; }; } diff --git a/src/IO/ReadSettings.h b/src/IO/ReadSettings.h index e639ecbedc2..fb093c52f67 100644 --- a/src/IO/ReadSettings.h +++ b/src/IO/ReadSettings.h @@ -3,7 +3,7 @@ #include #include #include -#include +#include #include namespace DB diff --git a/src/IO/S3/PocoHTTPClient.cpp b/src/IO/S3/PocoHTTPClient.cpp index 089d89cd8a7..34e7a7f6eb4 100644 --- a/src/IO/S3/PocoHTTPClient.cpp +++ b/src/IO/S3/PocoHTTPClient.cpp @@ -121,9 +121,29 @@ std::shared_ptr PocoHTTPClient::MakeRequest( Aws::Utils::RateLimits::RateLimiterInterface * readLimiter, Aws::Utils::RateLimits::RateLimiterInterface * writeLimiter) const { - auto response = Aws::MakeShared("PocoHTTPClient", request); - makeRequestInternal(*request, response, readLimiter, writeLimiter); - return response; + try + { + auto response = Aws::MakeShared("PocoHTTPClient", request); + makeRequestInternal(*request, response, readLimiter, writeLimiter); + return response; + } + catch (const Exception &) + { + throw; + } + catch (const Poco::Exception & e) + { + throw Exception(Exception::CreateFromPocoTag{}, e); + } + catch (const std::exception & e) + { + throw Exception(Exception::CreateFromSTDTag{}, e); + } + catch (...) + { + tryLogCurrentException(__PRETTY_FUNCTION__); + throw; + } } namespace diff --git a/src/IO/S3Common.cpp b/src/IO/S3Common.cpp index fb9cff5d109..1ff1c609952 100644 --- a/src/IO/S3Common.cpp +++ b/src/IO/S3Common.cpp @@ -24,7 +24,7 @@ # include # include # include -# include // Y_IGNORE +# include # include # include diff --git a/src/IO/WriteBufferFromS3.cpp b/src/IO/WriteBufferFromS3.cpp index 2f2479766d5..2510b86716c 100644 --- a/src/IO/WriteBufferFromS3.cpp +++ b/src/IO/WriteBufferFromS3.cpp @@ -4,7 +4,7 @@ #include #include -#include +#include #include #include @@ -431,7 +431,7 @@ void WriteBufferFromS3::waitForReadyBackGroundTasks() { if (schedule) { - std::lock_guard lock(bg_tasks_mutex); + std::unique_lock lock(bg_tasks_mutex); { while (!upload_object_tasks.empty() && upload_object_tasks.front().is_finised) { @@ -442,7 +442,7 @@ void WriteBufferFromS3::waitForReadyBackGroundTasks() if (exception) { - waitForAllBackGroundTasks(); + waitForAllBackGroundTasksUnlocked(lock); std::rethrow_exception(exception); } @@ -457,7 +457,15 @@ void WriteBufferFromS3::waitForAllBackGroundTasks() if (schedule) { std::unique_lock lock(bg_tasks_mutex); - bg_tasks_condvar.wait(lock, [this]() { return num_added_bg_tasks == num_finished_bg_tasks; }); + waitForAllBackGroundTasksUnlocked(lock); + } +} + +void WriteBufferFromS3::waitForAllBackGroundTasksUnlocked(std::unique_lock & bg_tasks_lock) +{ + if (schedule) + { + bg_tasks_condvar.wait(bg_tasks_lock, [this]() { return num_added_bg_tasks == num_finished_bg_tasks; }); while (!upload_object_tasks.empty()) { @@ -472,7 +480,7 @@ void WriteBufferFromS3::waitForAllBackGroundTasks() if (put_object_task) { - bg_tasks_condvar.wait(lock, [this]() { return put_object_task->is_finised; }); + bg_tasks_condvar.wait(bg_tasks_lock, [this]() { return put_object_task->is_finised; }); if (put_object_task->exception) std::rethrow_exception(put_object_task->exception); } diff --git a/src/IO/WriteBufferFromS3.h b/src/IO/WriteBufferFromS3.h index 99440654910..712044841d0 100644 --- a/src/IO/WriteBufferFromS3.h +++ b/src/IO/WriteBufferFromS3.h @@ -84,6 +84,7 @@ private: void waitForReadyBackGroundTasks(); void waitForAllBackGroundTasks(); + void waitForAllBackGroundTasksUnlocked(std::unique_lock & bg_tasks_lock); String bucket; String key; diff --git a/src/IO/WriteHelpers.h b/src/IO/WriteHelpers.h index f9892ac6194..7acea87c0d7 100644 --- a/src/IO/WriteHelpers.h +++ b/src/IO/WriteHelpers.h @@ -146,14 +146,14 @@ inline size_t writeFloatTextFastPath(T x, char * buffer) /// The library Ryu has low performance on integers. /// This workaround improves performance 6..10 times. - if (DecomposedFloat64(x).is_integer_in_representable_range()) + if (DecomposedFloat64(x).isIntegerInRepresentableRange()) result = itoa(Int64(x), buffer) - buffer; else result = jkj::dragonbox::to_chars_n(x, buffer) - buffer; } else { - if (DecomposedFloat32(x).is_integer_in_representable_range()) + if (DecomposedFloat32(x).isIntegerInRepresentableRange()) result = itoa(Int32(x), buffer) - buffer; else result = jkj::dragonbox::to_chars_n(x, buffer) - buffer; diff --git a/src/Interpreters/AggregationUtils.cpp b/src/Interpreters/AggregationUtils.cpp new file mode 100644 index 00000000000..43062546450 --- /dev/null +++ b/src/Interpreters/AggregationUtils.cpp @@ -0,0 +1,113 @@ +#include + +namespace DB +{ + +namespace ErrorCodes +{ + extern const int LOGICAL_ERROR; +} + +OutputBlockColumns prepareOutputBlockColumns( + const Aggregator::Params & params, + const Aggregator::AggregateFunctionsPlainPtrs & aggregate_functions, + const Block & res_header, + Arenas & aggregates_pools, + bool final, + size_t rows) +{ + MutableColumns key_columns(params.keys_size); + MutableColumns aggregate_columns(params.aggregates_size); + MutableColumns final_aggregate_columns(params.aggregates_size); + Aggregator::AggregateColumnsData aggregate_columns_data(params.aggregates_size); + + for (size_t i = 0; i < params.keys_size; ++i) + { + key_columns[i] = res_header.safeGetByPosition(i).type->createColumn(); + key_columns[i]->reserve(rows); + } + + for (size_t i = 0; i < params.aggregates_size; ++i) + { + if (!final) + { + const auto & aggregate_column_name = params.aggregates[i].column_name; + aggregate_columns[i] = res_header.getByName(aggregate_column_name).type->createColumn(); + + /// The ColumnAggregateFunction column captures the shared ownership of the arena with the aggregate function states. + ColumnAggregateFunction & column_aggregate_func = assert_cast(*aggregate_columns[i]); + + for (auto & pool : aggregates_pools) + column_aggregate_func.addArena(pool); + + aggregate_columns_data[i] = &column_aggregate_func.getData(); + aggregate_columns_data[i]->reserve(rows); + } + else + { + final_aggregate_columns[i] = aggregate_functions[i]->getReturnType()->createColumn(); + final_aggregate_columns[i]->reserve(rows); + + if (aggregate_functions[i]->isState()) + { + /// The ColumnAggregateFunction column captures the shared ownership of the arena with aggregate function states. + if (auto * column_aggregate_func = typeid_cast(final_aggregate_columns[i].get())) + for (auto & pool : aggregates_pools) + column_aggregate_func->addArena(pool); + + /// Aggregate state can be wrapped into array if aggregate function ends with -Resample combinator. + final_aggregate_columns[i]->forEachSubcolumn( + [&aggregates_pools](auto & subcolumn) + { + if (auto * column_aggregate_func = typeid_cast(subcolumn.get())) + for (auto & pool : aggregates_pools) + column_aggregate_func->addArena(pool); + }); + } + } + } + + if (key_columns.size() != params.keys_size) + throw Exception{"Aggregate. Unexpected key columns size.", ErrorCodes::LOGICAL_ERROR}; + + std::vector raw_key_columns; + raw_key_columns.reserve(key_columns.size()); + for (auto & column : key_columns) + raw_key_columns.push_back(column.get()); + + return { + .key_columns = std::move(key_columns), + .raw_key_columns = std::move(raw_key_columns), + .aggregate_columns = std::move(aggregate_columns), + .final_aggregate_columns = std::move(final_aggregate_columns), + .aggregate_columns_data = std::move(aggregate_columns_data), + }; +} + +Block finalizeBlock(const Aggregator::Params & params, const Block & res_header, OutputBlockColumns && out_cols, bool final, size_t rows) +{ + auto && [key_columns, raw_key_columns, aggregate_columns, final_aggregate_columns, aggregate_columns_data] = out_cols; + + Block res = res_header.cloneEmpty(); + + for (size_t i = 0; i < params.keys_size; ++i) + res.getByPosition(i).column = std::move(key_columns[i]); + + for (size_t i = 0; i < params.aggregates_size; ++i) + { + const auto & aggregate_column_name = params.aggregates[i].column_name; + if (final) + res.getByName(aggregate_column_name).column = std::move(final_aggregate_columns[i]); + else + res.getByName(aggregate_column_name).column = std::move(aggregate_columns[i]); + } + + /// Change the size of the columns-constants in the block. + size_t columns = res_header.columns(); + for (size_t i = 0; i < columns; ++i) + if (isColumnConst(*res.getByPosition(i).column)) + res.getByPosition(i).column = res.getByPosition(i).column->cut(0, rows); + + return res; +} +} diff --git a/src/Interpreters/AggregationUtils.h b/src/Interpreters/AggregationUtils.h new file mode 100644 index 00000000000..cc37cec0a69 --- /dev/null +++ b/src/Interpreters/AggregationUtils.h @@ -0,0 +1,27 @@ +#pragma once + +#include + +namespace DB +{ + +struct OutputBlockColumns +{ + MutableColumns key_columns; + std::vector raw_key_columns; + MutableColumns aggregate_columns; + MutableColumns final_aggregate_columns; + Aggregator::AggregateColumnsData aggregate_columns_data; +}; + + +OutputBlockColumns prepareOutputBlockColumns( + const Aggregator::Params & params, + const Aggregator::AggregateFunctionsPlainPtrs & aggregate_functions, + const Block & res_header, + Arenas & aggregates_pools, + bool final, + size_t rows); + +Block finalizeBlock(const Aggregator::Params & params, const Block & res_header, OutputBlockColumns && out_cols, bool final, size_t rows); +} diff --git a/src/Interpreters/Aggregator.cpp b/src/Interpreters/Aggregator.cpp index 3340170f71b..ef55f92f63a 100644 --- a/src/Interpreters/Aggregator.cpp +++ b/src/Interpreters/Aggregator.cpp @@ -34,6 +34,8 @@ #include +#include + namespace ProfileEvents { extern const Event ExternalAggregationWritePart; @@ -1587,16 +1589,10 @@ Block Aggregator::convertOneBucketToBlock( bool final, size_t bucket) const { - Block block = prepareBlockAndFill(data_variants, final, method.data.impls[bucket].size(), - [bucket, &method, arena, this] ( - MutableColumns & key_columns, - AggregateColumnsData & aggregate_columns, - MutableColumns & final_aggregate_columns, - bool final_) - { - convertToBlockImpl(method, method.data.impls[bucket], - key_columns, aggregate_columns, final_aggregate_columns, arena, final_); - }); + // Used in ConvertingAggregatedToChunksSource -> ConvertingAggregatedToChunksTransform (expects single chunk for each bucket_id). + constexpr bool return_single_block = true; + Block block = convertToBlockImpl( + method, method.data.impls[bucket], arena, data_variants.aggregates_pools, final, method.data.impls[bucket].size()); block.info.bucket_num = bucket; return block; @@ -1702,26 +1698,17 @@ bool Aggregator::checkLimits(size_t result_size, bool & no_more_keys) const } -template -void Aggregator::convertToBlockImpl( - Method & method, - Table & data, - MutableColumns & key_columns, - AggregateColumnsData & aggregate_columns, - MutableColumns & final_aggregate_columns, - Arena * arena, - bool final) const +template +Aggregator::ConvertToBlockRes +Aggregator::convertToBlockImpl(Method & method, Table & data, Arena * arena, Arenas & aggregates_pools, bool final, size_t rows) const { if (data.empty()) - return; + { + auto && out_cols = prepareOutputBlockColumns(params, aggregate_functions, getHeader(final), aggregates_pools, final, rows); + return {finalizeBlock(params, getHeader(final), std::move(out_cols), final, rows)}; + } - if (key_columns.size() != params.keys_size) - throw Exception{"Aggregate. Unexpected key columns size.", ErrorCodes::LOGICAL_ERROR}; - - std::vector raw_key_columns; - raw_key_columns.reserve(key_columns.size()); - for (auto & column : key_columns) - raw_key_columns.push_back(column.get()); + ConvertToBlockRes res; if (final) { @@ -1729,20 +1716,23 @@ void Aggregator::convertToBlockImpl( if (compiled_aggregate_functions_holder) { static constexpr bool use_compiled_functions = !Method::low_cardinality_optimization; - convertToBlockImplFinal(method, data, std::move(raw_key_columns), final_aggregate_columns, arena); + res = convertToBlockImplFinal(method, data, arena, aggregates_pools, rows); } else #endif { - convertToBlockImplFinal(method, data, std::move(raw_key_columns), final_aggregate_columns, arena); + res = convertToBlockImplFinal(method, data, arena, aggregates_pools, rows); } } else { - convertToBlockImplNotFinal(method, data, std::move(raw_key_columns), aggregate_columns); + res = convertToBlockImplNotFinal(method, data, aggregates_pools, rows); } + /// In order to release memory early. data.clearAndShrink(); + + return res; } @@ -1811,38 +1801,9 @@ inline void Aggregator::insertAggregatesIntoColumns(Mapped & mapped, MutableColu } -template -void NO_INLINE Aggregator::convertToBlockImplFinal( - Method & method, - Table & data, - std::vector key_columns, - MutableColumns & final_aggregate_columns, - Arena * arena) const +template +Block Aggregator::insertResultsIntoColumns(PaddedPODArray & places, OutputBlockColumns && out_cols, Arena * arena) const { - if constexpr (Method::low_cardinality_optimization) - { - if (data.hasNullKeyData()) - { - key_columns[0]->insertDefault(); - insertAggregatesIntoColumns(data.getNullKeyData(), final_aggregate_columns, arena); - } - } - - auto shuffled_key_sizes = method.shuffleKeyColumns(key_columns, key_sizes); - const auto & key_sizes_ref = shuffled_key_sizes ? *shuffled_key_sizes : key_sizes; - - PaddedPODArray places; - places.reserve(data.size()); - - data.forEachValue([&](const auto & key, auto & mapped) - { - method.insertKeyIntoColumns(key, key_columns, key_sizes_ref); - places.emplace_back(mapped); - - /// Mark the cell as destroyed so it will not be destroyed in destructor. - mapped = nullptr; - }); - std::exception_ptr exception; size_t aggregate_functions_destroy_index = 0; @@ -1863,7 +1824,7 @@ void NO_INLINE Aggregator::convertToBlockImplFinal( if (!is_aggregate_function_compiled[i]) continue; - auto & final_aggregate_column = final_aggregate_columns[i]; + auto & final_aggregate_column = out_cols.final_aggregate_columns[i]; final_aggregate_column = final_aggregate_column->cloneResized(places.size()); columns_data.emplace_back(getColumnData(final_aggregate_column.get())); } @@ -1884,7 +1845,7 @@ void NO_INLINE Aggregator::convertToBlockImplFinal( } } - auto & final_aggregate_column = final_aggregate_columns[aggregate_functions_destroy_index]; + auto & final_aggregate_column = out_cols.final_aggregate_columns[aggregate_functions_destroy_index]; size_t offset = offsets_of_aggregate_states[aggregate_functions_destroy_index]; /** We increase aggregate_functions_destroy_index because by function contract if insertResultIntoBatch @@ -1898,7 +1859,8 @@ void NO_INLINE Aggregator::convertToBlockImplFinal( bool is_state = aggregate_functions[destroy_index]->isState(); bool destroy_place_after_insert = !is_state; - aggregate_functions[destroy_index]->insertResultIntoBatch(0, places.size(), places.data(), offset, *final_aggregate_column, arena, destroy_place_after_insert); + aggregate_functions[destroy_index]->insertResultIntoBatch( + 0, places.size(), places.data(), offset, *final_aggregate_column, arena, destroy_place_after_insert); } } catch (...) @@ -1923,125 +1885,155 @@ void NO_INLINE Aggregator::convertToBlockImplFinal( if (exception) std::rethrow_exception(exception); + + return finalizeBlock(params, getHeader(/* final */ true), std::move(out_cols), /* final */ true, places.size()); } -template -void NO_INLINE Aggregator::convertToBlockImplNotFinal( - Method & method, - Table & data, - std::vector key_columns, - AggregateColumnsData & aggregate_columns) const +template +Aggregator::ConvertToBlockRes NO_INLINE +Aggregator::convertToBlockImplFinal(Method & method, Table & data, Arena * arena, Arenas & aggregates_pools, size_t) const { - if constexpr (Method::low_cardinality_optimization) + const size_t max_block_size = params.max_block_size; + const bool final = true; + ConvertToBlockRes res; + + std::optional out_cols; + std::optional shuffled_key_sizes; + PaddedPODArray places; + + auto init_out_cols = [&]() { - if (data.hasNullKeyData()) + out_cols = prepareOutputBlockColumns(params, aggregate_functions, getHeader(final), aggregates_pools, final, max_block_size); + + if constexpr (Method::low_cardinality_optimization) { - key_columns[0]->insertDefault(); - - for (size_t i = 0; i < params.aggregates_size; ++i) - aggregate_columns[i]->push_back(data.getNullKeyData() + offsets_of_aggregate_states[i]); - - data.getNullKeyData() = nullptr; - } - } - - auto shuffled_key_sizes = method.shuffleKeyColumns(key_columns, key_sizes); - const auto & key_sizes_ref = shuffled_key_sizes ? *shuffled_key_sizes : key_sizes; - - data.forEachValue([&](const auto & key, auto & mapped) - { - method.insertKeyIntoColumns(key, key_columns, key_sizes_ref); - - /// reserved, so push_back does not throw exceptions - for (size_t i = 0; i < params.aggregates_size; ++i) - aggregate_columns[i]->push_back(mapped + offsets_of_aggregate_states[i]); - - mapped = nullptr; - }); -} - - -template -Block Aggregator::prepareBlockAndFill( - AggregatedDataVariants & data_variants, - bool final, - size_t rows, - Filler && filler) const -{ - MutableColumns key_columns(params.keys_size); - MutableColumns aggregate_columns(params.aggregates_size); - MutableColumns final_aggregate_columns(params.aggregates_size); - AggregateColumnsData aggregate_columns_data(params.aggregates_size); - - Block res_header = getHeader(final); - - for (size_t i = 0; i < params.keys_size; ++i) - { - key_columns[i] = res_header.safeGetByPosition(i).type->createColumn(); - key_columns[i]->reserve(rows); - } - - for (size_t i = 0; i < params.aggregates_size; ++i) - { - if (!final) - { - const auto & aggregate_column_name = params.aggregates[i].column_name; - aggregate_columns[i] = res_header.getByName(aggregate_column_name).type->createColumn(); - - /// The ColumnAggregateFunction column captures the shared ownership of the arena with the aggregate function states. - ColumnAggregateFunction & column_aggregate_func = assert_cast(*aggregate_columns[i]); - - for (auto & pool : data_variants.aggregates_pools) - column_aggregate_func.addArena(pool); - - aggregate_columns_data[i] = &column_aggregate_func.getData(); - aggregate_columns_data[i]->reserve(rows); - } - else - { - final_aggregate_columns[i] = aggregate_functions[i]->getReturnType()->createColumn(); - final_aggregate_columns[i]->reserve(rows); - - if (aggregate_functions[i]->isState()) + if (data.hasNullKeyData()) { - /// The ColumnAggregateFunction column captures the shared ownership of the arena with aggregate function states. - if (auto * column_aggregate_func = typeid_cast(final_aggregate_columns[i].get())) - for (auto & pool : data_variants.aggregates_pools) - column_aggregate_func->addArena(pool); - - /// Aggregate state can be wrapped into array if aggregate function ends with -Resample combinator. - final_aggregate_columns[i]->forEachSubcolumn([&data_variants](auto & subcolumn) - { - if (auto * column_aggregate_func = typeid_cast(subcolumn.get())) - for (auto & pool : data_variants.aggregates_pools) - column_aggregate_func->addArena(pool); - }); + out_cols->key_columns[0]->insertDefault(); + insertAggregatesIntoColumns(data.getNullKeyData(), out_cols->final_aggregate_columns, arena); + data.hasNullKeyData() = false; } } - } - filler(key_columns, aggregate_columns_data, final_aggregate_columns, final); + shuffled_key_sizes = method.shuffleKeyColumns(out_cols->raw_key_columns, key_sizes); - Block res = res_header.cloneEmpty(); + places.reserve(max_block_size); + }; - for (size_t i = 0; i < params.keys_size; ++i) - res.getByPosition(i).column = std::move(key_columns[i]); + // should be invoked at least once, because null data might be the only content of the `data` + init_out_cols(); - for (size_t i = 0; i < params.aggregates_size; ++i) + data.forEachValue( + [&](const auto & key, auto & mapped) + { + if (!out_cols.has_value()) + init_out_cols(); + + const auto & key_sizes_ref = shuffled_key_sizes ? *shuffled_key_sizes : key_sizes; + method.insertKeyIntoColumns(key, out_cols->raw_key_columns, key_sizes_ref); + places.emplace_back(mapped); + + /// Mark the cell as destroyed so it will not be destroyed in destructor. + mapped = nullptr; + + if constexpr (!return_single_block) + { + if (places.size() >= max_block_size) + { + res.emplace_back(insertResultsIntoColumns(places, std::move(out_cols.value()), arena)); + places.clear(); + out_cols.reset(); + } + } + }); + + if constexpr (return_single_block) { - const auto & aggregate_column_name = params.aggregates[i].column_name; - if (final) - res.getByName(aggregate_column_name).column = std::move(final_aggregate_columns[i]); - else - res.getByName(aggregate_column_name).column = std::move(aggregate_columns[i]); + return insertResultsIntoColumns(places, std::move(out_cols.value()), arena); } + else + { + if (out_cols.has_value()) + res.emplace_back(insertResultsIntoColumns(places, std::move(out_cols.value()), arena)); + return res; + } +} - /// Change the size of the columns-constants in the block. - size_t columns = res_header.columns(); - for (size_t i = 0; i < columns; ++i) - if (isColumnConst(*res.getByPosition(i).column)) - res.getByPosition(i).column = res.getByPosition(i).column->cut(0, rows); +template +Aggregator::ConvertToBlockRes NO_INLINE +Aggregator::convertToBlockImplNotFinal(Method & method, Table & data, Arenas & aggregates_pools, size_t) const +{ + const size_t max_block_size = params.max_block_size; + const bool final = false; + ConvertToBlockRes res; + std::optional out_cols; + std::optional shuffled_key_sizes; + + auto init_out_cols = [&]() + { + out_cols = prepareOutputBlockColumns(params, aggregate_functions, getHeader(final), aggregates_pools, final, max_block_size); + + if constexpr (Method::low_cardinality_optimization) + { + if (data.hasNullKeyData()) + { + out_cols->raw_key_columns[0]->insertDefault(); + + for (size_t i = 0; i < params.aggregates_size; ++i) + out_cols->aggregate_columns_data[i]->push_back(data.getNullKeyData() + offsets_of_aggregate_states[i]); + + data.getNullKeyData() = nullptr; + data.hasNullKeyData() = false; + } + } + + shuffled_key_sizes = method.shuffleKeyColumns(out_cols->raw_key_columns, key_sizes); + }; + + // should be invoked at least once, because null data might be the only content of the `data` + init_out_cols(); + + size_t rows_in_current_block = 0; + + data.forEachValue( + [&](const auto & key, auto & mapped) + { + if (!out_cols.has_value()) + init_out_cols(); + + const auto & key_sizes_ref = shuffled_key_sizes ? *shuffled_key_sizes : key_sizes; + method.insertKeyIntoColumns(key, out_cols->raw_key_columns, key_sizes_ref); + + /// reserved, so push_back does not throw exceptions + for (size_t i = 0; i < params.aggregates_size; ++i) + out_cols->aggregate_columns_data[i]->push_back(mapped + offsets_of_aggregate_states[i]); + + mapped = nullptr; + + ++rows_in_current_block; + + if constexpr (!return_single_block) + { + if (rows_in_current_block >= max_block_size) + { + res.emplace_back(finalizeBlock(params, getHeader(final), std::move(out_cols.value()), final, rows_in_current_block)); + out_cols.reset(); + rows_in_current_block = 0; + } + } + }); + + if constexpr (return_single_block) + { + return finalizeBlock(params, getHeader(final), std::move(out_cols).value(), final, rows_in_current_block); + } + else + { + if (rows_in_current_block) + res.emplace_back(finalizeBlock(params, getHeader(final), std::move(out_cols).value(), final, rows_in_current_block)); + return res; + } return res; } @@ -2105,39 +2097,35 @@ void Aggregator::createStatesAndFillKeyColumnsWithSingleKey( Block Aggregator::prepareBlockAndFillWithoutKey(AggregatedDataVariants & data_variants, bool final, bool is_overflows) const { size_t rows = 1; + auto && out_cols + = prepareOutputBlockColumns(params, aggregate_functions, getHeader(final), data_variants.aggregates_pools, final, rows); + auto && [key_columns, raw_key_columns, aggregate_columns, final_aggregate_columns, aggregate_columns_data] = out_cols; - auto filler = [&data_variants, this]( - MutableColumns & key_columns, - AggregateColumnsData & aggregate_columns, - MutableColumns & final_aggregate_columns, - bool final_) + if (data_variants.type == AggregatedDataVariants::Type::without_key || params.overflow_row) { - if (data_variants.type == AggregatedDataVariants::Type::without_key || params.overflow_row) + AggregatedDataWithoutKey & data = data_variants.without_key; + + if (!data) + throw Exception("Wrong data variant passed.", ErrorCodes::LOGICAL_ERROR); + + if (!final) { - AggregatedDataWithoutKey & data = data_variants.without_key; - - if (!data) - throw Exception("Wrong data variant passed.", ErrorCodes::LOGICAL_ERROR); - - if (!final_) - { - for (size_t i = 0; i < params.aggregates_size; ++i) - aggregate_columns[i]->push_back(data + offsets_of_aggregate_states[i]); - data = nullptr; - } - else - { - /// Always single-thread. It's safe to pass current arena from 'aggregates_pool'. - insertAggregatesIntoColumns(data, final_aggregate_columns, data_variants.aggregates_pool); - } - - if (params.overflow_row) - for (size_t i = 0; i < params.keys_size; ++i) - key_columns[i]->insertDefault(); + for (size_t i = 0; i < params.aggregates_size; ++i) + aggregate_columns_data[i]->push_back(data + offsets_of_aggregate_states[i]); + data = nullptr; + } + else + { + /// Always single-thread. It's safe to pass current arena from 'aggregates_pool'. + insertAggregatesIntoColumns(data, final_aggregate_columns, data_variants.aggregates_pool); } - }; - Block block = prepareBlockAndFill(data_variants, final, rows, filler); + if (params.overflow_row) + for (size_t i = 0; i < params.keys_size; ++i) + key_columns[i]->insertDefault(); + } + + Block block = finalizeBlock(params, getHeader(final), std::move(out_cols), final, rows); if (is_overflows) block.info.is_overflows = true; @@ -2148,29 +2136,22 @@ Block Aggregator::prepareBlockAndFillWithoutKey(AggregatedDataVariants & data_va return block; } -Block Aggregator::prepareBlockAndFillSingleLevel(AggregatedDataVariants & data_variants, bool final) const +template +Aggregator::ConvertToBlockRes +Aggregator::prepareBlockAndFillSingleLevel(AggregatedDataVariants & data_variants, bool final) const { - size_t rows = data_variants.sizeWithoutOverflowRow(); + const size_t rows = data_variants.sizeWithoutOverflowRow(); +#define M(NAME) \ + else if (data_variants.type == AggregatedDataVariants::Type::NAME) \ + { \ + return convertToBlockImpl( \ + *data_variants.NAME, data_variants.NAME->data, data_variants.aggregates_pool, data_variants.aggregates_pools, final, rows); \ + } - auto filler = [&data_variants, this]( - MutableColumns & key_columns, - AggregateColumnsData & aggregate_columns, - MutableColumns & final_aggregate_columns, - bool final_) - { - #define M(NAME) \ - else if (data_variants.type == AggregatedDataVariants::Type::NAME) \ - convertToBlockImpl(*data_variants.NAME, data_variants.NAME->data, \ - key_columns, aggregate_columns, final_aggregate_columns, data_variants.aggregates_pool, final_); - - if (false) {} // NOLINT - APPLY_FOR_VARIANTS_SINGLE_LEVEL(M) - #undef M - else - throw Exception("Unknown aggregated data variant.", ErrorCodes::UNKNOWN_AGGREGATED_DATA_VARIANT); - }; - - return prepareBlockAndFill(data_variants, final, rows, filler); + if (false) {} // NOLINT + APPLY_FOR_VARIANTS_SINGLE_LEVEL(M) +#undef M + else throw Exception("Unknown aggregated data variant.", ErrorCodes::UNKNOWN_AGGREGATED_DATA_VARIANT); } @@ -2292,7 +2273,7 @@ BlocksList Aggregator::convertToBlocks(AggregatedDataVariants & data_variants, b if (data_variants.type != AggregatedDataVariants::Type::without_key) { if (!data_variants.isTwoLevel()) - blocks.emplace_back(prepareBlockAndFillSingleLevel(data_variants, final)); + blocks.splice(blocks.end(), prepareBlockAndFillSingleLevel(data_variants, final)); else blocks.splice(blocks.end(), prepareBlocksAndFillTwoLevel(data_variants, final, thread_pool.get())); } @@ -3044,9 +3025,15 @@ Block Aggregator::mergeBlocks(BlocksList & blocks, bool final) Block block; if (result.type == AggregatedDataVariants::Type::without_key || is_overflows) + { block = prepareBlockAndFillWithoutKey(result, final, is_overflows); + } else - block = prepareBlockAndFillSingleLevel(result, final); + { + // Used during memory efficient merging (SortingAggregatedTransform expects single chunk for each bucket_id). + constexpr bool return_single_block = true; + block = prepareBlockAndFillSingleLevel(result, final); + } /// NOTE: two-level data is not possible here - chooseAggregationMethod chooses only among single-level methods. if (!final) @@ -3247,4 +3234,6 @@ void Aggregator::destroyAllAggregateStates(AggregatedDataVariants & result) cons } +template Aggregator::ConvertToBlockRes +Aggregator::prepareBlockAndFillSingleLevel(AggregatedDataVariants & data_variants, bool final) const; } diff --git a/src/Interpreters/Aggregator.h b/src/Interpreters/Aggregator.h index ce63c24969a..1d317e0a93a 100644 --- a/src/Interpreters/Aggregator.h +++ b/src/Interpreters/Aggregator.h @@ -1,8 +1,9 @@ #pragma once -#include -#include #include +#include +#include +#include #include @@ -872,6 +873,7 @@ using ManyAggregatedDataVariantsPtr = std::shared_ptr - void convertToBlockImpl( - Method & method, - Table & data, - MutableColumns & key_columns, - AggregateColumnsData & aggregate_columns, - MutableColumns & final_aggregate_columns, - Arena * arena, - bool final) const; + template + using ConvertToBlockRes = std::conditional_t; + + template + ConvertToBlockRes + convertToBlockImpl(Method & method, Table & data, Arena * arena, Arenas & aggregates_pools, bool final, size_t rows) const; template void insertAggregatesIntoColumns( @@ -1293,27 +1296,16 @@ private: MutableColumns & final_aggregate_columns, Arena * arena) const; - template - void convertToBlockImplFinal( - Method & method, - Table & data, - std::vector key_columns, - MutableColumns & final_aggregate_columns, - Arena * arena) const; + template + Block insertResultsIntoColumns(PaddedPODArray & places, OutputBlockColumns && out_cols, Arena * arena) const; - template - void convertToBlockImplNotFinal( - Method & method, - Table & data, - std::vector key_columns, - AggregateColumnsData & aggregate_columns) const; + template + ConvertToBlockRes + convertToBlockImplFinal(Method & method, Table & data, Arena * arena, Arenas & aggregates_pools, size_t rows) const; - template - Block prepareBlockAndFill( - AggregatedDataVariants & data_variants, - bool final, - size_t rows, - Filler && filler) const; + template + ConvertToBlockRes + convertToBlockImplNotFinal(Method & method, Table & data, Arenas & aggregates_pools, size_t rows) const; template Block convertOneBucketToBlock( @@ -1331,9 +1323,11 @@ private: std::atomic * is_cancelled = nullptr) const; Block prepareBlockAndFillWithoutKey(AggregatedDataVariants & data_variants, bool final, bool is_overflows) const; - Block prepareBlockAndFillSingleLevel(AggregatedDataVariants & data_variants, bool final) const; BlocksList prepareBlocksAndFillTwoLevel(AggregatedDataVariants & data_variants, bool final, ThreadPool * thread_pool) const; + template + ConvertToBlockRes prepareBlockAndFillSingleLevel(AggregatedDataVariants & data_variants, bool final) const; + template BlocksList prepareBlocksAndFillTwoLevelImpl( AggregatedDataVariants & data_variants, diff --git a/src/Interpreters/AsynchronousMetrics.cpp b/src/Interpreters/AsynchronousMetrics.cpp index f9bc22dd110..bccfa8f5b1e 100644 --- a/src/Interpreters/AsynchronousMetrics.cpp +++ b/src/Interpreters/AsynchronousMetrics.cpp @@ -11,10 +11,10 @@ #include #include #include -#include +#include #include #include -#include +#include #include #include #include @@ -77,9 +77,11 @@ static std::unique_ptr openFileIfExists(const std::stri AsynchronousMetrics::AsynchronousMetrics( ContextPtr global_context_, int update_period_seconds, + int heavy_metrics_update_period_seconds, const ProtocolServerMetricsFunc & protocol_server_metrics_func_) : WithContext(global_context_) , update_period(update_period_seconds) + , heavy_metric_update_period(heavy_metrics_update_period_seconds) , protocol_server_metrics_func(protocol_server_metrics_func_) , log(&Poco::Logger::get("AsynchronousMetrics")) { @@ -563,7 +565,7 @@ AsynchronousMetrics::NetworkInterfaceStatValues::operator-(const AsynchronousMet #endif -void AsynchronousMetrics::update(std::chrono::system_clock::time_point update_time) +void AsynchronousMetrics::update(TimePoint update_time) { Stopwatch watch; @@ -1584,6 +1586,8 @@ void AsynchronousMetrics::update(std::chrono::system_clock::time_point update_ti saveAllArenasMetric(new_values, "muzzy_purged"); #endif + updateHeavyMetricsIfNeeded(current_time, update_time, new_values); + /// Add more metrics as you wish. new_values["AsynchronousMetricsCalculationTimeSpent"] = watch.elapsedSeconds(); @@ -1601,4 +1605,76 @@ void AsynchronousMetrics::update(std::chrono::system_clock::time_point update_ti values = new_values; } +void AsynchronousMetrics::updateDetachedPartsStats() +{ + DetachedPartsStats current_values{}; + + for (const auto & db : DatabaseCatalog::instance().getDatabases()) + { + if (!db.second->canContainMergeTreeTables()) + continue; + + for (auto iterator = db.second->getTablesIterator(getContext()); iterator->isValid(); iterator->next()) + { + const auto & table = iterator->table(); + if (!table) + continue; + + if (MergeTreeData * table_merge_tree = dynamic_cast(table.get())) + { + for (const auto & detached_part: table_merge_tree->getDetachedParts()) + { + if (!detached_part.valid_name) + continue; + + if (detached_part.prefix.empty()) + ++current_values.detached_by_user; + + ++current_values.count; + } + } + } + } + + detached_parts_stats = current_values; +} + +void AsynchronousMetrics::updateHeavyMetricsIfNeeded(TimePoint current_time, TimePoint update_time, AsynchronousMetricValues & new_values) +{ + const auto time_after_previous_update = current_time - heavy_metric_previous_update_time; + const bool update_heavy_metric = time_after_previous_update >= heavy_metric_update_period || first_run; + + if (update_heavy_metric) + { + heavy_metric_previous_update_time = update_time; + + Stopwatch watch; + + /// Test shows that listing 100000 entries consuming around 0.15 sec. + updateDetachedPartsStats(); + + watch.stop(); + + /// Normally heavy metrics don't delay the rest of the metrics calculation + /// otherwise log the warning message + auto log_level = std::make_pair(DB::LogsLevel::trace, Poco::Message::PRIO_TRACE); + if (watch.elapsedSeconds() > (update_period.count() / 2.)) + log_level = std::make_pair(DB::LogsLevel::debug, Poco::Message::PRIO_DEBUG); + else if (watch.elapsedSeconds() > (update_period.count() / 4. * 3)) + log_level = std::make_pair(DB::LogsLevel::warning, Poco::Message::PRIO_WARNING); + LOG_IMPL(log, log_level.first, log_level.second, + "Update heavy metrics. " + "Update period {} sec. " + "Update heavy metrics period {} sec. " + "Heavy metrics calculation elapsed: {} sec.", + update_period.count(), + heavy_metric_update_period.count(), + watch.elapsedSeconds()); + + } + + new_values["NumberOfDetachedParts"] = detached_parts_stats.count; + new_values["NumberOfDetachedByUserParts"] = detached_parts_stats.detached_by_user; +} + } diff --git a/src/Interpreters/AsynchronousMetrics.h b/src/Interpreters/AsynchronousMetrics.h index e4bcb2890f3..6e32bdb43b8 100644 --- a/src/Interpreters/AsynchronousMetrics.h +++ b/src/Interpreters/AsynchronousMetrics.h @@ -50,6 +50,7 @@ public: AsynchronousMetrics( ContextPtr global_context_, int update_period_seconds, + int heavy_metrics_update_period_seconds, const ProtocolServerMetricsFunc & protocol_server_metrics_func_); ~AsynchronousMetrics(); @@ -63,7 +64,11 @@ public: AsynchronousMetricValues getValues() const; private: - const std::chrono::seconds update_period; + using Duration = std::chrono::seconds; + using TimePoint = std::chrono::system_clock::time_point; + + const Duration update_period; + const Duration heavy_metric_update_period; ProtocolServerMetricsFunc protocol_server_metrics_func; mutable std::mutex mutex; @@ -74,7 +79,16 @@ private: /// Some values are incremental and we have to calculate the difference. /// On first run we will only collect the values to subtract later. bool first_run = true; - std::chrono::system_clock::time_point previous_update_time; + TimePoint previous_update_time; + TimePoint heavy_metric_previous_update_time; + + struct DetachedPartsStats + { + size_t count; + size_t detached_by_user; + }; + + DetachedPartsStats detached_parts_stats{}; #if defined(OS_LINUX) || defined(OS_FREEBSD) MemoryStatisticsOS memory_stat; @@ -185,7 +199,10 @@ private: std::unique_ptr thread; void run(); - void update(std::chrono::system_clock::time_point update_time); + void update(TimePoint update_time); + + void updateDetachedPartsStats(); + void updateHeavyMetricsIfNeeded(TimePoint current_time, TimePoint update_time, AsynchronousMetricValues & new_values); Poco::Logger * log; }; diff --git a/src/Common/FileCache.cpp b/src/Interpreters/Cache/FileCache.cpp similarity index 97% rename from src/Common/FileCache.cpp rename to src/Interpreters/Cache/FileCache.cpp index 44ecac2cc02..f51df9ae737 100644 --- a/src/Common/FileCache.cpp +++ b/src/Interpreters/Cache/FileCache.cpp @@ -2,7 +2,8 @@ #include #include -#include +#include +#include #include #include #include @@ -10,7 +11,6 @@ #include #include #include -#include namespace fs = std::filesystem; @@ -59,6 +59,24 @@ String FileCache::getPathInLocalCache(const Key & key) const return fs::path(cache_base_path) / key_str.substr(0, 3) / key_str; } +void FileCache::removeKeyDirectoryIfExists(const Key & key, std::lock_guard & /* cache_lock */) const +{ + /// Note: it is guaranteed that there is no concurrency here with files deletion + /// because cache key directories are create only in FileCache class under cache_lock. + + auto key_str = key.toString(); + auto key_prefix_path = fs::path(cache_base_path) / key_str.substr(0, 3); + auto key_path = key_prefix_path / key_str; + + if (!fs::exists(key_path)) + return; + + fs::remove_all(key_path); + + if (fs::is_empty(key_prefix_path)) + fs::remove(key_prefix_path); +} + static bool isQueryInitialized() { return CurrentThread::isInitialized() @@ -104,6 +122,7 @@ void FileCache::initialize() fs::create_directories(cache_base_path); } + status_file = make_unique(fs::path(cache_base_path) / "status", StatusFile::write_full_info); is_initialized = true; } } @@ -174,15 +193,8 @@ FileSegments FileCache::getImpl( const auto & file_segments = it->second; if (file_segments.empty()) { - auto key_path = getPathInLocalCache(key); - files.erase(key); - - /// Note: it is guaranteed that there is no concurrency with files deletion, - /// because cache files are deleted only inside FileCache and under cache lock. - if (fs::exists(key_path)) - fs::remove_all(key_path); - + removeKeyDirectoryIfExists(key, cache_lock); return {}; } @@ -827,14 +839,10 @@ void FileCache::removeIfExists(const Key & key) } } - auto key_path = getPathInLocalCache(key); - if (!some_cells_were_skipped) { files.erase(key); - - if (fs::exists(key_path)) - fs::remove_all(key_path); + removeKeyDirectoryIfExists(key, cache_lock); } } @@ -924,12 +932,8 @@ void FileCache::remove( if (is_initialized && offsets.empty()) { - auto key_path = getPathInLocalCache(key); - files.erase(key); - - if (fs::exists(key_path)) - fs::remove_all(key_path); + removeKeyDirectoryIfExists(key, cache_lock); } } catch (...) @@ -960,12 +964,19 @@ void FileCache::loadCacheInfoIntoMemory(std::lock_guard & cache_lock fs::directory_iterator key_prefix_it{cache_base_path}; for (; key_prefix_it != fs::directory_iterator(); ++key_prefix_it) { + if (!key_prefix_it->is_directory()) + { + if (key_prefix_it->path().filename() != "status") + LOG_DEBUG(log, "Unexpected file {} (not a directory), will skip it", key_prefix_it->path().string()); + continue; + } + fs::directory_iterator key_it{key_prefix_it->path()}; for (; key_it != fs::directory_iterator(); ++key_it) { if (!key_it->is_directory()) { - LOG_WARNING(log, "Unexpected file: {}. Expected a directory", key_it->path().string()); + LOG_DEBUG(log, "Unexpected file {} (not a directory), will skip it", key_it->path().string()); continue; } diff --git a/src/Common/FileCache.h b/src/Interpreters/Cache/FileCache.h similarity index 96% rename from src/Common/FileCache.h rename to src/Interpreters/Cache/FileCache.h index b5b1e917e76..3f5a5c9e1c5 100644 --- a/src/Common/FileCache.h +++ b/src/Interpreters/Cache/FileCache.h @@ -13,11 +13,12 @@ #include #include -#include -#include -#include +#include +#include +#include #include -#include +#include +#include namespace DB { @@ -143,6 +144,7 @@ private: bool is_initialized = false; std::exception_ptr initialization_exception; + std::unique_ptr status_file; mutable std::mutex mutex; @@ -261,6 +263,8 @@ private: void assertCacheCellsCorrectness(const FileSegmentsByOffset & cells_by_offset, std::lock_guard & cache_lock); + void removeKeyDirectoryIfExists(const Key & key, std::lock_guard & cache_lock) const; + /// Used to track and control the cache access of each query. /// Through it, we can realize the processing of different queries by the cache layer. struct QueryContext diff --git a/src/Common/FileCacheFactory.cpp b/src/Interpreters/Cache/FileCacheFactory.cpp similarity index 100% rename from src/Common/FileCacheFactory.cpp rename to src/Interpreters/Cache/FileCacheFactory.cpp diff --git a/src/Common/FileCacheFactory.h b/src/Interpreters/Cache/FileCacheFactory.h similarity index 92% rename from src/Common/FileCacheFactory.h rename to src/Interpreters/Cache/FileCacheFactory.h index 0d99da0343c..82e0ec8f928 100644 --- a/src/Common/FileCacheFactory.h +++ b/src/Interpreters/Cache/FileCacheFactory.h @@ -1,7 +1,7 @@ #pragma once -#include -#include +#include +#include #include #include diff --git a/src/Common/FileCacheType.h b/src/Interpreters/Cache/FileCacheKey.h similarity index 100% rename from src/Common/FileCacheType.h rename to src/Interpreters/Cache/FileCacheKey.h diff --git a/src/Common/FileCacheSettings.cpp b/src/Interpreters/Cache/FileCacheSettings.cpp similarity index 100% rename from src/Common/FileCacheSettings.cpp rename to src/Interpreters/Cache/FileCacheSettings.cpp diff --git a/src/Common/FileCacheSettings.h b/src/Interpreters/Cache/FileCacheSettings.h similarity index 94% rename from src/Common/FileCacheSettings.h rename to src/Interpreters/Cache/FileCacheSettings.h index 1e4f668e8bb..c6155edad85 100644 --- a/src/Common/FileCacheSettings.h +++ b/src/Interpreters/Cache/FileCacheSettings.h @@ -1,6 +1,6 @@ #pragma once -#include +#include namespace Poco { namespace Util { class AbstractConfiguration; } } // NOLINT(cppcoreguidelines-virtual-class-destructor) diff --git a/src/Common/FileCache_fwd.h b/src/Interpreters/Cache/FileCache_fwd.h similarity index 100% rename from src/Common/FileCache_fwd.h rename to src/Interpreters/Cache/FileCache_fwd.h diff --git a/src/Common/FileSegment.cpp b/src/Interpreters/Cache/FileSegment.cpp similarity index 99% rename from src/Common/FileSegment.cpp rename to src/Interpreters/Cache/FileSegment.cpp index 1e1a309a8c7..547e6849dd6 100644 --- a/src/Common/FileSegment.cpp +++ b/src/Interpreters/Cache/FileSegment.cpp @@ -3,7 +3,7 @@ #include #include #include -#include +#include #include #include #include diff --git a/src/Common/FileSegment.h b/src/Interpreters/Cache/FileSegment.h similarity index 99% rename from src/Common/FileSegment.h rename to src/Interpreters/Cache/FileSegment.h index 8b826576e1f..f3fb367792a 100644 --- a/src/Common/FileSegment.h +++ b/src/Interpreters/Cache/FileSegment.h @@ -5,7 +5,8 @@ #include #include #include -#include +#include + namespace Poco { class Logger; } diff --git a/src/Common/IFileCachePriority.h b/src/Interpreters/Cache/IFileCachePriority.h similarity index 98% rename from src/Common/IFileCachePriority.h rename to src/Interpreters/Cache/IFileCachePriority.h index f80266f9eea..de5e3f1428c 100644 --- a/src/Common/IFileCachePriority.h +++ b/src/Interpreters/Cache/IFileCachePriority.h @@ -4,7 +4,7 @@ #include #include #include -#include +#include namespace DB { diff --git a/src/Common/LRUFileCachePriority.cpp b/src/Interpreters/Cache/LRUFileCachePriority.cpp similarity index 98% rename from src/Common/LRUFileCachePriority.cpp rename to src/Interpreters/Cache/LRUFileCachePriority.cpp index c4d6313e4d6..17fbd2c2092 100644 --- a/src/Common/LRUFileCachePriority.cpp +++ b/src/Interpreters/Cache/LRUFileCachePriority.cpp @@ -1,4 +1,4 @@ -#include +#include #include namespace CurrentMetrics diff --git a/src/Common/LRUFileCachePriority.h b/src/Interpreters/Cache/LRUFileCachePriority.h similarity index 97% rename from src/Common/LRUFileCachePriority.h rename to src/Interpreters/Cache/LRUFileCachePriority.h index 2cdcc981e04..2345d3c47db 100644 --- a/src/Common/LRUFileCachePriority.h +++ b/src/Interpreters/Cache/LRUFileCachePriority.h @@ -1,7 +1,7 @@ #pragma once #include -#include +#include #include namespace DB diff --git a/src/Interpreters/ClientInfo.h b/src/Interpreters/ClientInfo.h index 3ea846101f5..a1096b99325 100644 --- a/src/Interpreters/ClientInfo.h +++ b/src/Interpreters/ClientInfo.h @@ -62,9 +62,8 @@ public: time_t initial_query_start_time{}; Decimal64 initial_query_start_time_microseconds{}; - // OpenTelemetry trace context we received from client, or which we are going - // to send to server. - OpenTelemetryTraceContext client_trace_context; + /// OpenTelemetry trace context we received from client, or which we are going to send to server. + OpenTelemetry::TracingContext client_trace_context; /// All below are parameters related to initial query. diff --git a/src/Interpreters/Context.cpp b/src/Interpreters/Context.cpp index d45d30f1957..91604c8cc82 100644 --- a/src/Interpreters/Context.cpp +++ b/src/Interpreters/Context.cpp @@ -1335,29 +1335,6 @@ void Context::setCurrentQueryId(const String & query_id) random.words.a = thread_local_rng(); //-V656 random.words.b = thread_local_rng(); //-V656 - if (client_info.client_trace_context.trace_id != UUID()) - { - // Use the OpenTelemetry trace context we received from the client, and - // create a new span for the query. - query_trace_context = client_info.client_trace_context; - query_trace_context.span_id = thread_local_rng(); - } - else if (client_info.query_kind == ClientInfo::QueryKind::INITIAL_QUERY) - { - // If this is an initial query without any parent OpenTelemetry trace, we - // might start the trace ourselves, with some configurable probability. - std::bernoulli_distribution should_start_trace{ - settings.opentelemetry_start_trace_probability}; - - if (should_start_trace(thread_local_rng)) - { - // Use the randomly generated default query id as the new trace id. - query_trace_context.trace_id = random.uuid; - query_trace_context.span_id = thread_local_rng(); - // Mark this trace as sampled in the flags. - query_trace_context.trace_flags = 1; - } - } String query_id_to_set = query_id; if (query_id_to_set.empty()) /// If the user did not submit his query_id, then we generate it ourselves. diff --git a/src/Interpreters/Context.h b/src/Interpreters/Context.h index ea03b8e6586..2997fc370bf 100644 --- a/src/Interpreters/Context.h +++ b/src/Interpreters/Context.h @@ -364,8 +364,26 @@ private: bool apply_deleted_mask = true; public: - // Top-level OpenTelemetry trace context for the query. Makes sense only for a query context. - OpenTelemetryTraceContext query_trace_context; + /// Some counters for current query execution. + /// Most of them are workarounds and should be removed in the future. + struct KitchenSink + { + std::atomic analyze_counter = 0; + + KitchenSink() = default; + + KitchenSink(const KitchenSink & rhs) + : analyze_counter(rhs.analyze_counter.load()) + {} + + KitchenSink & operator=(const KitchenSink & rhs) + { + analyze_counter = rhs.analyze_counter.load(); + return *this; + } + }; + + KitchenSink kitchen_sink; private: using SampleBlockCache = std::unordered_map; diff --git a/src/Interpreters/ExpressionAnalyzer.cpp b/src/Interpreters/ExpressionAnalyzer.cpp index 105d46eed1f..be32125edf8 100644 --- a/src/Interpreters/ExpressionAnalyzer.cpp +++ b/src/Interpreters/ExpressionAnalyzer.cpp @@ -725,7 +725,7 @@ void ExpressionAnalyzer::makeWindowDescriptionFromAST(const Context & context_, with_alias->getColumnName(), 1 /* direction */, 1 /* nulls_direction */)); - auto actions_dag = std::make_shared(columns_after_join); + auto actions_dag = std::make_shared(aggregated_columns); getRootActions(column_ast, false, actions_dag); desc.partition_by_actions.push_back(std::move(actions_dag)); } @@ -746,7 +746,7 @@ void ExpressionAnalyzer::makeWindowDescriptionFromAST(const Context & context_, order_by_element.direction, order_by_element.nulls_direction)); - auto actions_dag = std::make_shared(columns_after_join); + auto actions_dag = std::make_shared(aggregated_columns); getRootActions(column_ast, false, actions_dag); desc.order_by_actions.push_back(std::move(actions_dag)); } diff --git a/src/Interpreters/InterpreterDescribeCacheQuery.cpp b/src/Interpreters/InterpreterDescribeCacheQuery.cpp index d5e4a81e98e..7ebb6ef64c6 100644 --- a/src/Interpreters/InterpreterDescribeCacheQuery.cpp +++ b/src/Interpreters/InterpreterDescribeCacheQuery.cpp @@ -5,8 +5,8 @@ #include #include #include -#include -#include +#include +#include #include #include diff --git a/src/Interpreters/InterpreterFactory.cpp b/src/Interpreters/InterpreterFactory.cpp index 00183086bf6..170f3c463b4 100644 --- a/src/Interpreters/InterpreterFactory.cpp +++ b/src/Interpreters/InterpreterFactory.cpp @@ -114,8 +114,6 @@ namespace ErrorCodes std::unique_ptr InterpreterFactory::get(ASTPtr & query, ContextMutablePtr context, const SelectQueryOptions & options) { - OpenTelemetrySpanHolder span("InterpreterFactory::get()"); - ProfileEvents::increment(ProfileEvents::Query); if (query->as()) diff --git a/src/Interpreters/InterpreterInsertQuery.cpp b/src/Interpreters/InterpreterInsertQuery.cpp index 7b6066575ae..1e1b22cb5e0 100644 --- a/src/Interpreters/InterpreterInsertQuery.cpp +++ b/src/Interpreters/InterpreterInsertQuery.cpp @@ -346,7 +346,7 @@ BlockIO InterpreterInsertQuery::execute() const auto & union_modes = select_query.list_of_modes; /// ASTSelectWithUnionQuery is not normalized now, so it may pass some queries which can be Trivial select queries - const auto mode_is_all = [](const auto & mode) { return mode == SelectUnionMode::ALL; }; + const auto mode_is_all = [](const auto & mode) { return mode == SelectUnionMode::UNION_ALL; }; is_trivial_insert_select = std::all_of(union_modes.begin(), union_modes.end(), std::move(mode_is_all)) diff --git a/src/Interpreters/InterpreterSelectIntersectExceptQuery.cpp b/src/Interpreters/InterpreterSelectIntersectExceptQuery.cpp index d6add3f77a9..a134f7bb913 100644 --- a/src/Interpreters/InterpreterSelectIntersectExceptQuery.cpp +++ b/src/Interpreters/InterpreterSelectIntersectExceptQuery.cpp @@ -4,11 +4,14 @@ #include #include #include +#include +#include #include #include +#include +#include #include #include -#include #include @@ -134,10 +137,29 @@ void InterpreterSelectIntersectExceptQuery::buildQueryPlan(QueryPlan & query_pla data_streams[i] = plans[i]->getCurrentDataStream(); } - auto max_threads = context->getSettingsRef().max_threads; + const Settings & settings = context->getSettingsRef(); + auto max_threads = settings.max_threads; auto step = std::make_unique(std::move(data_streams), final_operator, max_threads); query_plan.unitePlans(std::move(step), std::move(plans)); + const auto & query = query_ptr->as(); + if (query.final_operator == ASTSelectIntersectExceptQuery::Operator::INTERSECT_DISTINCT + || query.final_operator == ASTSelectIntersectExceptQuery::Operator::EXCEPT_DISTINCT) + { + /// Add distinct transform + SizeLimits limits(settings.max_rows_in_distinct, settings.max_bytes_in_distinct, settings.distinct_overflow_mode); + + auto distinct_step = std::make_unique( + query_plan.getCurrentDataStream(), + limits, + 0, + result_header.getNames(), + false, + settings.optimize_distinct_in_order); + + query_plan.addStep(std::move(distinct_step)); + } + addAdditionalPostFilter(query_plan); query_plan.addInterpreterContext(context); } diff --git a/src/Interpreters/InterpreterSelectQuery.cpp b/src/Interpreters/InterpreterSelectQuery.cpp index 76f199d8e41..c73db82a27b 100644 --- a/src/Interpreters/InterpreterSelectQuery.cpp +++ b/src/Interpreters/InterpreterSelectQuery.cpp @@ -39,6 +39,7 @@ #include #include #include +#include #include #include #include @@ -639,7 +640,18 @@ InterpreterSelectQuery::InterpreterSelectQuery( analyze(shouldMoveToPrewhere()); bool need_analyze_again = false; - if (analysis_result.prewhere_constant_filter_description.always_false || analysis_result.prewhere_constant_filter_description.always_true) + bool can_analyze_again = false; + if (context->hasQueryContext()) + { + /// Check number of calls of 'analyze' function. + /// If it is too big, we will not analyze the query again not to have exponential blowup. + std::atomic & current_query_analyze_count = context->getQueryContext()->kitchen_sink.analyze_counter; + ++current_query_analyze_count; + can_analyze_again = settings.max_analyze_depth == 0 || current_query_analyze_count < settings.max_analyze_depth; + } + + if (can_analyze_again && (analysis_result.prewhere_constant_filter_description.always_false || + analysis_result.prewhere_constant_filter_description.always_true)) { if (analysis_result.prewhere_constant_filter_description.always_true) query.setExpression(ASTSelectQuery::Expression::PREWHERE, {}); @@ -647,7 +659,9 @@ InterpreterSelectQuery::InterpreterSelectQuery( query.setExpression(ASTSelectQuery::Expression::PREWHERE, std::make_shared(0u)); need_analyze_again = true; } - if (analysis_result.where_constant_filter_description.always_false || analysis_result.where_constant_filter_description.always_true) + + if (can_analyze_again && (analysis_result.where_constant_filter_description.always_false || + analysis_result.where_constant_filter_description.always_true)) { if (analysis_result.where_constant_filter_description.always_true) query.setExpression(ASTSelectQuery::Expression::WHERE, {}); @@ -658,7 +672,8 @@ InterpreterSelectQuery::InterpreterSelectQuery( if (need_analyze_again) { - LOG_TRACE(log, "Running 'analyze' second time"); + size_t current_query_analyze_count = context->getQueryContext()->kitchen_sink.analyze_counter.load(); + LOG_TRACE(log, "Running 'analyze' second time (current analyze depth: {})", current_query_analyze_count); /// Reuse already built sets for multiple passes of analysis prepared_sets = query_analyzer->getPreparedSets(); @@ -1422,7 +1437,7 @@ void InterpreterSelectQuery::executeImpl(QueryPlan & query_plan, std::optional

context->getTemporaryVolume(), settings.min_free_disk_space_for_temporary_data, settings.optimize_sorting_by_input_stream_properties); - sorting_step->setStepDescription(fmt::format("Sort {} before JOIN", is_right ? "right" : "left")); + sorting_step->setStepDescription(fmt::format("Sort {} before JOIN", join_pos)); plan.addStep(std::move(sorting_step)); }; + auto crosswise_connection = CreateSetAndFilterOnTheFlyStep::createCrossConnection(); + auto add_create_set = [&settings, crosswise_connection](QueryPlan & plan, const Names & key_names, JoinTableSide join_pos) + { + auto creating_set_step = std::make_unique( + plan.getCurrentDataStream(), key_names, settings.max_rows_in_set_to_optimize_join, crosswise_connection, join_pos); + creating_set_step->setStepDescription(fmt::format("Create set and filter {} joined stream", join_pos)); + + auto * step_raw_ptr = creating_set_step.get(); + plan.addStep(std::move(creating_set_step)); + return step_raw_ptr; + }; + if (expressions.join->pipelineType() == JoinPipelineType::YShaped) { - const auto & join_clause = expressions.join->getTableJoin().getOnlyClause(); - add_sorting(query_plan, join_clause.key_names_left, false); - add_sorting(*joined_plan, join_clause.key_names_right, true); + const auto & table_join = expressions.join->getTableJoin(); + const auto & join_clause = table_join.getOnlyClause(); + + auto join_kind = table_join.kind(); + bool kind_allows_filtering = isInner(join_kind) || isLeft(join_kind) || isRight(join_kind); + if (settings.max_rows_in_set_to_optimize_join > 0 && kind_allows_filtering) + { + auto * left_set = add_create_set(query_plan, join_clause.key_names_left, JoinTableSide::Left); + auto * right_set = add_create_set(*joined_plan, join_clause.key_names_right, JoinTableSide::Right); + + if (isInnerOrLeft(join_kind)) + right_set->setFiltering(left_set->getSet()); + + if (isInnerOrRight(join_kind)) + left_set->setFiltering(right_set->getSet()); + } + + add_sorting(query_plan, join_clause.key_names_left, JoinTableSide::Left); + add_sorting(*joined_plan, join_clause.key_names_right, JoinTableSide::Right); } QueryPlanStepPtr join_step = std::make_unique( @@ -1720,7 +1763,7 @@ static void executeMergeAggregatedImpl( * but it can work more slowly. */ - Aggregator::Params params(keys, aggregates, overflow_row, settings.max_threads); + Aggregator::Params params(keys, aggregates, overflow_row, settings.max_threads, settings.max_block_size); auto merging_aggregated = std::make_unique( query_plan.getCurrentDataStream(), @@ -2316,6 +2359,7 @@ static Aggregator::Params getAggregatorParams( settings.min_free_disk_space_for_temporary_data, settings.compile_aggregate_expressions, settings.min_count_to_compile_aggregate_expression, + settings.max_block_size, /* only_merge */ false, stats_collecting_params }; @@ -2592,7 +2636,7 @@ void InterpreterSelectQuery::executeOrderOptimized(QueryPlan & query_plan, Input auto finish_sorting_step = std::make_unique( query_plan.getCurrentDataStream(), - input_sorting_info->order_key_prefix_descr, + input_sorting_info->sort_description_for_merging, output_order_descr, settings.max_block_size, limit); diff --git a/src/Interpreters/InterpreterSelectWithUnionQuery.cpp b/src/Interpreters/InterpreterSelectWithUnionQuery.cpp index 87a182e70ae..a679b17a5bd 100644 --- a/src/Interpreters/InterpreterSelectWithUnionQuery.cpp +++ b/src/Interpreters/InterpreterSelectWithUnionQuery.cpp @@ -317,13 +317,13 @@ void InterpreterSelectWithUnionQuery::buildQueryPlan(QueryPlan & query_plan) data_streams[i] = plans[i]->getCurrentDataStream(); } - auto max_threads = context->getSettingsRef().max_threads; + auto max_threads = settings.max_threads; auto union_step = std::make_unique(std::move(data_streams), max_threads); query_plan.unitePlans(std::move(union_step), std::move(plans)); const auto & query = query_ptr->as(); - if (query.union_mode == SelectUnionMode::DISTINCT) + if (query.union_mode == SelectUnionMode::UNION_DISTINCT) { /// Add distinct transform SizeLimits limits(settings.max_rows_in_distinct, settings.max_bytes_in_distinct, settings.distinct_overflow_mode); diff --git a/src/Interpreters/InterpreterShowTablesQuery.cpp b/src/Interpreters/InterpreterShowTablesQuery.cpp index d623b6c71d6..35e1a27f24b 100644 --- a/src/Interpreters/InterpreterShowTablesQuery.cpp +++ b/src/Interpreters/InterpreterShowTablesQuery.cpp @@ -7,7 +7,7 @@ #include #include #include -#include +#include #include #include #include diff --git a/src/Interpreters/InterpreterSystemQuery.cpp b/src/Interpreters/InterpreterSystemQuery.cpp index 43bb15b70bb..106b1611f80 100644 --- a/src/Interpreters/InterpreterSystemQuery.cpp +++ b/src/Interpreters/InterpreterSystemQuery.cpp @@ -7,8 +7,8 @@ #include #include #include -#include -#include +#include +#include #include #include #include diff --git a/src/Interpreters/NormalizeSelectWithUnionQueryVisitor.cpp b/src/Interpreters/NormalizeSelectWithUnionQueryVisitor.cpp index 364828e6126..40c42f7728e 100644 --- a/src/Interpreters/NormalizeSelectWithUnionQueryVisitor.cpp +++ b/src/Interpreters/NormalizeSelectWithUnionQueryVisitor.cpp @@ -59,22 +59,22 @@ void NormalizeSelectWithUnionQueryMatcher::visit(ASTSelectWithUnionQuery & ast, continue; /// Rewrite UNION Mode - if (union_modes[i] == SelectUnionMode::Unspecified) + if (union_modes[i] == SelectUnionMode::UNION_DEFAULT) { - if (data.union_default_mode == UnionMode::ALL) - union_modes[i] = SelectUnionMode::ALL; - else if (data.union_default_mode == UnionMode::DISTINCT) - union_modes[i] = SelectUnionMode::DISTINCT; + if (data.union_default_mode == SetOperationMode::ALL) + union_modes[i] = SelectUnionMode::UNION_ALL; + else if (data.union_default_mode == SetOperationMode::DISTINCT) + union_modes[i] = SelectUnionMode::UNION_DISTINCT; else throw Exception( "Expected ALL or DISTINCT in SelectWithUnion query, because setting (union_default_mode) is empty", DB::ErrorCodes::EXPECTED_ALL_OR_DISTINCT); } - if (union_modes[i] == SelectUnionMode::ALL) + if (union_modes[i] == SelectUnionMode::UNION_ALL) { if (auto * inner_union = select_list[i + 1]->as(); - inner_union && inner_union->union_mode == SelectUnionMode::ALL) + inner_union && inner_union->union_mode == SelectUnionMode::UNION_ALL) { /// Inner_union is an UNION ALL list, just lift up for (auto child = inner_union->list_of_selects->children.rbegin(); child != inner_union->list_of_selects->children.rend(); @@ -85,7 +85,7 @@ void NormalizeSelectWithUnionQueryMatcher::visit(ASTSelectWithUnionQuery & ast, selects.push_back(select_list[i + 1]); } /// flatten all left nodes and current node to a UNION DISTINCT list - else if (union_modes[i] == SelectUnionMode::DISTINCT) + else if (union_modes[i] == SelectUnionMode::UNION_DISTINCT) { auto distinct_list = std::make_shared(); distinct_list->list_of_selects = std::make_shared(); @@ -96,7 +96,7 @@ void NormalizeSelectWithUnionQueryMatcher::visit(ASTSelectWithUnionQuery & ast, getSelectsFromUnionListNode(select_list[j], distinct_list->list_of_selects->children); } - distinct_list->union_mode = SelectUnionMode::DISTINCT; + distinct_list->union_mode = SelectUnionMode::UNION_DISTINCT; distinct_list->is_normalized = true; selects.push_back(std::move(distinct_list)); distinct_found = true; @@ -113,7 +113,7 @@ void NormalizeSelectWithUnionQueryMatcher::visit(ASTSelectWithUnionQuery & ast, if (!distinct_found) { if (auto * inner_union = select_list[0]->as(); - inner_union && inner_union->union_mode == SelectUnionMode::ALL) + inner_union && inner_union->union_mode == SelectUnionMode::UNION_ALL) { /// Inner_union is an UNION ALL list, just lift it up for (auto child = inner_union->list_of_selects->children.rbegin(); child != inner_union->list_of_selects->children.rend(); @@ -136,7 +136,7 @@ void NormalizeSelectWithUnionQueryMatcher::visit(ASTSelectWithUnionQuery & ast, std::reverse(selects.begin(), selects.end()); ast.is_normalized = true; - ast.union_mode = SelectUnionMode::ALL; + ast.union_mode = SelectUnionMode::UNION_ALL; ast.set_of_modes = std::move(current_set_of_modes); ast.list_of_selects->children = std::move(selects); diff --git a/src/Interpreters/NormalizeSelectWithUnionQueryVisitor.h b/src/Interpreters/NormalizeSelectWithUnionQueryVisitor.h index d035e90f5a5..e8194f0dfe1 100644 --- a/src/Interpreters/NormalizeSelectWithUnionQueryVisitor.h +++ b/src/Interpreters/NormalizeSelectWithUnionQueryVisitor.h @@ -18,7 +18,7 @@ class NormalizeSelectWithUnionQueryMatcher public: struct Data { - const UnionMode & union_default_mode; + const SetOperationMode union_default_mode; }; static void getSelectsFromUnionListNode(ASTPtr ast_select, ASTs & selects); diff --git a/src/Interpreters/OpenTelemetrySpanLog.cpp b/src/Interpreters/OpenTelemetrySpanLog.cpp index 2683a5f7955..57d5c11ad97 100644 --- a/src/Interpreters/OpenTelemetrySpanLog.cpp +++ b/src/Interpreters/OpenTelemetrySpanLog.cpp @@ -68,195 +68,5 @@ void OpenTelemetrySpanLogElement::appendToBlock(MutableColumns & columns) const columns[i++]->insert(attributes); } - -OpenTelemetrySpanHolder::OpenTelemetrySpanHolder(const std::string & _operation_name) -{ - trace_id = 0; - - if (!CurrentThread::isInitialized()) - { - // There may be no thread context if we're running inside the - // clickhouse-client, e.g. reading an external table provided with the - // `--external` option. - return; - } - - auto & thread = CurrentThread::get(); - - trace_id = thread.thread_trace_context.trace_id; - if (trace_id == UUID()) - return; - - parent_span_id = thread.thread_trace_context.span_id; - span_id = thread_local_rng(); - operation_name = _operation_name; - start_time_us = std::chrono::duration_cast( - std::chrono::system_clock::now().time_since_epoch()).count(); - - thread.thread_trace_context.span_id = span_id; } - -OpenTelemetrySpanHolder::~OpenTelemetrySpanHolder() -{ - try - { - if (trace_id == UUID()) - return; - - // First of all, return old value of current span. - auto & thread = CurrentThread::get(); - assert(thread.thread_trace_context.span_id == span_id); - thread.thread_trace_context.span_id = parent_span_id; - - // Not sure what's the best way to access the log from here. - auto * thread_group = CurrentThread::getGroup().get(); - // Not sure whether and when this can be null. - if (!thread_group) - return; - - ContextPtr context; - { - std::lock_guard lock(thread_group->mutex); - context = thread_group->query_context.lock(); - } - - if (!context) - { - // Both global and query contexts can be null when executing a - // background task, and global context can be null for some - // queries. - return; - } - - auto log = context->getOpenTelemetrySpanLog(); - if (!log) - { - // The log might be disabled. - return; - } - - finish_time_us = std::chrono::duration_cast( - std::chrono::system_clock::now().time_since_epoch()).count(); - - log->add(OpenTelemetrySpanLogElement( - static_cast(*this))); - } - catch (...) - { - tryLogCurrentException(__FUNCTION__); - } -} - -void OpenTelemetrySpanHolder::addAttribute(const std::string& name, UInt64 value) -{ - if (trace_id == UUID()) - return; - - this->attributes.push_back(Tuple{name, toString(value)}); -} - -void OpenTelemetrySpanHolder::addAttribute(const std::string& name, const std::string& value) -{ - if (trace_id == UUID()) - return; - - this->attributes.push_back(Tuple{name, value}); -} - -void OpenTelemetrySpanHolder::addAttribute(const Exception & e) -{ - if (trace_id == UUID()) - return; - - this->attributes.push_back(Tuple{"clickhouse.exception", getExceptionMessage(e, false)}); -} - -void OpenTelemetrySpanHolder::addAttribute(std::exception_ptr e) -{ - if (trace_id == UUID() || e == nullptr) - return; - - this->attributes.push_back(Tuple{"clickhouse.exception", getExceptionMessage(e, false)}); -} - -bool OpenTelemetryTraceContext::parseTraceparentHeader(const std::string & traceparent, - std::string & error) -{ - trace_id = 0; - - // Version 00, which is the only one we can parse, is fixed width. Use this - // fact for an additional sanity check. - const int expected_length = strlen("xx-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx-xxxxxxxxxxxxxxxx-xx"); - if (traceparent.length() != expected_length) - { - error = fmt::format("unexpected length {}, expected {}", - traceparent.length(), expected_length); - return false; - } - - const char * data = traceparent.data(); - - uint8_t version = unhex2(data); - data += 2; - - if (version != 0) - { - error = fmt::format("unexpected version {}, expected 00", version); - return false; - } - - if (*data != '-') - { - error = fmt::format("Malformed traceparant header: {}", traceparent); - return false; - } - - ++data; - UInt64 trace_id_higher_64 = unhexUInt(data); - UInt64 trace_id_lower_64 = unhexUInt(data + 16); - data += 32; - - if (*data != '-') - { - error = fmt::format("Malformed traceparant header: {}", traceparent); - return false; - } - - ++data; - UInt64 span_id_64 = unhexUInt(data); - data += 16; - - if (*data != '-') - { - error = fmt::format("Malformed traceparant header: {}", traceparent); - return false; - } - - ++data; - this->trace_flags = unhex2(data); - - // store the 128-bit trace id in big-endian order - this->trace_id.toUnderType().items[0] = trace_id_higher_64; - this->trace_id.toUnderType().items[1] = trace_id_lower_64; - this->span_id = span_id_64; - return true; -} - - -std::string OpenTelemetryTraceContext::composeTraceparentHeader() const -{ - // This span is a parent for its children, so we specify this span_id as a - // parent id. - return fmt::format("00-{:016x}{:016x}-{:016x}-{:02x}", - // Output the trace id in network byte order - trace_id.toUnderType().items[0], - trace_id.toUnderType().items[1], - span_id, - // This cast is needed because fmt is being weird and complaining that - // "mixing character types is not allowed". - static_cast(trace_flags)); -} - - -} diff --git a/src/Interpreters/OpenTelemetrySpanLog.h b/src/Interpreters/OpenTelemetrySpanLog.h index 34f4765c8c4..e5a5b082284 100644 --- a/src/Interpreters/OpenTelemetrySpanLog.h +++ b/src/Interpreters/OpenTelemetrySpanLog.h @@ -7,24 +7,11 @@ namespace DB { -struct OpenTelemetrySpan -{ - UUID trace_id; - UInt64 span_id; - UInt64 parent_span_id; - std::string operation_name; - UInt64 start_time_us; - UInt64 finish_time_us; - Map attributes; - // I don't understand how Links work, namely, which direction should they - // point to, and how they are related with parent_span_id, so no Links for now. -}; - -struct OpenTelemetrySpanLogElement : public OpenTelemetrySpan +struct OpenTelemetrySpanLogElement : public OpenTelemetry::Span { OpenTelemetrySpanLogElement() = default; - explicit OpenTelemetrySpanLogElement(const OpenTelemetrySpan & span) - : OpenTelemetrySpan(span) {} + OpenTelemetrySpanLogElement(const OpenTelemetry::Span & span) + : OpenTelemetry::Span(span) {} static std::string name() { return "OpenTelemetrySpanLog"; } static NamesAndTypesList getNamesAndTypes(); @@ -41,15 +28,4 @@ public: using SystemLog::SystemLog; }; -struct OpenTelemetrySpanHolder : public OpenTelemetrySpan -{ - explicit OpenTelemetrySpanHolder(const std::string & _operation_name); - void addAttribute(const std::string& name, UInt64 value); - void addAttribute(const std::string& name, const std::string& value); - void addAttribute(const Exception & e); - void addAttribute(std::exception_ptr e); - - ~OpenTelemetrySpanHolder(); -}; - } diff --git a/src/Interpreters/ReplaceQueryParameterVisitor.cpp b/src/Interpreters/ReplaceQueryParameterVisitor.cpp index 03de8aecc92..664cda74522 100644 --- a/src/Interpreters/ReplaceQueryParameterVisitor.cpp +++ b/src/Interpreters/ReplaceQueryParameterVisitor.cpp @@ -1,16 +1,17 @@ -#include -#include #include -#include #include +#include #include #include -#include -#include -#include #include #include #include +#include +#include +#include +#include +#include +#include namespace DB @@ -30,7 +31,12 @@ void ReplaceQueryParameterVisitor::visit(ASTPtr & ast) else if (ast->as() || ast->as()) visitIdentifier(ast); else - visitChildren(ast); + { + if (auto * describe_query = dynamic_cast(ast.get()); describe_query && describe_query->table_expression) + visitChildren(describe_query->table_expression); + else + visitChildren(ast); + } } diff --git a/src/Interpreters/RewriteCountDistinctVisitor.cpp b/src/Interpreters/RewriteCountDistinctVisitor.cpp index a7a26a63460..cf28d8abb87 100644 --- a/src/Interpreters/RewriteCountDistinctVisitor.cpp +++ b/src/Interpreters/RewriteCountDistinctVisitor.cpp @@ -62,7 +62,7 @@ void RewriteCountDistinctFunctionMatcher::visit(ASTPtr & ast, Data & /*data*/) auto expr = std::make_shared(); expr->children.emplace_back(cloned_select_query); auto select_with_union = std::make_shared(); - select_with_union->union_mode = SelectUnionMode::Unspecified; + select_with_union->union_mode = SelectUnionMode::UNION_DEFAULT; select_with_union->is_normalized = false; select_with_union->list_of_modes.clear(); select_with_union->set_of_modes.clear(); diff --git a/src/Interpreters/SelectIntersectExceptQueryVisitor.cpp b/src/Interpreters/SelectIntersectExceptQueryVisitor.cpp index 4d695263f26..756a8a48e25 100644 --- a/src/Interpreters/SelectIntersectExceptQueryVisitor.cpp +++ b/src/Interpreters/SelectIntersectExceptQueryVisitor.cpp @@ -2,6 +2,8 @@ #include #include #include +#include +#include namespace DB @@ -9,6 +11,7 @@ namespace DB namespace ErrorCodes { extern const int LOGICAL_ERROR; + extern const int EXPECTED_ALL_OR_DISTINCT; } /* @@ -27,9 +30,9 @@ void SelectIntersectExceptQueryMatcher::visit(ASTPtr & ast, Data & data) visit(*select_union, data); } -void SelectIntersectExceptQueryMatcher::visit(ASTSelectWithUnionQuery & ast, Data &) +void SelectIntersectExceptQueryMatcher::visit(ASTSelectWithUnionQuery & ast, Data & data) { - const auto & union_modes = ast.list_of_modes; + auto union_modes = std::move(ast.list_of_modes); if (union_modes.empty()) return; @@ -46,14 +49,39 @@ void SelectIntersectExceptQueryMatcher::visit(ASTSelectWithUnionQuery & ast, Dat selects.pop_back(); SelectUnionModes modes; - for (const auto & mode : union_modes) + for (auto & mode : union_modes) { + /// Rewrite intersect / except mode + if (mode == SelectUnionMode::EXCEPT_DEFAULT) + { + if (data.except_default_mode == SetOperationMode::ALL) + mode = SelectUnionMode::EXCEPT_ALL; + else if (data.except_default_mode == SetOperationMode::DISTINCT) + mode = SelectUnionMode::EXCEPT_DISTINCT; + else + throw Exception( + "Expected ALL or DISTINCT in EXCEPT query, because setting (except_default_mode) is empty", + DB::ErrorCodes::EXPECTED_ALL_OR_DISTINCT); + } + else if (mode == SelectUnionMode::INTERSECT_DEFAULT) + { + if (data.intersect_default_mode == SetOperationMode::ALL) + mode = SelectUnionMode::INTERSECT_ALL; + else if (data.intersect_default_mode == SetOperationMode::DISTINCT) + mode = SelectUnionMode::INTERSECT_DISTINCT; + else + throw Exception( + "Expected ALL or DISTINCT in INTERSECT query, because setting (intersect_default_mode) is empty", + DB::ErrorCodes::EXPECTED_ALL_OR_DISTINCT); + } + switch (mode) { - case SelectUnionMode::EXCEPT: + case SelectUnionMode::EXCEPT_ALL: + case SelectUnionMode::EXCEPT_DISTINCT: { auto left = std::make_shared(); - left->union_mode = SelectUnionMode::ALL; + left->union_mode = mode == SelectUnionMode::EXCEPT_ALL ? SelectUnionMode::UNION_ALL : SelectUnionMode::UNION_DISTINCT; left->list_of_selects = std::make_shared(); left->children.push_back(left->list_of_selects); @@ -66,17 +94,22 @@ void SelectIntersectExceptQueryMatcher::visit(ASTSelectWithUnionQuery & ast, Dat selects.pop_back(); auto except_node = std::make_shared(); - except_node->final_operator = ASTSelectIntersectExceptQuery::Operator::EXCEPT; + except_node->final_operator = mode == SelectUnionMode::EXCEPT_ALL + ? ASTSelectIntersectExceptQuery::Operator::EXCEPT_ALL + : ASTSelectIntersectExceptQuery::Operator::EXCEPT_DISTINCT; except_node->children = {left, right}; children = {except_node}; break; } - case SelectUnionMode::INTERSECT: + case SelectUnionMode::INTERSECT_ALL: + case SelectUnionMode::INTERSECT_DISTINCT: { bool from_except = false; const auto * except_ast = typeid_cast(children.back().get()); - if (except_ast && (except_ast->final_operator == ASTSelectIntersectExceptQuery::Operator::EXCEPT)) + if (except_ast + && (except_ast->final_operator == ASTSelectIntersectExceptQuery::Operator::EXCEPT_ALL + || except_ast->final_operator == ASTSelectIntersectExceptQuery::Operator::EXCEPT_DISTINCT)) from_except = true; ASTPtr left; @@ -94,7 +127,9 @@ void SelectIntersectExceptQueryMatcher::visit(ASTSelectWithUnionQuery & ast, Dat selects.pop_back(); auto intersect_node = std::make_shared(); - intersect_node->final_operator = ASTSelectIntersectExceptQuery::Operator::INTERSECT; + intersect_node->final_operator = mode == SelectUnionMode::INTERSECT_ALL + ? ASTSelectIntersectExceptQuery::Operator::INTERSECT_ALL + : ASTSelectIntersectExceptQuery::Operator::INTERSECT_DISTINCT; intersect_node->children = {left, right}; if (from_except) @@ -122,7 +157,6 @@ void SelectIntersectExceptQueryMatcher::visit(ASTSelectWithUnionQuery & ast, Dat children.emplace_back(std::move(right)); } - ast.union_mode = SelectUnionMode::Unspecified; ast.list_of_selects->children = std::move(children); ast.list_of_modes = std::move(modes); } diff --git a/src/Interpreters/SelectIntersectExceptQueryVisitor.h b/src/Interpreters/SelectIntersectExceptQueryVisitor.h index 5aeb15f70bb..daf6d2ea0df 100644 --- a/src/Interpreters/SelectIntersectExceptQueryVisitor.h +++ b/src/Interpreters/SelectIntersectExceptQueryVisitor.h @@ -6,6 +6,7 @@ #include #include +#include namespace DB @@ -17,7 +18,11 @@ class ASTSelectWithUnionQuery; class SelectIntersectExceptQueryMatcher { public: - struct Data {}; + struct Data + { + const SetOperationMode intersect_default_mode; + const SetOperationMode except_default_mode; + }; static bool needChildVisit(const ASTPtr &, const ASTPtr &) { return true; } diff --git a/src/Interpreters/Session.h b/src/Interpreters/Session.h index 8de76349b7d..ed4f7809dee 100644 --- a/src/Interpreters/Session.h +++ b/src/Interpreters/Session.h @@ -65,6 +65,8 @@ public: ContextMutablePtr sessionContext() { return session_context; } ContextPtr sessionContext() const { return session_context; } + ContextPtr sessionOrGlobalContext() const { return session_context ? session_context : global_context; } + /// Makes a query context, can be used multiple times, with or without makeSession() called earlier. /// The query context will be created from a copy of a session context if it exists, or from a copy of /// a global context otherwise. In the latter case the function also assigns an user to this context. diff --git a/src/Interpreters/Set.cpp b/src/Interpreters/Set.cpp index af05b33c1f6..ded8b04a589 100644 --- a/src/Interpreters/Set.cpp +++ b/src/Interpreters/Set.cpp @@ -22,6 +22,8 @@ #include #include +#include + #include #include @@ -162,8 +164,16 @@ void Set::setHeader(const ColumnsWithTypeAndName & header) data.init(data.chooseMethod(key_columns, key_sizes)); } - bool Set::insertFromBlock(const ColumnsWithTypeAndName & columns) +{ + Columns cols; + cols.reserve(columns.size()); + for (const auto & column : columns) + cols.emplace_back(column.column); + return insertFromBlock(cols); +} + +bool Set::insertFromBlock(const Columns & columns) { std::lock_guard lock(rwlock); @@ -179,11 +189,11 @@ bool Set::insertFromBlock(const ColumnsWithTypeAndName & columns) /// Remember the columns we will work with for (size_t i = 0; i < keys_size; ++i) { - materialized_columns.emplace_back(columns.at(i).column->convertToFullIfNeeded()); + materialized_columns.emplace_back(columns.at(i)->convertToFullIfNeeded()); key_columns.emplace_back(materialized_columns.back().get()); } - size_t rows = columns.at(0).column->size(); + size_t rows = columns.at(0)->size(); /// We will insert to the Set only keys, where all components are not NULL. ConstNullMapPtr null_map{}; @@ -393,7 +403,13 @@ void Set::checkColumnsNumber(size_t num_key_columns) const bool Set::areTypesEqual(size_t set_type_idx, const DataTypePtr & other_type) const { - return removeNullable(recursiveRemoveLowCardinality(data_types[set_type_idx]))->equals(*removeNullable(recursiveRemoveLowCardinality(other_type))); + /// Out-of-bound access can happen when same set expression built with different columns. + /// Caller may call this method to make sure that the set is indeed the one they want + /// without awaring data_types.size(). + if (set_type_idx >= data_types.size()) + return false; + return removeNullable(recursiveRemoveLowCardinality(data_types[set_type_idx])) + ->equals(*removeNullable(recursiveRemoveLowCardinality(other_type))); } void Set::checkTypesEqual(size_t set_type_idx, const DataTypePtr & other_type) const diff --git a/src/Interpreters/Set.h b/src/Interpreters/Set.h index 6a3b28407ee..44f543ce222 100644 --- a/src/Interpreters/Set.h +++ b/src/Interpreters/Set.h @@ -20,6 +20,7 @@ class Context; class IFunctionBase; using FunctionBasePtr = std::shared_ptr; +class Chunk; /** Data structure for implementation of IN expression. */ @@ -45,11 +46,14 @@ public: void setHeader(const ColumnsWithTypeAndName & header); /// Returns false, if some limit was exceeded and no need to insert more data. + bool insertFromBlock(const Columns & columns); bool insertFromBlock(const ColumnsWithTypeAndName & columns); + /// Call after all blocks were inserted. To get the information that set is already created. void finishInsert() { is_created = true; } - bool isCreated() const { return is_created; } + /// finishInsert and isCreated are thread-safe + bool isCreated() const { return is_created.load(); } /** For columns of 'block', check belonging of corresponding rows to the set. * Return UInt8 column with the result. @@ -111,7 +115,7 @@ private: bool transform_null_in; /// Check if set contains all the data. - bool is_created = false; + std::atomic is_created = false; /// If in the left part columns contains the same types as the elements of the set. void executeOrdinary( diff --git a/src/Interpreters/TableJoin.h b/src/Interpreters/TableJoin.h index d0bf64fdebe..3835ef77deb 100644 --- a/src/Interpreters/TableJoin.h +++ b/src/Interpreters/TableJoin.h @@ -73,16 +73,32 @@ public: return key_names_right.size(); } - String formatDebug() const + String formatDebug(bool short_format = false) const { - return fmt::format("Left keys: [{}] Right keys [{}] Condition columns: '{}', '{}'", - fmt::join(key_names_left, ", "), fmt::join(key_names_right, ", "), - condColumnNames().first, condColumnNames().second); + const auto & [left_cond, right_cond] = condColumnNames(); + + if (short_format) + { + return fmt::format("({}) = ({}){}{}", fmt::join(key_names_left, ", "), fmt::join(key_names_right, ", "), + !left_cond.empty() ? " AND " + left_cond : "", !right_cond.empty() ? " AND " + right_cond : ""); + } + + return fmt::format( + "Left keys: [{}] Right keys [{}] Condition columns: '{}', '{}'", + fmt::join(key_names_left, ", "), fmt::join(key_names_right, ", "), left_cond, right_cond); } }; using Clauses = std::vector; + static std::string formatClauses(const Clauses & clauses, bool short_format = false) + { + std::vector res; + for (const auto & clause : clauses) + res.push_back("[" + clause.formatDebug(short_format) + "]"); + return fmt::format("{}", fmt::join(res, "; ")); + } + private: /** Query of the form `SELECT expr(x) AS k FROM t1 ANY LEFT JOIN (SELECT expr(x) AS k FROM t2) USING k` * The join is made by column k. diff --git a/src/Interpreters/ThreadStatusExt.cpp b/src/Interpreters/ThreadStatusExt.cpp index 643b2fdf497..b1f5749da25 100644 --- a/src/Interpreters/ThreadStatusExt.cpp +++ b/src/Interpreters/ThreadStatusExt.cpp @@ -84,15 +84,6 @@ void ThreadStatus::attachQueryContext(ContextPtr query_context_) thread_group->global_context = global_context; } - // Generate new span for thread manually here, because we can't depend - // on OpenTelemetrySpanHolder due to link order issues. - // FIXME why and how is this different from setupState()? - thread_trace_context = query_context_->query_trace_context; - if (thread_trace_context.trace_id != UUID()) - { - thread_trace_context.span_id = thread_local_rng(); - } - applyQuerySettings(); } @@ -132,18 +123,6 @@ void ThreadStatus::setupState(const ThreadGroupStatusPtr & thread_group_) if (auto query_context_ptr = query_context.lock()) { applyQuerySettings(); - - // Generate new span for thread manually here, because we can't depend - // on OpenTelemetrySpanHolder due to link order issues. - thread_trace_context = query_context_ptr->query_trace_context; - if (thread_trace_context.trace_id != UUID()) - { - thread_trace_context.span_id = thread_local_rng(); - } - } - else - { - thread_trace_context.trace_id = 0; } initPerformanceCounters(); @@ -353,42 +332,6 @@ void ThreadStatus::detachQuery(bool exit_if_already_detached, bool thread_exits) assertState({ThreadState::AttachedToQuery}, __PRETTY_FUNCTION__); - std::shared_ptr opentelemetry_span_log; - auto query_context_ptr = query_context.lock(); - if (thread_trace_context.trace_id != UUID() && query_context_ptr) - { - opentelemetry_span_log = query_context_ptr->getOpenTelemetrySpanLog(); - } - - if (opentelemetry_span_log) - { - // Log the current thread span. - // We do this manually, because we can't use OpenTelemetrySpanHolder as a - // ThreadStatus member, because of linking issues. This file is linked - // separately, so we can reference OpenTelemetrySpanLog here, but if we had - // the span holder as a field, we would have to reference it in the - // destructor, which is in another library. - OpenTelemetrySpanLogElement span; - - span.trace_id = thread_trace_context.trace_id; - // All child span holders should be finished by the time we detach this - // thread, so the current span id should be the thread span id. If not, - // an assertion for a proper parent span in ~OpenTelemetrySpanHolder() - // is going to fail, because we're going to reset it to zero later in - // this function. - span.span_id = thread_trace_context.span_id; - assert(query_context_ptr); - span.parent_span_id = query_context_ptr->query_trace_context.span_id; - span.operation_name = getThreadName(); - span.start_time_us = query_start_time_microseconds; - span.finish_time_us = - std::chrono::duration_cast( - std::chrono::system_clock::now().time_since_epoch()).count(); - span.attributes.push_back(Tuple{"clickhouse.thread_id", toString(thread_id)}); - - opentelemetry_span_log->add(span); - } - finalizeQueryProfiler(); finalizePerformanceCounters(); @@ -404,8 +347,6 @@ void ThreadStatus::detachQuery(bool exit_if_already_detached, bool thread_exits) query_id.clear(); query_context.reset(); - thread_trace_context.trace_id = 0; - thread_trace_context.span_id = 0; /// Avoid leaking of ThreadGroupStatus::finished_threads_counters_memory /// (this is in case someone uses system thread but did not call getProfileEventsCountersAndMemoryForThreads()) diff --git a/src/Interpreters/TreeOptimizer.cpp b/src/Interpreters/TreeOptimizer.cpp index 07c7cd85d1b..eaf59731967 100644 --- a/src/Interpreters/TreeOptimizer.cpp +++ b/src/Interpreters/TreeOptimizer.cpp @@ -453,7 +453,7 @@ void optimizeMonotonousFunctionsInOrderBy(ASTSelectQuery * select_query, Context return; /// Do not apply optimization for Distributed and Merge storages, - /// because we can't get the sorting key of their undelying tables + /// because we can't get the sorting key of their underlying tables /// and we can break the matching of the sorting key for `read_in_order` /// optimization by removing monotonous functions from the prefix of key. if (result.is_remote_storage || (result.storage && result.storage->getName() == "Merge")) diff --git a/src/Interpreters/TreeRewriter.cpp b/src/Interpreters/TreeRewriter.cpp index 9248e8eecb6..73410a39ffd 100644 --- a/src/Interpreters/TreeRewriter.cpp +++ b/src/Interpreters/TreeRewriter.cpp @@ -521,10 +521,15 @@ void removeUnneededColumnsFromSelectClause(ASTSelectQuery * select_query, const ++new_elements_size; } /// removing aggregation can change number of rows, so `count()` result in outer sub-query would be wrong - if (func && AggregateUtils::isAggregateFunction(*func) && !select_query->groupBy()) + if (func && !select_query->groupBy()) { - new_elements[result_index] = elem; - ++new_elements_size; + GetAggregatesVisitor::Data data = {}; + GetAggregatesVisitor(data).visit(elem); + if (!data.aggregates.empty()) + { + new_elements[result_index] = elem; + ++new_elements_size; + } } } } diff --git a/src/Interpreters/executeQuery.cpp b/src/Interpreters/executeQuery.cpp index cdddd28adeb..c501c1722ba 100644 --- a/src/Interpreters/executeQuery.cpp +++ b/src/Interpreters/executeQuery.cpp @@ -233,7 +233,7 @@ inline UInt64 time_in_seconds(std::chrono::time_point return std::chrono::duration_cast(timepoint.time_since_epoch()).count(); } -static void onExceptionBeforeStart(const String & query_for_logging, ContextPtr context, UInt64 current_time_us, ASTPtr ast) +static void onExceptionBeforeStart(const String & query_for_logging, ContextPtr context, UInt64 current_time_us, ASTPtr ast, const std::shared_ptr & query_span) { /// Exception before the query execution. if (auto quota = context->getQuota()) @@ -291,29 +291,13 @@ static void onExceptionBeforeStart(const String & query_for_logging, ContextPtr if (auto query_log = context->getQueryLog()) query_log->add(elem); - if (auto opentelemetry_span_log = context->getOpenTelemetrySpanLog(); - context->query_trace_context.trace_id != UUID() - && opentelemetry_span_log) + if (query_span) { - OpenTelemetrySpanLogElement span; - span.trace_id = context->query_trace_context.trace_id; - span.span_id = context->query_trace_context.span_id; - span.parent_span_id = context->getClientInfo().client_trace_context.span_id; - span.operation_name = "query"; - span.start_time_us = current_time_us; - span.finish_time_us = time_in_microseconds(std::chrono::system_clock::now()); - span.attributes.reserve(6); - span.attributes.push_back(Tuple{"clickhouse.query_status", "ExceptionBeforeStart"}); - span.attributes.push_back(Tuple{"db.statement", elem.query}); - span.attributes.push_back(Tuple{"clickhouse.query_id", elem.client_info.current_query_id}); - span.attributes.push_back(Tuple{"clickhouse.exception", elem.exception}); - span.attributes.push_back(Tuple{"clickhouse.exception_code", toString(elem.exception_code)}); - if (!context->query_trace_context.tracestate.empty()) - { - span.attributes.push_back(Tuple{"clickhouse.tracestate", context->query_trace_context.tracestate}); - } - - opentelemetry_span_log->add(span); + query_span->addAttribute("clickhouse.exception_code", elem.exception_code); + query_span->addAttribute("clickhouse.exception", elem.exception); + query_span->addAttribute("db.statement", elem.query); + query_span->addAttribute("clickhouse.query_id", elem.client_info.current_query_id); + query_span->finish(); } ProfileEvents::increment(ProfileEvents::FailedQuery); @@ -364,6 +348,14 @@ static std::tuple executeQueryImpl( QueryProcessingStage::Enum stage, ReadBuffer * istr) { + /// query_span is a special span, when this function exits, it's lifetime is not ended, but ends when the query finishes. + /// Some internal queries might call this function recursively by setting 'internal' parameter to 'true', + /// to make sure SpanHolders in current stack ends in correct order, we disable this span for these internal queries + /// + /// This does not have impact on the final span logs, because these internal queries are issued by external queries, + /// we still have enough span logs for the execution of external queries. + std::shared_ptr query_span = internal ? nullptr : std::make_shared("query"); + const auto current_time = std::chrono::system_clock::now(); auto & client_info = context->getClientInfo(); @@ -465,7 +457,7 @@ static std::tuple executeQueryImpl( if (!internal) { - onExceptionBeforeStart(query_for_logging, context, time_in_microseconds(current_time), ast); + onExceptionBeforeStart(query_for_logging, context, time_in_microseconds(current_time), ast, query_span); } throw; @@ -521,13 +513,13 @@ static std::tuple executeQueryImpl( } { - SelectIntersectExceptQueryVisitor::Data data; + SelectIntersectExceptQueryVisitor::Data data{settings.intersect_default_mode, settings.except_default_mode}; SelectIntersectExceptQueryVisitor{data}.visit(ast); } { /// Normalize SelectWithUnionQuery - NormalizeSelectWithUnionQueryVisitor::Data data{context->getSettingsRef().union_default_mode}; + NormalizeSelectWithUnionQueryVisitor::Data data{settings.union_default_mode}; NormalizeSelectWithUnionQueryVisitor{data}.visit(ast); } @@ -684,12 +676,12 @@ static std::tuple executeQueryImpl( } { - std::unique_ptr span; - if (context->query_trace_context.trace_id != UUID()) + std::unique_ptr span; + if (OpenTelemetry::CurrentContext().isTraceEnabled()) { auto * raw_interpreter_ptr = interpreter.get(); std::string class_name(demangle(typeid(*raw_interpreter_ptr).name())); - span = std::make_unique(class_name + "::execute()"); + span = std::make_unique(class_name + "::execute()"); } res = interpreter->execute(); } @@ -841,7 +833,8 @@ static std::tuple executeQueryImpl( log_processors_profiles = settings.log_processors_profiles, status_info_to_query_log, implicit_txn_control, - pulling_pipeline = pipeline.pulling()](QueryPipeline & query_pipeline) mutable + pulling_pipeline = pipeline.pulling(), + query_span](QueryPipeline & query_pipeline) mutable { QueryStatus * process_list_elem = context->getProcessListElement(); @@ -944,28 +937,18 @@ static std::tuple executeQueryImpl( } } - if (auto opentelemetry_span_log = context->getOpenTelemetrySpanLog(); - context->query_trace_context.trace_id != UUID() - && opentelemetry_span_log) + if (query_span) { - OpenTelemetrySpanLogElement span; - span.trace_id = context->query_trace_context.trace_id; - span.span_id = context->query_trace_context.span_id; - span.parent_span_id = context->getClientInfo().client_trace_context.span_id; - span.operation_name = "query"; - span.start_time_us = elem.query_start_time_microseconds; - span.finish_time_us = time_in_microseconds(finish_time); - - span.attributes.reserve(4); - span.attributes.push_back(Tuple{"clickhouse.query_status", "QueryFinish"}); - span.attributes.push_back(Tuple{"db.statement", elem.query}); - span.attributes.push_back(Tuple{"clickhouse.query_id", elem.client_info.current_query_id}); - if (!context->query_trace_context.tracestate.empty()) - { - span.attributes.push_back(Tuple{"clickhouse.tracestate", context->query_trace_context.tracestate}); - } - - opentelemetry_span_log->add(span); + query_span->addAttribute("db.statement", elem.query); + query_span->addAttribute("clickhouse.query_id", elem.client_info.current_query_id); + query_span->addAttribute("clickhouse.query_status", "QueryFinish"); + query_span->addAttributeIfNotEmpty("clickhouse.tracestate", OpenTelemetry::CurrentContext().tracestate); + query_span->addAttributeIfNotZero("clickhouse.read_rows", elem.read_rows); + query_span->addAttributeIfNotZero("clickhouse.read_bytes", elem.read_bytes); + query_span->addAttributeIfNotZero("clickhouse.written_rows", info.written_rows); + query_span->addAttributeIfNotZero("clickhouse.written_bytes", elem.written_bytes); + query_span->addAttributeIfNotZero("clickhouse.memory_usage", elem.memory_usage); + query_span->finish(); } if (implicit_txn_control) @@ -993,7 +976,8 @@ static std::tuple executeQueryImpl( log_queries_min_query_duration_ms = settings.log_queries_min_query_duration_ms.totalMilliseconds(), quota(quota), status_info_to_query_log, - implicit_txn_control]() mutable + implicit_txn_control, + query_span]() mutable { if (implicit_txn_control) { @@ -1050,6 +1034,15 @@ static std::tuple executeQueryImpl( { ProfileEvents::increment(ProfileEvents::FailedInsertQuery); } + + if (query_span) + { + query_span->addAttribute("db.statement", elem.query); + query_span->addAttribute("clickhouse.query_id", elem.client_info.current_query_id); + query_span->addAttribute("clickhouse.exception", elem.exception); + query_span->addAttribute("clickhouse.exception_code", elem.exception_code); + query_span->finish(); + } }; res.finish_callback = std::move(finish_callback); @@ -1073,7 +1066,7 @@ static std::tuple executeQueryImpl( if (query_for_logging.empty()) query_for_logging = prepareQueryForLogging(query, context); - onExceptionBeforeStart(query_for_logging, context, time_in_microseconds(current_time), ast); + onExceptionBeforeStart(query_for_logging, context, time_in_microseconds(current_time), ast, query_span); } throw; diff --git a/src/Interpreters/tests/gtest_lru_file_cache.cpp b/src/Interpreters/tests/gtest_lru_file_cache.cpp new file mode 100644 index 00000000000..6460eeef8c5 --- /dev/null +++ b/src/Interpreters/tests/gtest_lru_file_cache.cpp @@ -0,0 +1,518 @@ +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +namespace fs = std::filesystem; + +fs::path caches_dir = fs::current_path() / "lru_cache_test"; +String cache_base_path = caches_dir / "cache1" / ""; + +void assertRange( + [[maybe_unused]] size_t assert_n, DB::FileSegmentPtr file_segment, + const DB::FileSegment::Range & expected_range, DB::FileSegment::State expected_state) +{ + auto range = file_segment->range(); + + std::cerr << fmt::format("\nAssert #{} : {} == {} (state: {} == {})\n", assert_n, + range.toString(), expected_range.toString(), + toString(file_segment->state()), toString(expected_state)); + + ASSERT_EQ(range.left, expected_range.left); + ASSERT_EQ(range.right, expected_range.right); + ASSERT_EQ(file_segment->state(), expected_state); +} + +void printRanges(const auto & segments) +{ + std::cerr << "\nHaving file segments: "; + for (const auto & segment : segments) + std::cerr << '\n' << segment->range().toString() << " (state: " + DB::FileSegment::stateToString(segment->state()) + ")" << "\n"; +} + +std::vector fromHolder(const DB::FileSegmentsHolder & holder) +{ + return std::vector(holder.file_segments.begin(), holder.file_segments.end()); +} + +String getFileSegmentPath(const String & base_path, const DB::FileCache::Key & key, size_t offset) +{ + auto key_str = key.toString(); + return fs::path(base_path) / key_str.substr(0, 3) / key_str / DB::toString(offset); +} + +void download(DB::FileSegmentPtr file_segment) +{ + const auto & key = file_segment->key(); + size_t size = file_segment->range().size(); + + auto key_str = key.toString(); + auto subdir = fs::path(cache_base_path) / key_str.substr(0, 3) / key_str; + if (!fs::exists(subdir)) + fs::create_directories(subdir); + + std::string data(size, '0'); + file_segment->write(data.data(), size, file_segment->getDownloadOffset()); +} + +void prepareAndDownload(DB::FileSegmentPtr file_segment) +{ + // std::cerr << "Reserving: " << file_segment->range().size() << " for: " << file_segment->range().toString() << "\n"; + ASSERT_TRUE(file_segment->reserve(file_segment->range().size())); + download(file_segment); +} + +void complete(const DB::FileSegmentsHolder & holder) +{ + for (const auto & file_segment : holder.file_segments) + { + ASSERT_TRUE(file_segment->getOrSetDownloader() == DB::FileSegment::getCallerId()); + prepareAndDownload(file_segment); + file_segment->completeWithState(DB::FileSegment::State::DOWNLOADED); + } +} + + +TEST(FileCache, get) +{ + if (fs::exists(cache_base_path)) + fs::remove_all(cache_base_path); + + DB::ThreadStatus thread_status; + + /// To work with cache need query_id and query context. + std::string query_id = "query_id"; + auto query_context = DB::Context::createCopy(getContext().context); + query_context->makeQueryContext(); + query_context->setCurrentQueryId(query_id); + DB::CurrentThread::QueryScope query_scope_holder(query_context); + + DB::FileCacheSettings settings; + settings.max_size = 30; + settings.max_elements = 5; + + { + auto cache = DB::FileCache(cache_base_path, settings); + cache.initialize(); + auto key = cache.hash("key1"); + + { + auto holder = cache.getOrSet(key, 0, 10, false); /// Add range [0, 9] + auto segments = fromHolder(holder); + /// Range was not present in cache. It should be added in cache as one while file segment. + ASSERT_EQ(segments.size(), 1); + + assertRange(1, segments[0], DB::FileSegment::Range(0, 9), DB::FileSegment::State::EMPTY); + + /// Exception because space not reserved. + /// EXPECT_THROW(download(segments[0]), DB::Exception); + /// Exception because space can be reserved only by downloader + /// EXPECT_THROW(segments[0]->reserve(segments[0]->range().size()), DB::Exception); + + ASSERT_TRUE(segments[0]->getOrSetDownloader() == DB::FileSegment::getCallerId()); + ASSERT_TRUE(segments[0]->reserve(segments[0]->range().size())); + assertRange(2, segments[0], DB::FileSegment::Range(0, 9), DB::FileSegment::State::DOWNLOADING); + + download(segments[0]); + segments[0]->completeWithState(DB::FileSegment::State::DOWNLOADED); + assertRange(3, segments[0], DB::FileSegment::Range(0, 9), DB::FileSegment::State::DOWNLOADED); + } + + /// Current cache: [__________] + /// ^ ^ + /// 0 9 + ASSERT_EQ(cache.getFileSegmentsNum(), 1); + ASSERT_EQ(cache.getUsedCacheSize(), 10); + + { + /// Want range [5, 14], but [0, 9] already in cache, so only [10, 14] will be put in cache. + auto holder = cache.getOrSet(key, 5, 10, false); + auto segments = fromHolder(holder); + ASSERT_EQ(segments.size(), 2); + + assertRange(4, segments[0], DB::FileSegment::Range(0, 9), DB::FileSegment::State::DOWNLOADED); + assertRange(5, segments[1], DB::FileSegment::Range(10, 14), DB::FileSegment::State::EMPTY); + + ASSERT_TRUE(segments[1]->getOrSetDownloader() == DB::FileSegment::getCallerId()); + prepareAndDownload(segments[1]); + segments[1]->completeWithState(DB::FileSegment::State::DOWNLOADED); + assertRange(6, segments[1], DB::FileSegment::Range(10, 14), DB::FileSegment::State::DOWNLOADED); + } + + /// Current cache: [__________][_____] + /// ^ ^^ ^ + /// 0 910 14 + ASSERT_EQ(cache.getFileSegmentsNum(), 2); + ASSERT_EQ(cache.getUsedCacheSize(), 15); + + { + auto holder = cache.getOrSet(key, 9, 1, false); /// Get [9, 9] + auto segments = fromHolder(holder); + ASSERT_EQ(segments.size(), 1); + assertRange(7, segments[0], DB::FileSegment::Range(0, 9), DB::FileSegment::State::DOWNLOADED); + } + + { + auto holder = cache.getOrSet(key, 9, 2, false); /// Get [9, 10] + auto segments = fromHolder(holder); + ASSERT_EQ(segments.size(), 2); + assertRange(8, segments[0], DB::FileSegment::Range(0, 9), DB::FileSegment::State::DOWNLOADED); + assertRange(9, segments[1], DB::FileSegment::Range(10, 14), DB::FileSegment::State::DOWNLOADED); + } + + { + auto holder = cache.getOrSet(key, 10, 1, false); /// Get [10, 10] + auto segments = fromHolder(holder); + ASSERT_EQ(segments.size(), 1); + assertRange(10, segments[0], DB::FileSegment::Range(10, 14), DB::FileSegment::State::DOWNLOADED); + } + + complete(cache.getOrSet(key, 17, 4, false)); /// Get [17, 20] + complete(cache.getOrSet(key, 24, 3, false)); /// Get [24, 26] + /// complete(cache.getOrSet(key, 27, 1, false)); /// Get [27, 27] + + /// Current cache: [__________][_____] [____] [___][] + /// ^ ^^ ^ ^ ^ ^ ^^^ + /// 0 910 14 17 20 24 2627 + /// + ASSERT_EQ(cache.getFileSegmentsNum(), 4); + ASSERT_EQ(cache.getUsedCacheSize(), 22); + + { + auto holder = cache.getOrSet(key, 0, 26, false); /// Get [0, 25] + auto segments = fromHolder(holder); + ASSERT_EQ(segments.size(), 6); + + assertRange(11, segments[0], DB::FileSegment::Range(0, 9), DB::FileSegment::State::DOWNLOADED); + assertRange(12, segments[1], DB::FileSegment::Range(10, 14), DB::FileSegment::State::DOWNLOADED); + + /// Missing [15, 16] should be added in cache. + assertRange(13, segments[2], DB::FileSegment::Range(15, 16), DB::FileSegment::State::EMPTY); + + ASSERT_TRUE(segments[2]->getOrSetDownloader() == DB::FileSegment::getCallerId()); + prepareAndDownload(segments[2]); + + segments[2]->completeWithState(DB::FileSegment::State::DOWNLOADED); + + assertRange(14, segments[3], DB::FileSegment::Range(17, 20), DB::FileSegment::State::DOWNLOADED); + + /// New [21, 23], but will not be added in cache because of elements limit (5) + assertRange(15, segments[4], DB::FileSegment::Range(21, 23), DB::FileSegment::State::EMPTY); + ASSERT_TRUE(segments[4]->getOrSetDownloader() == DB::FileSegment::getCallerId()); + ASSERT_FALSE(segments[4]->reserve(1)); + + assertRange(16, segments[5], DB::FileSegment::Range(24, 26), DB::FileSegment::State::DOWNLOADED); + + /// Current cache: [__________][_____][ ][____] [___] + /// ^ ^ ^ + /// 0 20 24 + /// + + /// Range [27, 27] must be evicted in previous getOrSet [0, 25]. + /// Let's not invalidate pointers to returned segments from range [0, 25] and + /// as max elements size is reached, next attempt to put something in cache should fail. + /// This will also check that [27, 27] was indeed evicted. + + auto holder1 = cache.getOrSet(key, 27, 1, false); + auto segments_1 = fromHolder(holder1); /// Get [27, 27] + ASSERT_EQ(segments_1.size(), 1); + assertRange(17, segments_1[0], DB::FileSegment::Range(27, 27), DB::FileSegment::State::EMPTY); + } + + { + auto holder = cache.getOrSet(key, 12, 10, false); /// Get [12, 21] + auto segments = fromHolder(holder); + ASSERT_EQ(segments.size(), 4); + + assertRange(18, segments[0], DB::FileSegment::Range(10, 14), DB::FileSegment::State::DOWNLOADED); + assertRange(19, segments[1], DB::FileSegment::Range(15, 16), DB::FileSegment::State::DOWNLOADED); + assertRange(20, segments[2], DB::FileSegment::Range(17, 20), DB::FileSegment::State::DOWNLOADED); + + assertRange(21, segments[3], DB::FileSegment::Range(21, 21), DB::FileSegment::State::EMPTY); + + ASSERT_TRUE(segments[3]->getOrSetDownloader() == DB::FileSegment::getCallerId()); + prepareAndDownload(segments[3]); + + segments[3]->completeWithState(DB::FileSegment::State::DOWNLOADED); + ASSERT_TRUE(segments[3]->state() == DB::FileSegment::State::DOWNLOADED); + } + + /// Current cache: [_____][__][____][_] [___] + /// ^ ^ ^ ^ ^ + /// 10 17 21 24 26 + + ASSERT_EQ(cache.getFileSegmentsNum(), 5); + + { + auto holder = cache.getOrSet(key, 23, 5, false); /// Get [23, 28] + auto segments = fromHolder(holder); + ASSERT_EQ(segments.size(), 3); + + assertRange(22, segments[0], DB::FileSegment::Range(23, 23), DB::FileSegment::State::EMPTY); + assertRange(23, segments[1], DB::FileSegment::Range(24, 26), DB::FileSegment::State::DOWNLOADED); + assertRange(24, segments[2], DB::FileSegment::Range(27, 27), DB::FileSegment::State::EMPTY); + + ASSERT_TRUE(segments[0]->getOrSetDownloader() == DB::FileSegment::getCallerId()); + ASSERT_TRUE(segments[2]->getOrSetDownloader() == DB::FileSegment::getCallerId()); + prepareAndDownload(segments[0]); + prepareAndDownload(segments[2]); + segments[0]->completeWithState(DB::FileSegment::State::DOWNLOADED); + segments[2]->completeWithState(DB::FileSegment::State::DOWNLOADED); + } + + /// Current cache: [____][_] [][___][__] + /// ^ ^ ^^^ ^^ ^ + /// 17 21 2324 26 28 + + { + auto holder5 = cache.getOrSet(key, 2, 3,false); /// Get [2, 4] + auto s5 = fromHolder(holder5); + ASSERT_EQ(s5.size(), 1); + assertRange(25, s5[0], DB::FileSegment::Range(2, 4), DB::FileSegment::State::EMPTY); + + auto holder1 = cache.getOrSet(key, 30, 2, false); /// Get [30, 31] + auto s1 = fromHolder(holder1); + ASSERT_EQ(s1.size(), 1); + assertRange(26, s1[0], DB::FileSegment::Range(30, 31), DB::FileSegment::State::EMPTY); + + ASSERT_TRUE(s5[0]->getOrSetDownloader() == DB::FileSegment::getCallerId()); + ASSERT_TRUE(s1[0]->getOrSetDownloader() == DB::FileSegment::getCallerId()); + prepareAndDownload(s5[0]); + prepareAndDownload(s1[0]); + s5[0]->completeWithState(DB::FileSegment::State::DOWNLOADED); + s1[0]->completeWithState(DB::FileSegment::State::DOWNLOADED); + + /// Current cache: [___] [_][___][_] [__] + /// ^ ^ ^ ^ ^ ^ ^ ^ + /// 2 4 23 24 26 27 30 31 + + auto holder2 = cache.getOrSet(key, 23, 1, false); /// Get [23, 23] + auto s2 = fromHolder(holder2); + ASSERT_EQ(s2.size(), 1); + + auto holder3 = cache.getOrSet(key, 24, 3, false); /// Get [24, 26] + auto s3 = fromHolder(holder3); + ASSERT_EQ(s3.size(), 1); + + auto holder4 = cache.getOrSet(key, 27, 1, false); /// Get [27, 27] + auto s4 = fromHolder(holder4); + ASSERT_EQ(s4.size(), 1); + + /// All cache is now unreleasable because pointers are still hold + auto holder6 = cache.getOrSet(key, 0, 40, false); + auto f = fromHolder(holder6); + ASSERT_EQ(f.size(), 9); + + assertRange(27, f[0], DB::FileSegment::Range(0, 1), DB::FileSegment::State::EMPTY); + assertRange(28, f[2], DB::FileSegment::Range(5, 22), DB::FileSegment::State::EMPTY); + assertRange(29, f[6], DB::FileSegment::Range(28, 29), DB::FileSegment::State::EMPTY); + assertRange(30, f[8], DB::FileSegment::Range(32, 39), DB::FileSegment::State::EMPTY); + + ASSERT_TRUE(f[0]->getOrSetDownloader() == DB::FileSegment::getCallerId()); + ASSERT_TRUE(f[2]->getOrSetDownloader() == DB::FileSegment::getCallerId()); + ASSERT_TRUE(f[6]->getOrSetDownloader() == DB::FileSegment::getCallerId()); + ASSERT_TRUE(f[8]->getOrSetDownloader() == DB::FileSegment::getCallerId()); + + ASSERT_FALSE(f[0]->reserve(1)); + ASSERT_FALSE(f[2]->reserve(1)); + ASSERT_FALSE(f[6]->reserve(1)); + ASSERT_FALSE(f[8]->reserve(1)); + } + + { + auto holder = cache.getOrSet(key, 2, 3, false); /// Get [2, 4] + auto segments = fromHolder(holder); + ASSERT_EQ(segments.size(), 1); + assertRange(31, segments[0], DB::FileSegment::Range(2, 4), DB::FileSegment::State::DOWNLOADED); + } + + /// Current cache: [___] [_][___][_] [__] + /// ^ ^ ^ ^ ^ ^ ^ ^ + /// 2 4 23 24 26 27 30 31 + + { + auto holder = cache.getOrSet(key, 25, 5, false); /// Get [25, 29] + auto segments = fromHolder(holder); + ASSERT_EQ(segments.size(), 3); + + assertRange(32, segments[0], DB::FileSegment::Range(24, 26), DB::FileSegment::State::DOWNLOADED); + assertRange(33, segments[1], DB::FileSegment::Range(27, 27), DB::FileSegment::State::DOWNLOADED); + + assertRange(34, segments[2], DB::FileSegment::Range(28, 29), DB::FileSegment::State::EMPTY); + ASSERT_TRUE(segments[2]->getOrSetDownloader() == DB::FileSegment::getCallerId()); + ASSERT_TRUE(segments[2]->state() == DB::FileSegment::State::DOWNLOADING); + + bool lets_start_download = false; + std::mutex mutex; + std::condition_variable cv; + + std::thread other_1([&] + { + DB::ThreadStatus thread_status_1; + auto query_context_1 = DB::Context::createCopy(getContext().context); + query_context_1->makeQueryContext(); + query_context_1->setCurrentQueryId("query_id_1"); + DB::CurrentThread::QueryScope query_scope_holder_1(query_context_1); + thread_status_1.attachQueryContext(query_context_1); + + auto holder_2 = cache.getOrSet(key, 25, 5, false); /// Get [25, 29] once again. + auto segments_2 = fromHolder(holder_2); + ASSERT_EQ(segments.size(), 3); + + assertRange(35, segments_2[0], DB::FileSegment::Range(24, 26), DB::FileSegment::State::DOWNLOADED); + assertRange(36, segments_2[1], DB::FileSegment::Range(27, 27), DB::FileSegment::State::DOWNLOADED); + assertRange(37, segments_2[2], DB::FileSegment::Range(28, 29), DB::FileSegment::State::DOWNLOADING); + + ASSERT_TRUE(segments[2]->getOrSetDownloader() != DB::FileSegment::getCallerId()); + ASSERT_TRUE(segments[2]->state() == DB::FileSegment::State::DOWNLOADING); + + { + std::lock_guard lock(mutex); + lets_start_download = true; + } + cv.notify_one(); + + segments_2[2]->wait(); + ASSERT_TRUE(segments_2[2]->state() == DB::FileSegment::State::DOWNLOADED); + }); + + { + std::unique_lock lock(mutex); + cv.wait(lock, [&]{ return lets_start_download; }); + } + + prepareAndDownload(segments[2]); + segments[2]->completeWithState(DB::FileSegment::State::DOWNLOADED); + ASSERT_TRUE(segments[2]->state() == DB::FileSegment::State::DOWNLOADED); + + other_1.join(); + } + + /// Current cache: [___] [___][_][__][__] + /// ^ ^ ^ ^ ^^ ^^ ^ + /// 2 4 24 26 27 2930 31 + + { + /// Now let's check the similar case but getting ERROR state after segment->wait(), when + /// state is changed not manually via segment->complete(state) but from destructor of holder + /// and notify_all() is also called from destructor of holder. + + std::optional holder; + holder.emplace(cache.getOrSet(key, 3, 23, false)); /// Get [3, 25] + + auto segments = fromHolder(*holder); + ASSERT_EQ(segments.size(), 3); + + assertRange(38, segments[0], DB::FileSegment::Range(2, 4), DB::FileSegment::State::DOWNLOADED); + + assertRange(39, segments[1], DB::FileSegment::Range(5, 23), DB::FileSegment::State::EMPTY); + ASSERT_TRUE(segments[1]->getOrSetDownloader() == DB::FileSegment::getCallerId()); + ASSERT_TRUE(segments[1]->state() == DB::FileSegment::State::DOWNLOADING); + + assertRange(40, segments[2], DB::FileSegment::Range(24, 26), DB::FileSegment::State::DOWNLOADED); + + bool lets_start_download = false; + std::mutex mutex; + std::condition_variable cv; + + std::thread other_1([&] + { + DB::ThreadStatus thread_status_1; + auto query_context_1 = DB::Context::createCopy(getContext().context); + query_context_1->makeQueryContext(); + query_context_1->setCurrentQueryId("query_id_1"); + DB::CurrentThread::QueryScope query_scope_holder_1(query_context_1); + thread_status_1.attachQueryContext(query_context_1); + + auto holder_2 = cache.getOrSet(key, 3, 23, false); /// Get [3, 25] once again + auto segments_2 = fromHolder(*holder); + ASSERT_EQ(segments_2.size(), 3); + + assertRange(41, segments_2[0], DB::FileSegment::Range(2, 4), DB::FileSegment::State::DOWNLOADED); + assertRange(42, segments_2[1], DB::FileSegment::Range(5, 23), DB::FileSegment::State::DOWNLOADING); + assertRange(43, segments_2[2], DB::FileSegment::Range(24, 26), DB::FileSegment::State::DOWNLOADED); + + ASSERT_TRUE(segments_2[1]->getDownloader() != DB::FileSegment::getCallerId()); + ASSERT_TRUE(segments_2[1]->state() == DB::FileSegment::State::DOWNLOADING); + + { + std::lock_guard lock(mutex); + lets_start_download = true; + } + cv.notify_one(); + + segments_2[1]->wait(); + printRanges(segments_2); + ASSERT_TRUE(segments_2[1]->state() == DB::FileSegment::State::PARTIALLY_DOWNLOADED); + + ASSERT_TRUE(segments_2[1]->getOrSetDownloader() == DB::FileSegment::getCallerId()); + prepareAndDownload(segments_2[1]); + segments_2[1]->completeWithState(DB::FileSegment::State::DOWNLOADED); + }); + + { + std::unique_lock lock(mutex); + cv.wait(lock, [&]{ return lets_start_download; }); + } + + holder.reset(); + other_1.join(); + printRanges(segments); + ASSERT_TRUE(segments[1]->state() == DB::FileSegment::State::DOWNLOADED); + } + } + + /// Current cache: [___][ ][___][_][__] + /// ^ ^^ ^ ^^ ^ ^ + /// 2 45 24 2627 28 29 + + { + /// Test LRUCache::restore(). + + auto cache2 = DB::FileCache(cache_base_path, settings); + cache2.initialize(); + auto key = cache2.hash("key1"); + + auto holder1 = cache2.getOrSet(key, 2, 28, false); /// Get [2, 29] + + auto segments1 = fromHolder(holder1); + ASSERT_EQ(segments1.size(), 5); + + assertRange(44, segments1[0], DB::FileSegment::Range(2, 4), DB::FileSegment::State::DOWNLOADED); + assertRange(45, segments1[1], DB::FileSegment::Range(5, 23), DB::FileSegment::State::DOWNLOADED); + assertRange(45, segments1[2], DB::FileSegment::Range(24, 26), DB::FileSegment::State::DOWNLOADED); + assertRange(46, segments1[3], DB::FileSegment::Range(27, 27), DB::FileSegment::State::DOWNLOADED); + assertRange(47, segments1[4], DB::FileSegment::Range(28, 29), DB::FileSegment::State::DOWNLOADED); + } + + { + /// Test max file segment size + + auto settings2 = settings; + settings2.max_file_segment_size = 10; + auto cache2 = DB::FileCache(caches_dir / "cache2", settings2); + cache2.initialize(); + auto key = cache2.hash("key1"); + + auto holder1 = cache2.getOrSet(key, 0, 25, false); /// Get [0, 24] + auto segments1 = fromHolder(holder1); + + ASSERT_EQ(segments1.size(), 3); + assertRange(48, segments1[0], DB::FileSegment::Range(0, 9), DB::FileSegment::State::EMPTY); + assertRange(49, segments1[1], DB::FileSegment::Range(10, 19), DB::FileSegment::State::EMPTY); + assertRange(50, segments1[2], DB::FileSegment::Range(20, 24), DB::FileSegment::State::EMPTY); + } + +} diff --git a/src/Parsers/ASTSelectIntersectExceptQuery.cpp b/src/Parsers/ASTSelectIntersectExceptQuery.cpp index 62eeefba385..75fbe2b5280 100644 --- a/src/Parsers/ASTSelectIntersectExceptQuery.cpp +++ b/src/Parsers/ASTSelectIntersectExceptQuery.cpp @@ -27,7 +27,7 @@ void ASTSelectIntersectExceptQuery::formatImpl(const FormatSettings & settings, if (it != children.begin()) { settings.ostr << settings.nl_or_ws << indent_str << (settings.hilite ? hilite_keyword : "") - << (final_operator == Operator::INTERSECT ? "INTERSECT" : "EXCEPT") + << fromOperator(final_operator) << (settings.hilite ? hilite_none : "") << settings.nl_or_ws; } @@ -53,4 +53,20 @@ ASTs ASTSelectIntersectExceptQuery::getListOfSelects() const return selects; } +const char * ASTSelectIntersectExceptQuery::fromOperator(Operator op) +{ + switch (op) + { + case Operator::EXCEPT_ALL: + return "EXCEPT ALL"; + case Operator::EXCEPT_DISTINCT: + return "EXCEPT DISTINCT"; + case Operator::INTERSECT_ALL: + return "INTERSECT ALL"; + case Operator::INTERSECT_DISTINCT: + return "INTERSECT DISTINCT"; + default: + return ""; + } +} } diff --git a/src/Parsers/ASTSelectIntersectExceptQuery.h b/src/Parsers/ASTSelectIntersectExceptQuery.h index c95944a0c35..ad962fe25e2 100644 --- a/src/Parsers/ASTSelectIntersectExceptQuery.h +++ b/src/Parsers/ASTSelectIntersectExceptQuery.h @@ -1,6 +1,7 @@ #pragma once #include +#include "Parsers/ExpressionListParsers.h" namespace DB @@ -16,8 +17,10 @@ public: enum class Operator { UNKNOWN, - INTERSECT, - EXCEPT + EXCEPT_ALL, + EXCEPT_DISTINCT, + INTERSECT_ALL, + INTERSECT_DISTINCT, }; void formatImpl(const FormatSettings & settings, FormatState & state, FormatStateStacked frame) const override; @@ -26,6 +29,8 @@ public: ASTs getListOfSelects() const; + static const char * fromOperator(Operator op); + /// Final operator after applying visitor. Operator final_operator = Operator::UNKNOWN; }; diff --git a/src/Parsers/ASTSelectWithUnionQuery.cpp b/src/Parsers/ASTSelectWithUnionQuery.cpp index 7718b2dc2de..bc413fbe16d 100644 --- a/src/Parsers/ASTSelectWithUnionQuery.cpp +++ b/src/Parsers/ASTSelectWithUnionQuery.cpp @@ -1,6 +1,7 @@ #include #include #include +#include #include #include @@ -32,14 +33,18 @@ void ASTSelectWithUnionQuery::formatQueryImpl(const FormatSettings & settings, F auto mode_to_str = [&](auto mode) { - if (mode == SelectUnionMode::ALL) + if (mode == SelectUnionMode::UNION_ALL) return "UNION ALL"; - else if (mode == SelectUnionMode::DISTINCT) + else if (mode == SelectUnionMode::UNION_DISTINCT) return "UNION DISTINCT"; - else if (mode == SelectUnionMode::INTERSECT) - return "INTERSECT"; - else if (mode == SelectUnionMode::EXCEPT) - return "EXCEPT"; + else if (mode == SelectUnionMode::EXCEPT_ALL) + return "EXCEPT ALL"; + else if (mode == SelectUnionMode::EXCEPT_DISTINCT) + return "EXCEPT DISTINCT"; + else if (mode == SelectUnionMode::INTERSECT_ALL) + return "INTERSECT ALL"; + else if (mode == SelectUnionMode::INTERSECT_DISTINCT) + return "INTERSECT DISTINCT"; return ""; }; @@ -77,8 +82,8 @@ void ASTSelectWithUnionQuery::formatQueryImpl(const FormatSettings & settings, F bool ASTSelectWithUnionQuery::hasNonDefaultUnionMode() const { - return set_of_modes.contains(SelectUnionMode::DISTINCT) || set_of_modes.contains(SelectUnionMode::INTERSECT) - || set_of_modes.contains(SelectUnionMode::EXCEPT); + return set_of_modes.contains(SelectUnionMode::UNION_DISTINCT) || set_of_modes.contains(SelectUnionMode::INTERSECT_DISTINCT) + || set_of_modes.contains(SelectUnionMode::EXCEPT_DISTINCT); } } diff --git a/src/Parsers/ExpressionListParsers.cpp b/src/Parsers/ExpressionListParsers.cpp index ea51367ee5d..4e88e5c68e6 100644 --- a/src/Parsers/ExpressionListParsers.cpp +++ b/src/Parsers/ExpressionListParsers.cpp @@ -139,36 +139,38 @@ bool ParserUnionList::parseImpl(Pos & pos, ASTPtr & node, Expected & expected) return true; }; - /// Parse UNION type + /// Parse UNION / INTERSECT / EXCEPT mode + /// The mode can be DEFAULT (unspecified) / DISTINCT / ALL auto parse_separator = [&] { if (s_union_parser.ignore(pos, expected)) { - // SELECT ... UNION ALL SELECT ... if (s_all_parser.check(pos, expected)) - { - union_modes.push_back(SelectUnionMode::ALL); - } - // SELECT ... UNION DISTINCT SELECT ... + union_modes.push_back(SelectUnionMode::UNION_ALL); else if (s_distinct_parser.check(pos, expected)) - { - union_modes.push_back(SelectUnionMode::DISTINCT); - } - // SELECT ... UNION SELECT ... + union_modes.push_back(SelectUnionMode::UNION_DISTINCT); else - { - union_modes.push_back(SelectUnionMode::Unspecified); - } + union_modes.push_back(SelectUnionMode::UNION_DEFAULT); return true; } else if (s_except_parser.check(pos, expected)) { - union_modes.push_back(SelectUnionMode::EXCEPT); + if (s_all_parser.check(pos, expected)) + union_modes.push_back(SelectUnionMode::EXCEPT_ALL); + else if (s_distinct_parser.check(pos, expected)) + union_modes.push_back(SelectUnionMode::EXCEPT_DISTINCT); + else + union_modes.push_back(SelectUnionMode::EXCEPT_DEFAULT); return true; } else if (s_intersect_parser.check(pos, expected)) { - union_modes.push_back(SelectUnionMode::INTERSECT); + if (s_all_parser.check(pos, expected)) + union_modes.push_back(SelectUnionMode::INTERSECT_ALL); + else if (s_distinct_parser.check(pos, expected)) + union_modes.push_back(SelectUnionMode::INTERSECT_DISTINCT); + else + union_modes.push_back(SelectUnionMode::INTERSECT_DEFAULT); return true; } return false; diff --git a/src/Parsers/InsertQuerySettingsPushDownVisitor.cpp b/src/Parsers/InsertQuerySettingsPushDownVisitor.cpp index a3bca76816f..1cebdfde957 100644 --- a/src/Parsers/InsertQuerySettingsPushDownVisitor.cpp +++ b/src/Parsers/InsertQuerySettingsPushDownVisitor.cpp @@ -55,7 +55,7 @@ void InsertQuerySettingsPushDownMatcher::visit(ASTSelectQuery & select_query, AS insert_settings.push_back(setting); else { - /// Do not ovewrite setting that was passed for INSERT + /// Do not overwrite setting that was passed for INSERT /// by settings that was passed for SELECT } } diff --git a/src/Parsers/QueryWithOutputSettingsPushDownVisitor.h b/src/Parsers/QueryWithOutputSettingsPushDownVisitor.h index 2a7ed0125fa..fde8a07b555 100644 --- a/src/Parsers/QueryWithOutputSettingsPushDownVisitor.h +++ b/src/Parsers/QueryWithOutputSettingsPushDownVisitor.h @@ -11,7 +11,7 @@ struct SettingChange; class SettingsChanges; /// Pushdown SETTINGS clause that goes after FORMAT to the SELECT query: -/// (since settings after FORMAT parsed separatelly not in the ParserSelectQuery but in ParserQueryWithOutput) +/// (since settings after FORMAT parsed separately not in the ParserSelectQuery but in ParserQueryWithOutput) /// /// SELECT 1 FORMAT Null SETTINGS max_block_size = 1 -> /// SELECT 1 SETTINGS max_block_size = 1 FORMAT Null SETTINGS max_block_size = 1 diff --git a/src/Parsers/SelectUnionMode.h b/src/Parsers/SelectUnionMode.h index f4ca858d043..ca3637612aa 100644 --- a/src/Parsers/SelectUnionMode.h +++ b/src/Parsers/SelectUnionMode.h @@ -7,11 +7,15 @@ namespace DB { enum class SelectUnionMode { - Unspecified, - ALL, - DISTINCT, - EXCEPT, - INTERSECT + UNION_DEFAULT, + UNION_ALL, + UNION_DISTINCT, + EXCEPT_DEFAULT, + EXCEPT_ALL, + EXCEPT_DISTINCT, + INTERSECT_DEFAULT, + INTERSECT_ALL, + INTERSECT_DISTINCT }; using SelectUnionModes = std::vector; diff --git a/src/Processors/Executors/ExecutionThreadContext.cpp b/src/Processors/Executors/ExecutionThreadContext.cpp index 7631cb09f61..eddc1b76d8a 100644 --- a/src/Processors/Executors/ExecutionThreadContext.cpp +++ b/src/Processors/Executors/ExecutionThreadContext.cpp @@ -71,11 +71,11 @@ static void executeJob(ExecutingGraph::Node * node, ReadProgressCallback * read_ bool ExecutionThreadContext::executeTask() { - std::unique_ptr span; + std::unique_ptr span; if (trace_processors) { - span = std::make_unique("ExecutionThreadContext::executeTask() " + node->processor->getName()); + span = std::make_unique("ExecutionThreadContext::executeTask() " + node->processor->getName()); span->addAttribute("thread_number", thread_number); } std::optional execution_time_watch; diff --git a/src/Processors/Formats/Impl/ParquetBlockInputFormat.cpp b/src/Processors/Formats/Impl/ParquetBlockInputFormat.cpp index 12fa9710c42..427c159314b 100644 --- a/src/Processors/Formats/Impl/ParquetBlockInputFormat.cpp +++ b/src/Processors/Formats/Impl/ParquetBlockInputFormat.cpp @@ -55,7 +55,16 @@ Chunk ParquetBlockInputFormat::generate() return res; std::shared_ptr table; - arrow::Status read_status = file_reader->ReadRowGroup(row_group_current, column_indices, &table); + + std::unique_ptr<::arrow::RecordBatchReader> rbr; + std::vector row_group_indices { row_group_current }; + arrow::Status get_batch_reader_status = file_reader->GetRecordBatchReader(row_group_indices, column_indices, &rbr); + + if (!get_batch_reader_status.ok()) + throw ParsingException{"Error while reading Parquet data: " + get_batch_reader_status.ToString(), ErrorCodes::CANNOT_READ_ALL_DATA}; + + arrow::Status read_status = rbr->ReadAll(&table); + if (!read_status.ok()) throw ParsingException{"Error while reading Parquet data: " + read_status.ToString(), ErrorCodes::CANNOT_READ_ALL_DATA}; diff --git a/src/Processors/Merges/Algorithms/Graphite.cpp b/src/Processors/Merges/Algorithms/Graphite.cpp index 2448a1e2a94..c5c611366ff 100644 --- a/src/Processors/Merges/Algorithms/Graphite.cpp +++ b/src/Processors/Merges/Algorithms/Graphite.cpp @@ -103,17 +103,17 @@ Graphite::RollupRule selectPatternForPath( if (first_match->type == first_match->TypeUndef && pattern.type == pattern.TypeAll) { /// There is only default pattern for both retention and aggregation - return std::pair(&pattern, &pattern); + return {&pattern, &pattern}; } if (pattern.type != first_match->type) { if (first_match->type == first_match->TypeRetention) { - return std::pair(first_match, &pattern); + return {first_match, &pattern}; } if (first_match->type == first_match->TypeAggregation) { - return std::pair(&pattern, first_match); + return {&pattern, first_match}; } } } @@ -125,7 +125,7 @@ Graphite::RollupRule selectPatternForPath( if (pattern.type == pattern.TypeAll) { /// Only for not default patterns with both function and retention parameters - return std::pair(&pattern, &pattern); + return {&pattern, &pattern}; } if (first_match->type == first_match->TypeUndef) { @@ -136,11 +136,11 @@ Graphite::RollupRule selectPatternForPath( { if (first_match->type == first_match->TypeRetention) { - return std::pair(first_match, &pattern); + return {first_match, &pattern}; } if (first_match->type == first_match->TypeAggregation) { - return std::pair(&pattern, first_match); + return {&pattern, first_match}; } } } diff --git a/src/Processors/PingPongProcessor.cpp b/src/Processors/PingPongProcessor.cpp new file mode 100644 index 00000000000..e9d61386314 --- /dev/null +++ b/src/Processors/PingPongProcessor.cpp @@ -0,0 +1,198 @@ +#include + +namespace DB +{ + +/// Create list with `num_ports` of regular ports and 1 auxiliary port with empty header. +template requires std::is_same_v || std::is_same_v +static T createPortsWithSpecial(const Block & header, size_t num_ports) +{ + T res(num_ports, header); + res.emplace_back(Block()); + return res; +} + +PingPongProcessor::PingPongProcessor(const Block & header, size_t num_ports, Order order_) + : IProcessor(createPortsWithSpecial(header, num_ports), + createPortsWithSpecial(header, num_ports)) + , aux_in_port(inputs.back()) + , aux_out_port(outputs.back()) + , order(order_) +{ + assert(order == First || order == Second); + + port_pairs.resize(num_ports); + + auto input_it = inputs.begin(); + auto output_it = outputs.begin(); + for (size_t i = 0; i < num_ports; ++i) + { + port_pairs[i].input_port = &*input_it; + ++input_it; + + port_pairs[i].output_port = &*output_it; + ++output_it; + } +} + +void PingPongProcessor::finishPair(PortsPair & pair) +{ + if (!pair.is_finished) + { + pair.output_port->finish(); + pair.input_port->close(); + + pair.is_finished = true; + ++num_finished_pairs; + } +} + +bool PingPongProcessor::processPair(PortsPair & pair) +{ + if (pair.output_port->isFinished()) + { + finishPair(pair); + return false; + } + + if (pair.input_port->isFinished()) + { + finishPair(pair); + return false; + } + + if (!pair.output_port->canPush()) + { + pair.input_port->setNotNeeded(); + return false; + } + + pair.input_port->setNeeded(); + if (pair.input_port->hasData()) + { + Chunk chunk = pair.input_port->pull(true); + ready_to_send |= consume(chunk); + pair.output_port->push(std::move(chunk)); + } + + return true; +} + +bool PingPongProcessor::isPairsFinished() const +{ + return num_finished_pairs == port_pairs.size(); +} + +IProcessor::Status PingPongProcessor::processRegularPorts() +{ + if (isPairsFinished()) + return Status::Finished; + + bool need_data = false; + + for (auto & pair : port_pairs) + need_data = processPair(pair) || need_data; + + if (isPairsFinished()) + return Status::Finished; + + if (need_data) + return Status::NeedData; + + return Status::PortFull; +} + +bool PingPongProcessor::sendPing() +{ + if (aux_out_port.canPush()) + { + Chunk chunk(aux_out_port.getHeader().cloneEmpty().getColumns(), 0); + aux_out_port.push(std::move(chunk)); + is_send = true; + aux_out_port.finish(); + return true; + } + return false; +} + +bool PingPongProcessor::recievePing() +{ + if (aux_in_port.hasData()) + { + aux_in_port.pull(); + is_received = true; + aux_in_port.close(); + return true; + } + return false; +} + +bool PingPongProcessor::canSend() const +{ + return !is_send && (ready_to_send || isPairsFinished()); +} + +IProcessor::Status PingPongProcessor::prepare() +{ + if (!set_needed_once && !is_received && !aux_in_port.isFinished()) + { + set_needed_once = true; + aux_in_port.setNeeded(); + } + + if (order == First || is_send) + { + if (!is_received) + { + bool received = recievePing(); + if (!received) + { + return Status::NeedData; + } + } + } + + if (order == Second || is_received) + { + if (!is_send && canSend()) + { + bool sent = sendPing(); + if (!sent) + return Status::PortFull; + } + } + + auto status = processRegularPorts(); + if (status == Status::Finished) + { + if (order == First || is_send) + { + if (!is_received) + { + bool received = recievePing(); + if (!received) + { + return Status::NeedData; + } + } + } + + if (order == Second || is_received) + { + if (!is_send && canSend()) + { + bool sent = sendPing(); + if (!sent) + return Status::PortFull; + } + } + } + return status; +} + +std::pair PingPongProcessor::getAuxPorts() +{ + return std::make_pair(&aux_in_port, &aux_out_port); +} + +} diff --git a/src/Processors/PingPongProcessor.h b/src/Processors/PingPongProcessor.h new file mode 100644 index 00000000000..3dbe1178332 --- /dev/null +++ b/src/Processors/PingPongProcessor.h @@ -0,0 +1,105 @@ +#pragma once + +#include +#include +#include +#include + +namespace DB +{ + +/* + * Processor with N inputs and N outputs. Moves data from i-th input to i-th output as is. + * It has a pair of auxiliary ports to notify another instance by sending empty chunk after some condition holds. + * You should use this processor in pair of instances and connect auxiliary ports crosswise. + * + * ╭─┴───┴───┴───┴───┴─╮ ╭─┴───┴───┴───┴───┴─╮ + * │ ├─ aux ⟶│ │ + * │ PingPongProcessor │ │ PingPongProcessor │ + * │ │⟵ aux ─┤ │ + * ╰─┬───┬───┬───┬───┬─╯ ╰─┬───┬───┬───┬───┬─╯ + * + * One of the processors starts processing data, and another waits for notification. + * When `consume` returns true, the first stops processing, sends a ping to another and waits for notification. + * After that, the second one also processes data until `consume`, then send a notification back to the first one. + * After this roundtrip, processors bypass data from regular inputs to outputs. + */ +class PingPongProcessor : public IProcessor +{ +public: + enum class Order : uint8_t + { + /// Processor that starts processing data. + First, + /// Processor that waits for notification. + Second, + }; + + using enum Order; + + PingPongProcessor(const Block & header, size_t num_ports, Order order_); + + Status prepare() override; + + std::pair getAuxPorts(); + + /// Returns `true` when enough data consumed + virtual bool consume(const Chunk & chunk) = 0; + +protected: + struct PortsPair + { + InputPort * input_port = nullptr; + OutputPort * output_port = nullptr; + bool is_finished = false; + }; + + bool sendPing(); + bool recievePing(); + bool canSend() const; + + bool isPairsFinished() const; + bool processPair(PortsPair & pair); + void finishPair(PortsPair & pair); + Status processRegularPorts(); + + std::vector port_pairs; + size_t num_finished_pairs = 0; + + InputPort & aux_in_port; + OutputPort & aux_out_port; + + bool is_send = false; + bool is_received = false; + + bool ready_to_send = false; + + /// Used to set 'needed' flag once for auxiliary input at first `prepare` call. + bool set_needed_once = false; + + Order order; +}; + +/// Reads first N rows from two streams evenly. +class ReadHeadBalancedProcessor : public PingPongProcessor +{ +public: + ReadHeadBalancedProcessor(const Block & header, size_t num_ports, size_t size_to_wait_, Order order_) + : PingPongProcessor(header, num_ports, order_) , data_consumed(0) , size_to_wait(size_to_wait_) + { + } + + String getName() const override { return "ReadHeadBalancedProcessor"; } + + bool consume(const Chunk & chunk) override + { + data_consumed += chunk.getNumRows(); + return data_consumed > size_to_wait; + } + +private: + size_t data_consumed; + size_t size_to_wait; +}; + +} diff --git a/src/Processors/Port.cpp b/src/Processors/Port.cpp index 86431dbc6e6..79532dd4d6c 100644 --- a/src/Processors/Port.cpp +++ b/src/Processors/Port.cpp @@ -8,18 +8,18 @@ namespace ErrorCodes extern const int LOGICAL_ERROR; } -void connect(OutputPort & output, InputPort & input) +void connect(OutputPort & output, InputPort & input, bool reconnect) { - if (input.state) + if (!reconnect && input.state) throw Exception(ErrorCodes::LOGICAL_ERROR, "Port is already connected, (header: [{}])", input.header.dumpStructure()); - if (output.state) + if (!reconnect && output.state) throw Exception(ErrorCodes::LOGICAL_ERROR, "Port is already connected, (header: [{}])", output.header.dumpStructure()); - auto out_name = output.getProcessor().getName(); - auto in_name = input.getProcessor().getName(); + auto out_name = output.processor ? output.getProcessor().getName() : "null"; + auto in_name = input.processor ? input.getProcessor().getName() : "null"; - assertCompatibleHeader(output.getHeader(), input.getHeader(), fmt::format(" function connect between {} and {}", out_name, in_name)); + assertCompatibleHeader(output.getHeader(), input.getHeader(), fmt::format("function connect between {} and {}", out_name, in_name)); input.output_port = &output; output.input_port = &input; diff --git a/src/Processors/Port.h b/src/Processors/Port.h index e3fb0e3e342..9163402f600 100644 --- a/src/Processors/Port.h +++ b/src/Processors/Port.h @@ -25,7 +25,7 @@ namespace ErrorCodes class Port { - friend void connect(OutputPort &, InputPort &); + friend void connect(OutputPort &, InputPort &, bool); friend class IProcessor; public: @@ -267,7 +267,7 @@ protected: /// * You can pull only if port hasData(). class InputPort : public Port { - friend void connect(OutputPort &, InputPort &); + friend void connect(OutputPort &, InputPort &, bool); private: OutputPort * output_port = nullptr; @@ -390,7 +390,7 @@ public: /// * You can push only if port doesn't hasData(). class OutputPort : public Port { - friend void connect(OutputPort &, InputPort &); + friend void connect(OutputPort &, InputPort &, bool); private: InputPort * input_port = nullptr; @@ -483,6 +483,6 @@ using InputPorts = std::list; using OutputPorts = std::list; -void connect(OutputPort & output, InputPort & input); +void connect(OutputPort & output, InputPort & input, bool reconnect = false); } diff --git a/src/Processors/QueryPlan/AggregatingStep.cpp b/src/Processors/QueryPlan/AggregatingStep.cpp index f0374d2419b..5a5326091e6 100644 --- a/src/Processors/QueryPlan/AggregatingStep.cpp +++ b/src/Processors/QueryPlan/AggregatingStep.cpp @@ -182,6 +182,7 @@ void AggregatingStep::transformPipeline(QueryPipelineBuilder & pipeline, const B transform_params->params.min_free_disk_space, transform_params->params.compile_aggregate_expressions, transform_params->params.min_count_to_compile_aggregate_expression, + transform_params->params.max_block_size, /* only_merge */ false, transform_params->params.stats_collecting_params}; auto transform_params_for_set = std::make_shared(src_header, std::move(params_for_set), final); @@ -376,16 +377,15 @@ void AggregatingStep::transformPipeline(QueryPipelineBuilder & pipeline, const B }); /// We add the explicit resize here, but not in case of aggregating in order, since AIO don't use two-level hash tables and thus returns only buckets with bucket_number = -1. - pipeline.resize(should_produce_results_in_order_of_bucket_number ? 1 : pipeline.getNumStreams(), true /* force */); + pipeline.resize(should_produce_results_in_order_of_bucket_number ? 1 : params.max_threads, true /* force */); aggregating = collector.detachProcessors(0); } else { - pipeline.addSimpleTransform([&](const Block & header) - { - return std::make_shared(header, transform_params); - }); + pipeline.addSimpleTransform([&](const Block & header) { return std::make_shared(header, transform_params); }); + + pipeline.resize(should_produce_results_in_order_of_bucket_number ? 1 : params.max_threads, false /* force */); aggregating = collector.detachProcessors(0); } diff --git a/src/Processors/QueryPlan/CreateSetAndFilterOnTheFlyStep.cpp b/src/Processors/QueryPlan/CreateSetAndFilterOnTheFlyStep.cpp new file mode 100644 index 00000000000..e42642ceff8 --- /dev/null +++ b/src/Processors/QueryPlan/CreateSetAndFilterOnTheFlyStep.cpp @@ -0,0 +1,205 @@ +#include +#include + +#include +#include +#include +#include +#include +#include +#include + +namespace DB +{ + +namespace ErrorCodes +{ + extern const int LOGICAL_ERROR; +} + +static void connectAllInputs(OutputPortRawPtrs ports, InputPorts & inputs, size_t num_ports) +{ + auto input_it = inputs.begin(); + for (size_t i = 0; i < num_ports; ++i) + { + connect(*ports[i], *input_it); + input_it++; + } +} + +static ColumnsWithTypeAndName getColumnSubset(const Block & block, const Names & column_names) +{ + ColumnsWithTypeAndName result; + for (const auto & name : column_names) + result.emplace_back(block.getByName(name)); + return result; +} + +static ITransformingStep::Traits getTraits() +{ + return ITransformingStep::Traits + { + { + .preserves_distinct_columns = true, + .returns_single_stream = false, + .preserves_number_of_streams = true, + .preserves_sorting = true, + }, + { + .preserves_number_of_rows = false, + } + }; +} + +class CreateSetAndFilterOnTheFlyStep::CrosswiseConnection : public boost::noncopyable +{ +public: + using PortPair = std::pair; + + /// Remember ports passed on the first call and connect with ones from second call. + /// Thread-safe. + void connectPorts(PortPair rhs_ports, IProcessor * proc) + { + assert(!rhs_ports.first->isConnected() && !rhs_ports.second->isConnected()); + + std::lock_guard lock(mux); + if (input_port || output_port) + { + assert(input_port && output_port); + assert(!input_port->isConnected()); + connect(*rhs_ports.second, *input_port); + connect(*output_port, *rhs_ports.first, /* reconnect= */ true); + } + else + { + std::tie(input_port, output_port) = rhs_ports; + assert(input_port && output_port); + assert(!input_port->isConnected() && !output_port->isConnected()); + + dummy_input_port = std::make_unique(output_port->getHeader(), proc); + connect(*output_port, *dummy_input_port); + } + } + +private: + std::mutex mux; + InputPort * input_port = nullptr; + OutputPort * output_port = nullptr; + + /// Output ports should always be connected, and we can't add a step to the pipeline without them. + /// So, connect the port from the first processor to this dummy port and then reconnect to the second processor. + std::unique_ptr dummy_input_port; +}; + +CreateSetAndFilterOnTheFlyStep::CrosswiseConnectionPtr CreateSetAndFilterOnTheFlyStep::createCrossConnection() +{ + return std::make_shared(); +} + +CreateSetAndFilterOnTheFlyStep::CreateSetAndFilterOnTheFlyStep( + const DataStream & input_stream_, + const Names & column_names_, + size_t max_rows_in_set_, + CrosswiseConnectionPtr crosswise_connection_, + JoinTableSide position_) + : ITransformingStep(input_stream_, input_stream_.header, getTraits()) + , column_names(column_names_) + , max_rows_in_set(max_rows_in_set_) + , own_set(std::make_shared(SizeLimits(max_rows_in_set, 0, OverflowMode::BREAK), false, true)) + , filtering_set(nullptr) + , crosswise_connection(crosswise_connection_) + , position(position_) +{ + if (crosswise_connection == nullptr) + throw Exception(ErrorCodes::LOGICAL_ERROR, "Crosswise connection is not initialized"); + + if (input_streams.size() != 1) + throw Exception(ErrorCodes::LOGICAL_ERROR, "Step requires exactly one input stream, got {}", input_streams.size()); + + own_set->setHeader(getColumnSubset(input_streams[0].header, column_names)); +} + +void CreateSetAndFilterOnTheFlyStep::transformPipeline(QueryPipelineBuilder & pipeline, const BuildQueryPipelineSettings &) +{ + size_t num_streams = pipeline.getNumStreams(); + pipeline.addSimpleTransform([this, num_streams](const Block & header, QueryPipelineBuilder::StreamType stream_type) -> ProcessorPtr + { + if (stream_type != QueryPipelineBuilder::StreamType::Main) + return nullptr; + auto res = std::make_shared(header, column_names, num_streams, own_set); + res->setDescription(this->getStepDescription()); + return res; + }); + + Block input_header = pipeline.getHeader(); + auto pipeline_transform = [&input_header, this](OutputPortRawPtrs ports) + { + Processors result_transforms; + + size_t num_ports = ports.size(); + + /// Add balancing transform + auto idx = position == JoinTableSide::Left ? PingPongProcessor::First : PingPongProcessor::Second; + auto stream_balancer = std::make_shared(input_header, num_ports, max_rows_in_set, idx); + stream_balancer->setDescription(getStepDescription()); + + /// Regular inputs just bypass data for respective ports + connectAllInputs(ports, stream_balancer->getInputs(), num_ports); + + /// Connect auxiliary ports + crosswise_connection->connectPorts(stream_balancer->getAuxPorts(), stream_balancer.get()); + + if (!filtering_set) + { + LOG_DEBUG(log, "Skip filtering {} stream", position); + result_transforms.emplace_back(std::move(stream_balancer)); + return result_transforms; + } + + /// Add filtering transform, ports just connected respectively + auto & outputs = stream_balancer->getOutputs(); + auto output_it = outputs.begin(); + for (size_t i = 0; i < outputs.size() - 1; ++i) + { + auto & port = *output_it++; + auto transform = std::make_shared(port.getHeader(), column_names, filtering_set); + transform->setDescription(this->getStepDescription()); + connect(port, transform->getInputPort()); + result_transforms.emplace_back(std::move(transform)); + } + assert(output_it == std::prev(outputs.end())); + result_transforms.emplace_back(std::move(stream_balancer)); + + return result_transforms; + }; + + /// Auxiliary port stream_balancer can be connected later (by crosswise_connection). + /// So, use unsafe `transform` with `check_ports = false` to avoid assertions + pipeline.transform(std::move(pipeline_transform), /* check_ports= */ false); +} + +void CreateSetAndFilterOnTheFlyStep::describeActions(JSONBuilder::JSONMap & map) const +{ + map.add(getName(), true); +} + +void CreateSetAndFilterOnTheFlyStep::describeActions(FormatSettings & settings) const +{ + String prefix(settings.offset, ' '); + settings.out << prefix << getName(); + + settings.out << '\n'; +} + +void CreateSetAndFilterOnTheFlyStep::updateOutputStream() +{ + if (input_streams.size() != 1) + throw Exception(ErrorCodes::LOGICAL_ERROR, "{} requires exactly one input stream, got {}", getName(), input_streams.size()); + + own_set->setHeader(getColumnSubset(input_streams[0].header, column_names)); + + output_stream = input_streams[0]; +} + + +} diff --git a/src/Processors/QueryPlan/CreateSetAndFilterOnTheFlyStep.h b/src/Processors/QueryPlan/CreateSetAndFilterOnTheFlyStep.h new file mode 100644 index 00000000000..8c2eef00af0 --- /dev/null +++ b/src/Processors/QueryPlan/CreateSetAndFilterOnTheFlyStep.h @@ -0,0 +1,59 @@ +#pragma once +#include +#include +#include + + +namespace DB +{ + +/* + * Used to optimize JOIN when joining a small table over a large table. + * Currently applied only for the full sorting join. + * It tries to build a set for each stream. + * Once one stream is finished, it starts to filter another stream with this set. + */ +class CreateSetAndFilterOnTheFlyStep : public ITransformingStep +{ +public: + /// Two instances of step need some shared state to connect processors crosswise + class CrosswiseConnection; + using CrosswiseConnectionPtr = std::shared_ptr; + static CrosswiseConnectionPtr createCrossConnection(); + + CreateSetAndFilterOnTheFlyStep( + const DataStream & input_stream_, + const Names & column_names_, + size_t max_rows_in_set_, + CrosswiseConnectionPtr crosswise_connection_, + JoinTableSide position_); + + String getName() const override { return "CreateSetAndFilterOnTheFlyStep"; } + void transformPipeline(QueryPipelineBuilder & pipeline, const BuildQueryPipelineSettings & settings) override; + + void describeActions(JSONBuilder::JSONMap & map) const override; + void describeActions(FormatSettings & settings) const override; + + SetWithStatePtr getSet() const { return own_set; } + + /// Set for another stream. + void setFiltering(SetWithStatePtr filtering_set_) { filtering_set = filtering_set_; } + +private: + void updateOutputStream() override; + + Names column_names; + + size_t max_rows_in_set; + + SetWithStatePtr own_set; + SetWithStatePtr filtering_set; + + CrosswiseConnectionPtr crosswise_connection; + + JoinTableSide position; + + Poco::Logger * log = &Poco::Logger::get("CreateSetAndFilterOnTheFlyStep"); +}; + +} diff --git a/src/Processors/QueryPlan/JoinStep.cpp b/src/Processors/QueryPlan/JoinStep.cpp index 909933fbed2..6e212a53bc6 100644 --- a/src/Processors/QueryPlan/JoinStep.cpp +++ b/src/Processors/QueryPlan/JoinStep.cpp @@ -34,8 +34,12 @@ QueryPipelineBuilderPtr JoinStep::updatePipeline(QueryPipelineBuilders pipelines throw Exception(ErrorCodes::LOGICAL_ERROR, "JoinStep expect two input steps"); if (join->pipelineType() == JoinPipelineType::YShaped) - return QueryPipelineBuilder::joinPipelinesYShaped( + { + auto joined_pipeline = QueryPipelineBuilder::joinPipelinesYShaped( std::move(pipelines[0]), std::move(pipelines[1]), join, output_stream->header, max_block_size, &processors); + joined_pipeline->resize(max_streams); + return joined_pipeline; + } return QueryPipelineBuilder::joinPipelinesRightLeft( std::move(pipelines[0]), diff --git a/src/Processors/QueryPlan/Optimizations/filterPushDown.cpp b/src/Processors/QueryPlan/Optimizations/filterPushDown.cpp index 6c6c8954ea4..dbf389163be 100644 --- a/src/Processors/QueryPlan/Optimizations/filterPushDown.cpp +++ b/src/Processors/QueryPlan/Optimizations/filterPushDown.cpp @@ -8,6 +8,7 @@ #include #include #include +#include #include #include #include @@ -22,6 +23,7 @@ #include #include #include +#include namespace DB::ErrorCodes { @@ -134,10 +136,24 @@ tryAddNewFilterStep(QueryPlan::Node * parent_node, QueryPlan::Nodes & nodes, con static size_t tryAddNewFilterStep(QueryPlan::Node * parent_node, QueryPlan::Nodes & nodes, const Names & allowed_inputs, - bool can_remove_filter = true) + bool can_remove_filter = true, size_t child_idx = 0) { - if (auto split_filter = splitFilter(parent_node, allowed_inputs, 0)) - return tryAddNewFilterStep(parent_node, nodes, split_filter, can_remove_filter, 0); + if (auto split_filter = splitFilter(parent_node, allowed_inputs, child_idx)) + return tryAddNewFilterStep(parent_node, nodes, split_filter, can_remove_filter, child_idx); + return 0; +} + + +/// Push down filter through specified type of step +template +static size_t simplePushDownOverStep(QueryPlan::Node * parent_node, QueryPlan::Nodes & nodes, QueryPlanStepPtr & child) +{ + if (typeid_cast(child.get())) + { + Names allowed_inputs = child->getOutputStream().header.getNames(); + if (auto updated_steps = tryAddNewFilterStep(parent_node, nodes, allowed_inputs)) + return updated_steps; + } return 0; } @@ -234,12 +250,8 @@ size_t tryPushDownFilter(QueryPlan::Node * parent_node, QueryPlan::Nodes & nodes return updated_steps; } - if (auto * distinct = typeid_cast(child.get())) - { - Names allowed_inputs = distinct->getOutputStream().header.getNames(); - if (auto updated_steps = tryAddNewFilterStep(parent_node, nodes, allowed_inputs)) - return updated_steps; - } + if (auto updated_steps = simplePushDownOverStep(parent_node, nodes, child)) + return updated_steps; if (auto * join = typeid_cast(child.get())) { @@ -290,7 +302,7 @@ size_t tryPushDownFilter(QueryPlan::Node * parent_node, QueryPlan::Nodes & nodes const size_t updated_steps = tryAddNewFilterStep(parent_node, nodes, split_filter, can_remove_filter, child_idx); if (updated_steps > 0) { - LOG_DEBUG(&Poco::Logger::get("QueryPlanOptimizations"), "Pushed down filter to {} side of join", kind); + LOG_DEBUG(&Poco::Logger::get("QueryPlanOptimizations"), "Pushed down filter {} to the {} side of join", split_filter_column_name, kind); } return updated_steps; }; @@ -321,12 +333,11 @@ size_t tryPushDownFilter(QueryPlan::Node * parent_node, QueryPlan::Nodes & nodes // { // } - if (typeid_cast(child.get())) - { - Names allowed_inputs = child->getOutputStream().header.getNames(); - if (auto updated_steps = tryAddNewFilterStep(parent_node, nodes, allowed_inputs)) - return updated_steps; - } + if (auto updated_steps = simplePushDownOverStep(parent_node, nodes, child)) + return updated_steps; + + if (auto updated_steps = simplePushDownOverStep(parent_node, nodes, child)) + return updated_steps; if (auto * union_step = typeid_cast(child.get())) { diff --git a/src/Processors/QueryPlan/Optimizations/reuseStorageOrderingForWindowFunctions.cpp b/src/Processors/QueryPlan/Optimizations/reuseStorageOrderingForWindowFunctions.cpp index 4d3c268ab7e..401774b390e 100644 --- a/src/Processors/QueryPlan/Optimizations/reuseStorageOrderingForWindowFunctions.cpp +++ b/src/Processors/QueryPlan/Optimizations/reuseStorageOrderingForWindowFunctions.cpp @@ -29,7 +29,7 @@ size_t tryReuseStorageOrderingForWindowFunctions(QueryPlan::Node * parent_node, { /// Find the following sequence of steps, add InputOrderInfo and apply prefix sort description to /// SortingStep: - /// WindowStep <- SortingStep <- [Expression] <- [SettingQuotaAndLimits] <- ReadFromMergeTree + /// WindowStep <- SortingStep <- [Expression] <- ReadFromMergeTree auto * window_node = parent_node; auto * window = typeid_cast(window_node->step.get()); @@ -104,7 +104,7 @@ size_t tryReuseStorageOrderingForWindowFunctions(QueryPlan::Node * parent_node, if (order_info) { read_from_merge_tree->setQueryInfoInputOrderInfo(order_info); - sorting->convertToFinishSorting(order_info->order_key_prefix_descr); + sorting->convertToFinishSorting(order_info->sort_description_for_merging); } return 0; diff --git a/src/Processors/QueryPlan/ReadFromMergeTree.cpp b/src/Processors/QueryPlan/ReadFromMergeTree.cpp index 67ea36aeeef..1f6c6ee2a3f 100644 --- a/src/Processors/QueryPlan/ReadFromMergeTree.cpp +++ b/src/Processors/QueryPlan/ReadFromMergeTree.cpp @@ -548,9 +548,7 @@ Pipe ReadFromMergeTree::spreadMarkRangesAmongStreamsWithOrder( if (need_preliminary_merge) { - size_t fixed_prefix_size = input_order_info->order_key_fixed_prefix_descr.size(); - size_t prefix_size = fixed_prefix_size + input_order_info->order_key_prefix_descr.size(); - + size_t prefix_size = input_order_info->used_prefix_of_sorting_key_size; auto order_key_prefix_ast = metadata_for_reading->getSortingKey().expression_list_ast->clone(); order_key_prefix_ast->children.resize(prefix_size); diff --git a/src/Processors/ResizeProcessor.h b/src/Processors/ResizeProcessor.h index f9c188e041a..364d1b4c883 100644 --- a/src/Processors/ResizeProcessor.h +++ b/src/Processors/ResizeProcessor.h @@ -85,6 +85,13 @@ public: { } + StrictResizeProcessor(InputPorts inputs_, OutputPorts outputs_) + : IProcessor(inputs_, outputs_) + , current_input(inputs.begin()) + , current_output(outputs.begin()) + { + } + String getName() const override { return "StrictResize"; } Status prepare(const PortNumbers &, const PortNumbers &) override; diff --git a/src/Processors/Sources/SQLiteSource.cpp b/src/Processors/Sources/SQLiteSource.cpp index 814480b63e3..60d39966659 100644 --- a/src/Processors/Sources/SQLiteSource.cpp +++ b/src/Processors/Sources/SQLiteSource.cpp @@ -39,7 +39,7 @@ SQLiteSource::SQLiteSource( if (status != SQLITE_OK) throw Exception(ErrorCodes::SQLITE_ENGINE_ERROR, - "Cannot prepate sqlite statement. Status: {}. Message: {}", + "Cannot prepare sqlite statement. Status: {}. Message: {}", status, sqlite3_errstr(status)); compiled_statement = std::unique_ptr(compiled_stmt, StatementDeleter()); diff --git a/src/Processors/TTL/TTLAggregationAlgorithm.cpp b/src/Processors/TTL/TTLAggregationAlgorithm.cpp index 0d160b8d32d..6a813a770cf 100644 --- a/src/Processors/TTL/TTLAggregationAlgorithm.cpp +++ b/src/Processors/TTL/TTLAggregationAlgorithm.cpp @@ -38,7 +38,8 @@ TTLAggregationAlgorithm::TTLAggregationAlgorithm( settings.max_threads, settings.min_free_disk_space_for_temporary_data, settings.compile_aggregate_expressions, - settings.min_count_to_compile_aggregate_expression); + settings.min_count_to_compile_aggregate_expression, + settings.max_block_size); aggregator = std::make_unique(header, params); diff --git a/src/Processors/Transforms/AggregatingInOrderTransform.cpp b/src/Processors/Transforms/AggregatingInOrderTransform.cpp index ce50ae5eeee..c2de0c3a23a 100644 --- a/src/Processors/Transforms/AggregatingInOrderTransform.cpp +++ b/src/Processors/Transforms/AggregatingInOrderTransform.cpp @@ -41,13 +41,13 @@ AggregatingInOrderTransform::AggregatingInOrderTransform( /// We won't finalize states in order to merge same states (generated due to multi-thread execution) in AggregatingSortedTransform res_header = params->getCustomHeader(/* final_= */ false); - for (size_t i = 0; i < group_by_info->order_key_prefix_descr.size(); ++i) + for (size_t i = 0; i < group_by_info->sort_description_for_merging.size(); ++i) { const auto & column_description = group_by_description_[i]; group_by_description.emplace_back(column_description, res_header.getPositionByName(column_description.column_name)); } - if (group_by_info->order_key_prefix_descr.size() < group_by_description_.size()) + if (group_by_info->sort_description_for_merging.size() < group_by_description_.size()) { group_by_key = true; /// group_by_description may contains duplicates, so we use keys_size from Aggregator::params @@ -182,7 +182,8 @@ void AggregatingInOrderTransform::consume(Chunk chunk) if (cur_block_size >= max_block_size || cur_block_bytes + current_memory_usage >= max_block_bytes) { if (group_by_key) - group_by_block = params->aggregator.prepareBlockAndFillSingleLevel(variants, /* final= */ false); + group_by_block + = params->aggregator.prepareBlockAndFillSingleLevel(variants, /* final= */ false); cur_block_bytes += current_memory_usage; finalizeCurrentChunk(std::move(chunk), key_end); return; @@ -293,7 +294,8 @@ void AggregatingInOrderTransform::generate() if (cur_block_size && is_consume_finished) { if (group_by_key) - group_by_block = params->aggregator.prepareBlockAndFillSingleLevel(variants, /* final= */ false); + group_by_block + = params->aggregator.prepareBlockAndFillSingleLevel(variants, /* final= */ false); else params->aggregator.addSingleKeyToAggregateColumns(variants, res_aggregate_columns); variants.invalidate(); diff --git a/src/Processors/Transforms/AggregatingTransform.cpp b/src/Processors/Transforms/AggregatingTransform.cpp index 7f5896f5e97..4e55081ca48 100644 --- a/src/Processors/Transforms/AggregatingTransform.cpp +++ b/src/Processors/Transforms/AggregatingTransform.cpp @@ -203,7 +203,7 @@ public: { auto & output = outputs.front(); - if (finished && !has_input) + if (finished && single_level_chunks.empty()) { output.finish(); return Status::Finished; @@ -230,7 +230,7 @@ public: if (!processors.empty()) return Status::ExpandPipeline; - if (has_input) + if (!single_level_chunks.empty()) return preparePushToOutput(); /// Single level case. @@ -244,11 +244,14 @@ public: private: IProcessor::Status preparePushToOutput() { - auto & output = outputs.front(); - output.push(std::move(current_chunk)); - has_input = false; + if (single_level_chunks.empty()) + throw Exception(ErrorCodes::LOGICAL_ERROR, "Some ready chunks expected"); - if (finished) + auto & output = outputs.front(); + output.push(std::move(single_level_chunks.back())); + single_level_chunks.pop_back(); + + if (finished && single_level_chunks.empty()) { output.finish(); return Status::Finished; @@ -268,17 +271,17 @@ private: { auto chunk = input.pull(); auto bucket = getInfoFromChunk(chunk)->bucket_num; - chunks[bucket] = std::move(chunk); + two_level_chunks[bucket] = std::move(chunk); } } if (!shared_data->is_bucket_processed[current_bucket_num]) return Status::NeedData; - if (!chunks[current_bucket_num]) + if (!two_level_chunks[current_bucket_num]) return Status::NeedData; - output.push(std::move(chunks[current_bucket_num])); + output.push(std::move(two_level_chunks[current_bucket_num])); ++current_bucket_num; if (current_bucket_num == NUM_BUCKETS) @@ -298,27 +301,16 @@ private: size_t num_threads; bool is_initialized = false; - bool has_input = false; bool finished = false; - Chunk current_chunk; + Chunks single_level_chunks; UInt32 current_bucket_num = 0; static constexpr Int32 NUM_BUCKETS = 256; - std::array chunks; + std::array two_level_chunks; Processors processors; - void setCurrentChunk(Chunk chunk) - { - if (has_input) - throw Exception("Current chunk was already set in " - "ConvertingAggregatedToChunksTransform.", ErrorCodes::LOGICAL_ERROR); - - has_input = true; - current_chunk = std::move(chunk); - } - void initialize() { is_initialized = true; @@ -339,7 +331,7 @@ private: auto block = params->aggregator.prepareBlockAndFillWithoutKey( *first, params->final, first->type != AggregatedDataVariants::Type::without_key); - setCurrentChunk(convertToChunk(block)); + single_level_chunks.emplace_back(convertToChunk(block)); } } @@ -364,9 +356,10 @@ private: else throw Exception("Unknown aggregated data variant.", ErrorCodes::UNKNOWN_AGGREGATED_DATA_VARIANT); - auto block = params->aggregator.prepareBlockAndFillSingleLevel(*first, params->final); + auto blocks = params->aggregator.prepareBlockAndFillSingleLevel(*first, params->final); + for (auto & block : blocks) + single_level_chunks.emplace_back(convertToChunk(block)); - setCurrentChunk(convertToChunk(block)); finished = true; } diff --git a/src/Processors/Transforms/CreateSetAndFilterOnTheFlyTransform.cpp b/src/Processors/Transforms/CreateSetAndFilterOnTheFlyTransform.cpp new file mode 100644 index 00000000000..4278eb8e8b2 --- /dev/null +++ b/src/Processors/Transforms/CreateSetAndFilterOnTheFlyTransform.cpp @@ -0,0 +1,195 @@ +#include + +#include +#include + +#include +#include +#include +#include +#include +#include +#include + +namespace DB +{ + +namespace ErrorCodes +{ + extern const int LOGICAL_ERROR; +} + +namespace +{ + +std::vector getColumnIndices(const Block & block, const Names & column_names) +{ + std::vector indices; + for (const auto & name : column_names) + indices.push_back(block.getPositionByName(name)); + return indices; +} + +Columns getColumnsByIndices(const Chunk & chunk, const std::vector & indices) +{ + Columns columns; + const Columns & all_cols = chunk.getColumns(); + for (const auto & index : indices) + columns.push_back(all_cols.at(index)); + return columns; +} + +ColumnsWithTypeAndName getColumnsByIndices(const Block & sample_block, const Chunk & chunk, const std::vector & indices) +{ + Block block = sample_block.cloneEmpty(); + block.setColumns(getColumnsByIndices(chunk, indices)); + return block.getColumnsWithTypeAndName(); +} + +} + +CreatingSetsOnTheFlyTransform::CreatingSetsOnTheFlyTransform( + const Block & header_, const Names & column_names_, size_t num_streams_, SetWithStatePtr set_) + : ISimpleTransform(header_, header_, true) + , column_names(column_names_) + , key_column_indices(getColumnIndices(inputs.front().getHeader(), column_names)) + , num_streams(num_streams_) + , set(set_) +{ +} + +IProcessor::Status CreatingSetsOnTheFlyTransform::prepare() +{ + IProcessor::Status status = ISimpleTransform::prepare(); + + if (!set || status != Status::Finished) + /// Nothing to do with set + return status; + + /// Finalize set + if (set->state == SetWithState::State::Creating) + { + if (input.isFinished()) + { + set->finished_count++; + if (set->finished_count != num_streams) + /// Not all instances of processor are finished + return status; + + set->finishInsert(); + set->state = SetWithState::State::Finished; + LOG_DEBUG(log, "{}: finish building set for [{}] with {} rows, set size is {}", + getDescription(), fmt::join(column_names, ", "), set->getTotalRowCount(), + formatReadableSizeWithBinarySuffix(set->getTotalByteCount())); + set.reset(); + } + else + { + /// Should not happen because processor inserted before join that reads all the data + throw Exception(ErrorCodes::LOGICAL_ERROR, "Processor finished, but not all input was read"); + } + } + + return status; +} + +void CreatingSetsOnTheFlyTransform::transform(Chunk & chunk) +{ + if (!set || set->state != SetWithState::State::Creating) + { + /// If set building suspended by another processor, release pointer + if (set != nullptr) + set.reset(); + return; + } + + if (chunk.getNumRows()) + { + Columns key_columns = getColumnsByIndices(chunk, key_column_indices); + bool limit_exceeded = !set->insertFromBlock(key_columns); + if (limit_exceeded) + { + auto prev_state = set->state.exchange(SetWithState::State::Suspended); + /// Print log only after first state switch + if (prev_state == SetWithState::State::Creating) + { + LOG_DEBUG(log, "{}: set limit exceeded, give up building set, after reading {} rows and using {}", + getDescription(), set->getTotalRowCount(), formatReadableSizeWithBinarySuffix(set->getTotalByteCount())); + } + /// Probaply we need to clear set here, because it's unneeded anymore + /// But now `Set` doesn't have such method, so reset pointer in all processors and then it should be freed + set.reset(); + } + } +} + +FilterBySetOnTheFlyTransform::FilterBySetOnTheFlyTransform(const Block & header_, const Names & column_names_, SetWithStatePtr set_) + : ISimpleTransform(header_, header_, true) + , column_names(column_names_) + , key_column_indices(getColumnIndices(inputs.front().getHeader(), column_names)) + , set(set_) +{ + const auto & header = inputs.front().getHeader(); + for (size_t idx : key_column_indices) + key_sample_block.insert(header.getByPosition(idx)); +} + +IProcessor::Status FilterBySetOnTheFlyTransform::prepare() +{ + auto status = ISimpleTransform::prepare(); + + if (set && set->state == SetWithState::State::Suspended) + set.reset(); + + if (status == Status::Finished) + { + bool has_filter = set && set->state == SetWithState::State::Finished; + if (has_filter) + { + LOG_DEBUG(log, "Finished {} by [{}]: consumed {} rows in total, {} rows bypassed, result {} rows, {:.2f}% filtered", + Poco::toLower(getDescription()), fmt::join(column_names, ", "), + stat.consumed_rows, stat.consumed_rows_before_set, stat.result_rows, + 100 - 100.0 * stat.result_rows / stat.consumed_rows); + } + else + { + LOG_DEBUG(log, "Finished {}: bypass {} rows", Poco::toLower(getDescription()), stat.consumed_rows); + } + + /// Release set to free memory + set = nullptr; + } + return status; +} + +void FilterBySetOnTheFlyTransform::transform(Chunk & chunk) +{ + stat.consumed_rows += chunk.getNumRows(); + stat.result_rows += chunk.getNumRows(); + + bool can_filter = set && set->state == SetWithState::State::Finished; + if (!can_filter) + stat.consumed_rows_before_set += chunk.getNumRows(); + + if (can_filter && chunk.getNumRows()) + { + auto key_columns = getColumnsByIndices(key_sample_block, chunk, key_column_indices); + ColumnPtr mask_col = set->execute(key_columns, false); + const auto & mask = assert_cast(mask_col.get())->getData(); + + stat.result_rows -= chunk.getNumRows(); + + Columns columns = chunk.detachColumns(); + size_t result_num_rows = 0; + for (auto & col : columns) + { + col = col->filter(mask, /* negative */ false); + result_num_rows = col->size(); + } + stat.result_rows += result_num_rows; + + chunk.setColumns(std::move(columns), result_num_rows); + } +} + +} diff --git a/src/Processors/Transforms/CreateSetAndFilterOnTheFlyTransform.h b/src/Processors/Transforms/CreateSetAndFilterOnTheFlyTransform.h new file mode 100644 index 00000000000..d214a310a8c --- /dev/null +++ b/src/Processors/Transforms/CreateSetAndFilterOnTheFlyTransform.h @@ -0,0 +1,114 @@ +#pragma once + +#include +#include +#include +#include +#include +#include + +namespace DB +{ + +struct SetWithState : public Set +{ + using Set::Set; + + /// Flow: Creating -> Finished or Suspended + enum class State + { + /// Set is not yet created, + /// Creating processor continues to build set. + /// Filtering bypasses data. + Creating, + + /// Set is finished. + /// Creating processor is finished. + /// Filtering filter stream with this set. + Finished, + + /// Set building is canceled (due to limit exceeded). + /// Creating and filtering processors bypass data. + Suspended, + }; + + std::atomic state = State::Creating; + + /// Track number of processors that are currently working on this set. + /// Last one finalizes set. + std::atomic_size_t finished_count = 0; +}; + +using SetWithStatePtr = std::shared_ptr; + +/* + * Create a set on the fly for incoming stream. + * The set is created from the key columns of the input block. + * Data is not changed and returned as is. + * Can be executed in parallel, but blocks on operations with set. + */ +class CreatingSetsOnTheFlyTransform : public ISimpleTransform +{ +public: + CreatingSetsOnTheFlyTransform(const Block & header_, const Names & column_names_, size_t num_streams_, SetWithStatePtr set_); + + String getName() const override { return "CreatingSetsOnTheFlyTransform"; } + + Status prepare() override; + + void transform(Chunk & chunk) override; + +private: + Names column_names; + std::vector key_column_indices; + + size_t num_streams; + + /// Set to fill + SetWithStatePtr set; + + Poco::Logger * log = &Poco::Logger::get("CreatingSetsOnTheFlyTransform"); +}; + +/* + * Filter the input chunk by the set. + * When set building is not completed, just return the source data. + */ +class FilterBySetOnTheFlyTransform : public ISimpleTransform +{ +public: + FilterBySetOnTheFlyTransform(const Block & header_, const Names & column_names_, SetWithStatePtr set_); + + String getName() const override { return "FilterBySetOnTheFlyTransform"; } + + Status prepare() override; + + void transform(Chunk & chunk) override; + +private: + /// Set::execute requires ColumnsWithTypesAndNames, so we need to convert Chunk to that format + Block key_sample_block; + + Names column_names; + std::vector key_column_indices; + + /// Filter by this set when it's created + SetWithStatePtr set; + + /// Statistics to log + struct Stat + { + /// Total number of rows + size_t consumed_rows = 0; + + /// Number of bypassed rows (processed before set is created) + size_t consumed_rows_before_set = 0; + + /// Number of rows that passed the filter + size_t result_rows = 0; + } stat; + + Poco::Logger * log = &Poco::Logger::get("FilterBySetOnTheFlyTransform"); +}; + +} diff --git a/src/Processors/Transforms/IntersectOrExceptTransform.cpp b/src/Processors/Transforms/IntersectOrExceptTransform.cpp index 1ac82e99cf2..31a3e304505 100644 --- a/src/Processors/Transforms/IntersectOrExceptTransform.cpp +++ b/src/Processors/Transforms/IntersectOrExceptTransform.cpp @@ -111,7 +111,10 @@ size_t IntersectOrExceptTransform::buildFilter( for (size_t i = 0; i < rows; ++i) { auto find_result = state.findKey(method.data, i, variants.string_pool); - filter[i] = current_operator == ASTSelectIntersectExceptQuery::Operator::EXCEPT ? !find_result.isFound() : find_result.isFound(); + filter[i] = (current_operator == ASTSelectIntersectExceptQuery::Operator::EXCEPT_ALL + || current_operator == ASTSelectIntersectExceptQuery::Operator::EXCEPT_DISTINCT) + ? !find_result.isFound() + : find_result.isFound(); if (filter[i]) ++new_rows_num; } diff --git a/src/Processors/Transforms/MergeJoinTransform.cpp b/src/Processors/Transforms/MergeJoinTransform.cpp index 6f842bec939..315fc4810ba 100644 --- a/src/Processors/Transforms/MergeJoinTransform.cpp +++ b/src/Processors/Transforms/MergeJoinTransform.cpp @@ -513,7 +513,7 @@ MergeJoinAlgorithm::Status MergeJoinAlgorithm::allJoin(JoinKind kind) Columns lcols; if (!left_to_right_key_remap.empty()) { - /// If we have remapped columns, then we need to get values from right columns insead of defaults + /// If we have remapped columns, then we need to get values from right columns instead of defaults const auto & indices = idx_map[0]; const auto & left_src = cursors[0]->getCurrent().getColumns(); diff --git a/src/QueryPipeline/Pipe.cpp b/src/QueryPipeline/Pipe.cpp index ae342abeea5..291739079a2 100644 --- a/src/QueryPipeline/Pipe.cpp +++ b/src/QueryPipeline/Pipe.cpp @@ -770,7 +770,7 @@ void Pipe::setSinks(const Pipe::ProcessorGetterWithStreamKind & getter) header.clear(); } -void Pipe::transform(const Transformer & transformer) +void Pipe::transform(const Transformer & transformer, bool check_ports) { if (output_ports.empty()) throw Exception("Cannot transform empty Pipe", ErrorCodes::LOGICAL_ERROR); @@ -784,6 +784,9 @@ void Pipe::transform(const Transformer & transformer) for (const auto & port : output_ports) { + if (!check_ports) + break; + if (!port->isConnected()) throw Exception( ErrorCodes::LOGICAL_ERROR, @@ -799,6 +802,9 @@ void Pipe::transform(const Transformer & transformer) { for (const auto & port : processor->getInputs()) { + if (!check_ports) + break; + if (!port.isConnected()) throw Exception( ErrorCodes::LOGICAL_ERROR, @@ -806,7 +812,7 @@ void Pipe::transform(const Transformer & transformer) processor->getName()); const auto * connected_processor = &port.getOutputPort().getProcessor(); - if (!set.contains(connected_processor)) + if (check_ports && !set.contains(connected_processor)) throw Exception( ErrorCodes::LOGICAL_ERROR, "Transformation of Pipe is not valid because processor {} has input port which is connected with unknown processor {}", @@ -823,7 +829,7 @@ void Pipe::transform(const Transformer & transformer) } const auto * connected_processor = &port.getInputPort().getProcessor(); - if (!set.contains(connected_processor)) + if (check_ports && !set.contains(connected_processor)) throw Exception( ErrorCodes::LOGICAL_ERROR, "Transformation of Pipe is not valid because processor {} has output port which is connected with unknown processor {}", diff --git a/src/QueryPipeline/Pipe.h b/src/QueryPipeline/Pipe.h index 52059f4ad19..79d19a18193 100644 --- a/src/QueryPipeline/Pipe.h +++ b/src/QueryPipeline/Pipe.h @@ -85,13 +85,13 @@ public: /// Add chain to every output port. void addChains(std::vector chains); - /// Changes the number of output ports if needed. Adds ResizeTransform. + /// Changes the number of output ports if needed. Adds (Strict)ResizeProcessor. void resize(size_t num_streams, bool force = false, bool strict = false); using Transformer = std::function; /// Transform Pipe in general way. - void transform(const Transformer & transformer); + void transform(const Transformer & transformer, bool check_ports = true); /// Unite several pipes together. They should have same header. static Pipe unitePipes(Pipes pipes); diff --git a/src/QueryPipeline/QueryPipelineBuilder.cpp b/src/QueryPipeline/QueryPipelineBuilder.cpp index 82907d883bc..877c19a493e 100644 --- a/src/QueryPipeline/QueryPipelineBuilder.cpp +++ b/src/QueryPipeline/QueryPipelineBuilder.cpp @@ -159,10 +159,10 @@ void QueryPipelineBuilder::addChain(Chain chain) pipe.addChains(std::move(chains)); } -void QueryPipelineBuilder::transform(const Transformer & transformer) +void QueryPipelineBuilder::transform(const Transformer & transformer, bool check_ports) { checkInitializedAndNotCompleted(); - pipe.transform(transformer); + pipe.transform(transformer, check_ports); } void QueryPipelineBuilder::setSinks(const Pipe::ProcessorGetterWithStreamKind & getter) @@ -348,8 +348,7 @@ std::unique_ptr QueryPipelineBuilder::joinPipelinesYShaped left->pipe.dropExtremes(); right->pipe.dropExtremes(); - - if (left->pipe.output_ports.size() != 1 || right->pipe.output_ports.size() != 1) + if (left->getNumStreams() != 1 || right->getNumStreams() != 1) throw Exception("Join is supported only for pipelines with one output port", ErrorCodes::LOGICAL_ERROR); if (left->hasTotals() || right->hasTotals()) @@ -359,8 +358,7 @@ std::unique_ptr QueryPipelineBuilder::joinPipelinesYShaped auto joining = std::make_shared(join, inputs, out_header, max_block_size); - auto result = mergePipelines(std::move(left), std::move(right), std::move(joining), collected_processors); - return result; + return mergePipelines(std::move(left), std::move(right), std::move(joining), collected_processors); } std::unique_ptr QueryPipelineBuilder::joinPipelinesRightLeft( diff --git a/src/QueryPipeline/QueryPipelineBuilder.h b/src/QueryPipeline/QueryPipelineBuilder.h index 100a2e07341..4edae83fe86 100644 --- a/src/QueryPipeline/QueryPipelineBuilder.h +++ b/src/QueryPipeline/QueryPipelineBuilder.h @@ -69,7 +69,7 @@ public: using Transformer = std::function; /// Transform pipeline in general way. - void transform(const Transformer & transformer); + void transform(const Transformer & transformer, bool check_ports = true); /// Add TotalsHavingTransform. Resize pipeline to single input. Adds totals port. void addTotalsHavingTransform(ProcessorPtr transform); diff --git a/src/QueryPipeline/RemoteInserter.cpp b/src/QueryPipeline/RemoteInserter.cpp index cd0485ec8e3..b8a878b56c3 100644 --- a/src/QueryPipeline/RemoteInserter.cpp +++ b/src/QueryPipeline/RemoteInserter.cpp @@ -31,22 +31,6 @@ RemoteInserter::RemoteInserter( { ClientInfo modified_client_info = client_info_; modified_client_info.query_kind = ClientInfo::QueryKind::SECONDARY_QUERY; - if (CurrentThread::isInitialized()) - { - auto& thread_trace_context = CurrentThread::get().thread_trace_context; - - if (thread_trace_context.trace_id != UUID()) - { - // overwrite the trace context only if current thread trace context is available - modified_client_info.client_trace_context = thread_trace_context; - } - else - { - // if the trace on the thread local is not enabled(for example running in a background thread) - // we should not clear the trace context on the client info because the client info may hold trace context - // and this trace context should be propagated to the remote server so that the tracing of distributed table insert is complete. - } - } Settings settings = settings_; /// With current protocol it is impossible to avoid deadlock in case of send_logs_level!=none. diff --git a/src/QueryPipeline/RemoteQueryExecutor.cpp b/src/QueryPipeline/RemoteQueryExecutor.cpp index 44e844fbe40..e42b0141a27 100644 --- a/src/QueryPipeline/RemoteQueryExecutor.cpp +++ b/src/QueryPipeline/RemoteQueryExecutor.cpp @@ -242,10 +242,6 @@ void RemoteQueryExecutor::sendQuery(ClientInfo::QueryKind query_kind) auto timeouts = ConnectionTimeouts::getTCPTimeoutsWithFailover(settings); ClientInfo modified_client_info = context->getClientInfo(); modified_client_info.query_kind = query_kind; - if (CurrentThread::isInitialized()) - { - modified_client_info.client_trace_context = CurrentThread::get().thread_trace_context; - } { std::lock_guard lock(duplicated_part_uuids_mutex); diff --git a/src/Server/GRPCServer.cpp b/src/Server/GRPCServer.cpp index 165dc7e1344..f9b3041ad40 100644 --- a/src/Server/GRPCServer.cpp +++ b/src/Server/GRPCServer.cpp @@ -662,6 +662,7 @@ namespace std::optional session; ContextMutablePtr query_context; std::optional query_scope; + OpenTelemetry::TracingContextHolderPtr thread_trace_context; String query_text; ASTPtr ast; ASTInsertQuery * insert_query = nullptr; @@ -840,6 +841,12 @@ namespace query_context->setCurrentQueryId(query_info.query_id()); query_scope.emplace(query_context); + /// Set up tracing context for this query on current thread + thread_trace_context = std::make_unique("GRPCServer", + query_context->getClientInfo().client_trace_context, + query_context->getSettingsRef(), + query_context->getOpenTelemetrySpanLog()); + /// Prepare for sending exceptions and logs. const Settings & settings = query_context->getSettingsRef(); send_exception_with_stacktrace = settings.calculate_text_stack_trace; @@ -1359,6 +1366,7 @@ namespace io = {}; query_scope.reset(); query_context.reset(); + thread_trace_context.reset(); session.reset(); } diff --git a/src/Server/HTTPHandler.cpp b/src/Server/HTTPHandler.cpp index 5b8e17eb279..442233ab408 100644 --- a/src/Server/HTTPHandler.cpp +++ b/src/Server/HTTPHandler.cpp @@ -534,19 +534,7 @@ void HTTPHandler::processQuery( session->makeSessionContext(session_id, session_timeout, session_check == "1"); } - // Parse the OpenTelemetry traceparent header. - ClientInfo client_info = session->getClientInfo(); - if (request.has("traceparent")) - { - std::string opentelemetry_traceparent = request.get("traceparent"); - std::string error; - if (!client_info.client_trace_context.parseTraceparentHeader(opentelemetry_traceparent, error)) - { - LOG_DEBUG(log, "Failed to parse OpenTelemetry traceparent header '{}': {}", opentelemetry_traceparent, error); - } - client_info.client_trace_context.tracestate = request.get("tracestate", ""); - } - + auto client_info = session->getClientInfo(); auto context = session->makeQueryContext(std::move(client_info)); /// The client can pass a HTTP header indicating supported compression method (gzip or deflate). @@ -945,6 +933,13 @@ void HTTPHandler::handleRequest(HTTPServerRequest & request, HTTPServerResponse /// In case of exception, send stack trace to client. bool with_stacktrace = false; + OpenTelemetry::TracingContextHolderPtr thread_trace_context; + SCOPE_EXIT({ + // make sure the response status is recorded + if (thread_trace_context) + thread_trace_context->root_span.addAttribute("clickhouse.http_status", response.getStatus()); + }); + try { if (request.getMethod() == HTTPServerRequest::HTTP_OPTIONS) @@ -952,6 +947,28 @@ void HTTPHandler::handleRequest(HTTPServerRequest & request, HTTPServerResponse processOptionsRequest(response, server.config()); return; } + + // Parse the OpenTelemetry traceparent header. + ClientInfo& client_info = session->getClientInfo(); + if (request.has("traceparent")) + { + std::string opentelemetry_traceparent = request.get("traceparent"); + std::string error; + if (!client_info.client_trace_context.parseTraceparentHeader(opentelemetry_traceparent, error)) + { + LOG_DEBUG(log, "Failed to parse OpenTelemetry traceparent header '{}': {}", opentelemetry_traceparent, error); + } + client_info.client_trace_context.tracestate = request.get("tracestate", ""); + } + + // Setup tracing context for this thread + auto context = session->sessionOrGlobalContext(); + thread_trace_context = std::make_unique("HTTPHandler", + client_info.client_trace_context, + context->getSettingsRef(), + context->getOpenTelemetrySpanLog()); + thread_trace_context->root_span.addAttribute("clickhouse.uri", request.getURI()); + response.setContentType("text/plain; charset=UTF-8"); response.set("X-ClickHouse-Server-Display-Name", server_display_name); /// For keep-alive to work. @@ -987,6 +1004,9 @@ void HTTPHandler::handleRequest(HTTPServerRequest & request, HTTPServerResponse /// cannot write in it anymore. So, just log this exception. if (used_output.isFinalized()) { + if (thread_trace_context) + thread_trace_context->root_span.addAttribute("clickhouse.exception", "Cannot flush data to client"); + tryLogCurrentException(log, "Cannot flush data to client"); return; } @@ -1000,6 +1020,9 @@ void HTTPHandler::handleRequest(HTTPServerRequest & request, HTTPServerResponse int exception_code = getCurrentExceptionCode(); trySendExceptionToClient(exception_message, exception_code, request, response, used_output); + + if (thread_trace_context) + thread_trace_context->root_span.addAttribute("clickhouse.exception_code", exception_code); } used_output.finalize(); diff --git a/src/Server/TCPHandler.cpp b/src/Server/TCPHandler.cpp index 1fc88168b35..550ae1bff31 100644 --- a/src/Server/TCPHandler.cpp +++ b/src/Server/TCPHandler.cpp @@ -219,6 +219,7 @@ void TCPHandler::runImpl() /// Initialized later. std::optional query_scope; + OpenTelemetry::TracingContextHolderPtr thread_trace_context; /** An exception during the execution of request (it must be sent over the network to the client). * The client will be able to accept it, if it did not happen while sending another packet and the client has not disconnected yet. @@ -244,6 +245,12 @@ void TCPHandler::runImpl() if (state.empty() && state.part_uuids_to_ignore && !receivePacket()) continue; + /// Set up tracing context for this query on current thread + thread_trace_context = std::make_unique("TCPHandler", + query_context->getClientInfo().client_trace_context, + query_context->getSettingsRef(), + query_context->getOpenTelemetrySpanLog()); + query_scope.emplace(query_context); /// If query received, then settings in query_context has been updated. @@ -419,6 +426,7 @@ void TCPHandler::runImpl() /// (i.e. deallocations from the Aggregator with two-level aggregation) state.reset(); query_scope.reset(); + thread_trace_context.reset(); } catch (const Exception & e) { @@ -484,6 +492,9 @@ void TCPHandler::runImpl() { if (exception) { + if (thread_trace_context) + thread_trace_context->root_span.addAttribute(*exception); + try { /// Try to send logs to client, but it could be risky too @@ -532,6 +543,7 @@ void TCPHandler::runImpl() /// (i.e. deallocations from the Aggregator with two-level aggregation) state.reset(); query_scope.reset(); + thread_trace_context.reset(); } catch (...) { diff --git a/src/Storages/AlterCommands.cpp b/src/Storages/AlterCommands.cpp index 7f99abf31fc..d370a67bfcc 100644 --- a/src/Storages/AlterCommands.cpp +++ b/src/Storages/AlterCommands.cpp @@ -1274,7 +1274,7 @@ void AlterCommands::validate(const StoragePtr & table, ContextPtr context) const throw Exception{"Table doesn't have SAMPLE BY, cannot remove", ErrorCodes::BAD_ARGUMENTS}; } - /// Collect default expressions for MODIFY and ADD comands + /// Collect default expressions for MODIFY and ADD commands if (command.type == AlterCommand::MODIFY_COLUMN || command.type == AlterCommand::ADD_COLUMN) { if (command.default_expression) diff --git a/src/Storages/Distributed/DirectoryMonitor.cpp b/src/Storages/Distributed/DirectoryMonitor.cpp index ff5a38fcc52..16981d26146 100644 --- a/src/Storages/Distributed/DirectoryMonitor.cpp +++ b/src/Storages/Distributed/DirectoryMonitor.cpp @@ -609,6 +609,8 @@ bool StorageDistributedDirectoryMonitor::processFiles(const std::mapgetSettingsRef()); @@ -627,6 +629,10 @@ void StorageDistributedDirectoryMonitor::processFile(const std::string & file_pa formatReadableQuantity(distributed_header.rows), formatReadableSizeWithBinarySuffix(distributed_header.bytes)); + thread_trace_context = std::make_unique(__PRETTY_FUNCTION__, + distributed_header.client_info.client_trace_context, + this->storage.getContext()->getOpenTelemetrySpanLog()); + RemoteInserter remote{*connection, timeouts, distributed_header.insert_query, distributed_header.insert_settings, @@ -637,10 +643,20 @@ void StorageDistributedDirectoryMonitor::processFile(const std::string & file_pa } catch (Exception & e) { + if (thread_trace_context) + thread_trace_context->root_span.addAttribute(std::current_exception()); + e.addMessage(fmt::format("While sending {}", file_path)); maybeMarkAsBroken(file_path, e); throw; } + catch (...) + { + if (thread_trace_context) + thread_trace_context->root_span.addAttribute(std::current_exception()); + + throw; + } auto dir_sync_guard = getDirectorySyncGuard(dir_fsync, disk, relative_path); markAsSend(file_path); @@ -854,6 +870,10 @@ private: ReadBufferFromFile in(file_path->second); const auto & distributed_header = readDistributedHeader(in, parent.log); + OpenTelemetry::TracingContextHolder thread_trace_context(__PRETTY_FUNCTION__, + distributed_header.client_info.client_trace_context, + parent.storage.getContext()->getOpenTelemetrySpanLog()); + if (!remote) { remote = std::make_unique(connection, timeouts, @@ -888,6 +908,11 @@ private: ReadBufferFromFile in(file_path->second); const auto & distributed_header = readDistributedHeader(in, parent.log); + // this function is called in a separated thread, so we set up the trace context from the file + OpenTelemetry::TracingContextHolder thread_trace_context(__PRETTY_FUNCTION__, + distributed_header.client_info.client_trace_context, + parent.storage.getContext()->getOpenTelemetrySpanLog()); + RemoteInserter remote(connection, timeouts, distributed_header.insert_query, distributed_header.insert_settings, diff --git a/src/Storages/Distributed/DistributedSink.cpp b/src/Storages/Distributed/DistributedSink.cpp index 8ea2954368e..ae72fdd84e2 100644 --- a/src/Storages/Distributed/DistributedSink.cpp +++ b/src/Storages/Distributed/DistributedSink.cpp @@ -336,7 +336,7 @@ DistributedSink::runWritingJob(JobReplica & job, const Block & current_block, si if (rows == 0) return; - OpenTelemetrySpanHolder span(__PRETTY_FUNCTION__); + OpenTelemetry::SpanHolder span(__PRETTY_FUNCTION__); span.addAttribute("clickhouse.shard_num", shard_info.shard_num); span.addAttribute("clickhouse.written_rows", rows); @@ -419,7 +419,7 @@ DistributedSink::runWritingJob(JobReplica & job, const Block & current_block, si void DistributedSink::writeSync(const Block & block) { - OpenTelemetrySpanHolder span(__PRETTY_FUNCTION__); + OpenTelemetry::SpanHolder span(__PRETTY_FUNCTION__); const Settings & settings = context->getSettingsRef(); const auto & shards_info = cluster->getShardsInfo(); @@ -610,7 +610,7 @@ void DistributedSink::writeSplitAsync(const Block & block) void DistributedSink::writeAsyncImpl(const Block & block, size_t shard_id) { - OpenTelemetrySpanHolder span("DistributedSink::writeAsyncImpl()"); + OpenTelemetry::SpanHolder span("DistributedSink::writeAsyncImpl()"); const auto & shard_info = cluster->getShardsInfo()[shard_id]; const auto & settings = context->getSettingsRef(); @@ -652,7 +652,7 @@ void DistributedSink::writeAsyncImpl(const Block & block, size_t shard_id) void DistributedSink::writeToLocal(const Block & block, size_t repeats) { - OpenTelemetrySpanHolder span(__PRETTY_FUNCTION__); + OpenTelemetry::SpanHolder span(__PRETTY_FUNCTION__); span.addAttribute("db.statement", this->query_string); InterpreterInsertQuery interp(query_ast, context, allow_materialized); @@ -668,7 +668,7 @@ void DistributedSink::writeToLocal(const Block & block, size_t repeats) void DistributedSink::writeToShard(const Block & block, const std::vector & dir_names) { - OpenTelemetrySpanHolder span(__PRETTY_FUNCTION__); + OpenTelemetry::SpanHolder span(__PRETTY_FUNCTION__); const auto & settings = context->getSettingsRef(); const auto & distributed_settings = storage.getDistributedSettingsRef(); @@ -737,11 +737,11 @@ void DistributedSink::writeToShard(const Block & block, const std::vectorgetSettingsRef().write(header_buf); - if (context->getClientInfo().client_trace_context.trace_id != UUID() && CurrentThread::isInitialized()) + if (OpenTelemetry::CurrentContext().isTraceEnabled()) { // if the distributed tracing is enabled, use the trace context in current thread as parent of next span auto client_info = context->getClientInfo(); - client_info.client_trace_context = CurrentThread::get().thread_trace_context; + client_info.client_trace_context = OpenTelemetry::CurrentContext(); client_info.write(header_buf, DBMS_TCP_PROTOCOL_VERSION); } else diff --git a/src/Storages/Kafka/StorageKafka.cpp b/src/Storages/Kafka/StorageKafka.cpp index b51457ba5d5..06ce4fb308d 100644 --- a/src/Storages/Kafka/StorageKafka.cpp +++ b/src/Storages/Kafka/StorageKafka.cpp @@ -810,7 +810,7 @@ void registerStorageKafka(StorageFactory & factory) /** Arguments of engine is following: * - Kafka broker list * - List of topics - * - Group ID (may be a constaint expression with a string result) + * - Group ID (may be a constant expression with a string result) * - Message format (string) * - Row delimiter * - Schema (optional, if the format supports it) diff --git a/src/Storages/LiveView/StorageLiveView.cpp b/src/Storages/LiveView/StorageLiveView.cpp index 6a079aa832f..b37dec6cbf1 100644 --- a/src/Storages/LiveView/StorageLiveView.cpp +++ b/src/Storages/LiveView/StorageLiveView.cpp @@ -382,7 +382,7 @@ bool StorageLiveView::getNewBlocks() BlocksMetadataPtr new_blocks_metadata = std::make_shared(); /// can't set mergeable_blocks here or anywhere else outside the writeIntoLiveView function - /// as there could be a race codition when the new block has been inserted into + /// as there could be a race condition when the new block has been inserted into /// the source table by the PushingToViews chain and this method /// called before writeIntoLiveView function is called which can lead to /// the same block added twice to the mergeable_blocks leading to diff --git a/src/Storages/MergeTree/DataPartsExchange.cpp b/src/Storages/MergeTree/DataPartsExchange.cpp index 02821de8629..9f8313a4700 100644 --- a/src/Storages/MergeTree/DataPartsExchange.cpp +++ b/src/Storages/MergeTree/DataPartsExchange.cpp @@ -773,6 +773,7 @@ MergeTreeData::MutableDataPartPtr Fetcher::downloadPartToDisk( ThrottlerPtr throttler) { assert(!tmp_prefix.empty()); + const auto data_settings = data.getSettings(); /// We will remove directory if it's already exists. Make precautions. if (tmp_prefix.empty() //-V560 @@ -800,7 +801,14 @@ MergeTreeData::MutableDataPartPtr Fetcher::downloadPartToDisk( { LOG_WARNING(log, "Directory {} already exists, probably result of a failed fetch. Will remove it before fetching part.", data_part_storage_builder->getFullPath()); - data_part_storage_builder->removeRecursive(); + + /// Even if it's a temporary part it could be downloaded with zero copy replication and this function + /// is executed as a callback. + /// + /// We don't control the amount of refs for temporary parts so we cannot decide can we remove blobs + /// or not. So we are not doing it + bool keep_shared = disk->supportZeroCopyReplication() && data_settings->allow_remote_fs_zero_copy_replication; + data_part_storage_builder->removeSharedRecursive(keep_shared); } data_part_storage_builder->createDirectories(); diff --git a/src/Storages/MergeTree/KeyCondition.cpp b/src/Storages/MergeTree/KeyCondition.cpp index 04007ce9356..f5eeb4ed35c 100644 --- a/src/Storages/MergeTree/KeyCondition.cpp +++ b/src/Storages/MergeTree/KeyCondition.cpp @@ -297,8 +297,10 @@ public: assert(indexes_mapping.size() == data_types.size()); for (size_t i = 0; i < indexes_mapping.size(); ++i) + { if (!candidate_set->areTypesEqual(indexes_mapping[i].tuple_index, data_types[i])) return false; + } return true; }; diff --git a/src/Storages/MergeTree/MergeTreeBackgroundExecutor.h b/src/Storages/MergeTree/MergeTreeBackgroundExecutor.h index a2f97d6bbc6..0fc888dd6ad 100644 --- a/src/Storages/MergeTree/MergeTreeBackgroundExecutor.h +++ b/src/Storages/MergeTree/MergeTreeBackgroundExecutor.h @@ -155,7 +155,7 @@ private: * We use boost::circular_buffer as a container for queues not to do any allocations. * * Another nuisance that we faces with is than background operations always interact with an associated Storage. - * So, when a Storage want to shutdown, it must wait until all its background operaions are finished. + * So, when a Storage want to shutdown, it must wait until all its background operations are finished. */ template class MergeTreeBackgroundExecutor final : boost::noncopyable diff --git a/src/Storages/MergeTree/MergeTreeData.cpp b/src/Storages/MergeTree/MergeTreeData.cpp index ee2cbf4fe8a..239451a16bb 100644 --- a/src/Storages/MergeTree/MergeTreeData.cpp +++ b/src/Storages/MergeTree/MergeTreeData.cpp @@ -1592,7 +1592,21 @@ size_t MergeTreeData::clearOldTemporaryDirectories(size_t custom_directories_lif else { LOG_WARNING(log, "Removing temporary directory {}", full_path); - disk->removeRecursive(it->path()); + + /// Even if it's a temporary part it could be downloaded with zero copy replication and this function + /// is executed as a callback. + /// + /// We don't control the amount of refs for temporary parts so we cannot decide can we remove blobs + /// or not. So we are not doing it + bool keep_shared = false; + if (it->path().find("fetch") != std::string::npos) + { + keep_shared = disk->supportZeroCopyReplication() && settings->allow_remote_fs_zero_copy_replication; + if (keep_shared) + LOG_WARNING(log, "Since zero-copy replication is enabled we are not going to remove blobs from shared storage for {}", full_path); + } + + disk->removeSharedRecursive(it->path(), keep_shared, {}); ++cleared_count; } } @@ -5541,6 +5555,10 @@ std::optional MergeTreeData::getQueryProcessingStageWithAgg if (select_query->interpolate() && !select_query->interpolate()->children.empty()) return std::nullopt; + // Currently projections don't support GROUPING SET yet. + if (select_query->group_by_with_grouping_sets) + return std::nullopt; + auto query_options = SelectQueryOptions( QueryProcessingStage::WithMergeableState, /* depth */ 1, diff --git a/src/Storages/MergeTree/MergeTreeDataPartTTLInfo.h b/src/Storages/MergeTree/MergeTreeDataPartTTLInfo.h index 71ef6edd7fb..3080e285ac8 100644 --- a/src/Storages/MergeTree/MergeTreeDataPartTTLInfo.h +++ b/src/Storages/MergeTree/MergeTreeDataPartTTLInfo.h @@ -16,7 +16,7 @@ struct MergeTreeDataPartTTLInfo time_t max = 0; /// This TTL was computed on completely expired part. It doesn't make sense - /// to select such parts for TTL again. But make sense to recalcuate TTL + /// to select such parts for TTL again. But make sense to recalculate TTL /// again for merge with multiple parts. std::optional ttl_finished; bool finished() const { return ttl_finished.value_or(false); } diff --git a/src/Storages/MergeTree/MergeTreeDataSelectExecutor.cpp b/src/Storages/MergeTree/MergeTreeDataSelectExecutor.cpp index c5f546a9c36..709a8babcdd 100644 --- a/src/Storages/MergeTree/MergeTreeDataSelectExecutor.cpp +++ b/src/Storages/MergeTree/MergeTreeDataSelectExecutor.cpp @@ -242,7 +242,7 @@ QueryPlanPtr MergeTreeDataSelectExecutor::read( auto sorting_step = std::make_unique( projection_plan->getCurrentDataStream(), - query_info.projection->input_order_info->order_key_prefix_descr, + query_info.projection->input_order_info->sort_description_for_merging, output_order_descr, settings.max_block_size, limit); @@ -313,6 +313,7 @@ QueryPlanPtr MergeTreeDataSelectExecutor::read( settings.min_free_disk_space_for_temporary_data, settings.compile_aggregate_expressions, settings.min_count_to_compile_aggregate_expression, + settings.max_block_size, only_merge); return std::make_pair(params, only_merge); diff --git a/src/Storages/MergeTree/PartMetadataManagerWithCache.cpp b/src/Storages/MergeTree/PartMetadataManagerWithCache.cpp index 9930aca2576..5a291373e6c 100644 --- a/src/Storages/MergeTree/PartMetadataManagerWithCache.cpp +++ b/src/Storages/MergeTree/PartMetadataManagerWithCache.cpp @@ -191,6 +191,7 @@ void PartMetadataManagerWithCache::getKeysAndCheckSums(Strings & keys, std::vect { ReadBufferFromString rbuf(values[i]); HashingReadBuffer hbuf(rbuf); + hbuf.ignoreAll(); checksums.push_back(hbuf.getHash()); } } diff --git a/src/Storages/MergeTree/ReplicatedMergeTreeQueue.h b/src/Storages/MergeTree/ReplicatedMergeTreeQueue.h index e8362e5cc6b..6d1a3efb01d 100644 --- a/src/Storages/MergeTree/ReplicatedMergeTreeQueue.h +++ b/src/Storages/MergeTree/ReplicatedMergeTreeQueue.h @@ -279,7 +279,7 @@ private: /// Very large queue entries may appear occasionally. /// We cannot process MAX_MULTI_OPS at once because it will fail. /// But we have to process more than one entry at once because otherwise lagged replicas keep up slowly. - /// Let's start with one entry per transaction and icrease it exponentially towards MAX_MULTI_OPS. + /// Let's start with one entry per transaction and increase it exponentially towards MAX_MULTI_OPS. /// It will allow to make some progress before failing and remain operational even in extreme cases. size_t current_multi_batch_size = 1; diff --git a/src/Storages/NATS/StorageNATS.cpp b/src/Storages/NATS/StorageNATS.cpp index 3c1a04c7824..fc3079a7aa7 100644 --- a/src/Storages/NATS/StorageNATS.cpp +++ b/src/Storages/NATS/StorageNATS.cpp @@ -144,6 +144,8 @@ ContextMutablePtr StorageNATS::addSettings(ContextPtr local_context) const modified_context->setSetting("input_format_skip_unknown_fields", true); modified_context->setSetting("input_format_allow_errors_ratio", 0.); modified_context->setSetting("input_format_allow_errors_num", nats_settings->nats_skip_broken_messages.value); + /// Since we are reusing the same context for all queries executed simultaneously, we don't want to used shared `analyze_count` + modified_context->setSetting("max_analyze_depth", Field{0}); if (!schema_name.empty()) modified_context->setSetting("format_schema", schema_name); diff --git a/src/Storages/PartitionCommands.h b/src/Storages/PartitionCommands.h index 9807c90bc23..4921cf8e53b 100644 --- a/src/Storages/PartitionCommands.h +++ b/src/Storages/PartitionCommands.h @@ -104,7 +104,7 @@ struct PartitionCommandResultInfo using PartitionCommandsResultInfo = std::vector; -/// Convert partition comands result to Source from single Chunk, which will be +/// Convert partition commands result to Source from single Chunk, which will be /// used to print info to the user. Tries to create narrowest table for given /// results. For example, if all commands were FREEZE commands, than /// old_part_name column will be absent. diff --git a/src/Storages/PostgreSQL/MaterializedPostgreSQLConsumer.h b/src/Storages/PostgreSQL/MaterializedPostgreSQLConsumer.h index 91bf5eeccde..37caa66aae5 100644 --- a/src/Storages/PostgreSQL/MaterializedPostgreSQLConsumer.h +++ b/src/Storages/PostgreSQL/MaterializedPostgreSQLConsumer.h @@ -126,7 +126,7 @@ private: static void assertCorrectInsertion(StorageData::Buffer & buffer, size_t column_idx); - /// lsn - log sequnce nuumber, like wal offset (64 bit). + /// lsn - log sequence number, like wal offset (64 bit). static Int64 getLSNValue(const std::string & lsn) { UInt32 upper_half, lower_half; diff --git a/src/Storages/RabbitMQ/StorageRabbitMQ.cpp b/src/Storages/RabbitMQ/StorageRabbitMQ.cpp index f831f81cd22..70838daec24 100644 --- a/src/Storages/RabbitMQ/StorageRabbitMQ.cpp +++ b/src/Storages/RabbitMQ/StorageRabbitMQ.cpp @@ -242,6 +242,8 @@ ContextMutablePtr StorageRabbitMQ::addSettings(ContextPtr local_context) const modified_context->setSetting("input_format_skip_unknown_fields", true); modified_context->setSetting("input_format_allow_errors_ratio", 0.); modified_context->setSetting("input_format_allow_errors_num", rabbitmq_settings->rabbitmq_skip_broken_messages.value); + /// Since we are reusing the same context for all queries executed simultaneously, we don't want to used shared `analyze_count` + modified_context->setSetting("max_analyze_depth", Field{0}); if (!schema_name.empty()) modified_context->setSetting("format_schema", schema_name); diff --git a/src/Storages/ReadInOrderOptimizer.cpp b/src/Storages/ReadInOrderOptimizer.cpp index b188cef065e..b67da14365e 100644 --- a/src/Storages/ReadInOrderOptimizer.cpp +++ b/src/Storages/ReadInOrderOptimizer.cpp @@ -5,10 +5,12 @@ #include #include #include +#include #include #include #include #include +#include namespace DB { @@ -21,26 +23,46 @@ namespace ErrorCodes namespace { -ASTPtr getFixedPoint(const ASTPtr & ast) +/// Finds expression like x = 'y' or f(x) = 'y', +/// where `x` is identifier, 'y' is literal and `f` is injective functions. +ASTPtr getFixedPoint(const ASTPtr & ast, const ContextPtr & context) { const auto * func = ast->as(); if (!func || func->name != "equals") return nullptr; + if (!func->arguments || func->arguments->children.size() != 2) + return nullptr; + const auto & lhs = func->arguments->children[0]; const auto & rhs = func->arguments->children[1]; - if (lhs->as()) - return rhs; + if (!lhs->as() && !rhs->as()) + return nullptr; - if (rhs->as()) - return lhs; + /// Case of two literals doesn't make sense. + if (lhs->as() && rhs->as()) + return nullptr; - return nullptr; + /// If indetifier is wrapped into injective functions, remove them. + auto argument = lhs->as() ? rhs : lhs; + while (const auto * arg_func = argument->as()) + { + if (!arg_func->arguments || arg_func->arguments->children.size() != 1) + return nullptr; + + auto func_resolver = FunctionFactory::instance().tryGet(arg_func->name, context); + if (!func_resolver || !func_resolver->isInjective({})) + return nullptr; + + argument = arg_func->arguments->children[0]; + } + + return argument->as() ? argument : nullptr; } -size_t calculateFixedPrefixSize( - const ASTSelectQuery & query, const Names & sorting_key_columns) +NameSet getFixedSortingColumns( + const ASTSelectQuery & query, const Names & sorting_key_columns, const ContextPtr & context) { ASTPtr condition; if (query.where() && query.prewhere()) @@ -51,14 +73,15 @@ size_t calculateFixedPrefixSize( condition = query.prewhere(); if (!condition) - return 0; + return {}; /// Convert condition to CNF for more convenient analysis. auto cnf = TreeCNFConverter::tryConvertToCNF(condition); if (!cnf) - return 0; + return {}; NameSet fixed_points; + NameSet sorting_key_columns_set(sorting_key_columns.begin(), sorting_key_columns.end()); /// If we met expression like 'column = x', where 'x' is literal, /// in clause of size 1 in CNF, then we can guarantee @@ -67,22 +90,17 @@ size_t calculateFixedPrefixSize( { if (group.size() == 1 && !group.begin()->negative) { - auto fixed_point = getFixedPoint(group.begin()->ast); + auto fixed_point = getFixedPoint(group.begin()->ast, context); if (fixed_point) - fixed_points.insert(fixed_point->getColumnName()); + { + auto column_name = fixed_point->getColumnName(); + if (sorting_key_columns_set.contains(column_name)) + fixed_points.insert(column_name); + } } }); - size_t prefix_size = 0; - for (const auto & column_name : sorting_key_columns) - { - if (!fixed_points.contains(column_name)) - break; - - ++prefix_size; - } - - return prefix_size; + return fixed_points; } /// Optimize in case of exact match with order key element @@ -181,46 +199,54 @@ InputOrderInfoPtr ReadInOrderOptimizer::getInputOrderImpl( const StorageMetadataPtr & metadata_snapshot, const SortDescription & description, const ManyExpressionActions & actions, + const ContextPtr & context, UInt64 limit) const { auto sorting_key_columns = metadata_snapshot->getSortingKeyColumns(); int read_direction = description.at(0).direction; - size_t fixed_prefix_size = calculateFixedPrefixSize(query, sorting_key_columns); - size_t descr_prefix_size = std::min(description.size(), sorting_key_columns.size() - fixed_prefix_size); + auto fixed_sorting_columns = getFixedSortingColumns(query, sorting_key_columns, context); - SortDescription order_key_prefix_descr; - order_key_prefix_descr.reserve(descr_prefix_size); + SortDescription sort_description_for_merging; + sort_description_for_merging.reserve(description.size()); - for (size_t i = 0; i < descr_prefix_size; ++i) + size_t desc_pos = 0; + size_t key_pos = 0; + + while (desc_pos < description.size() && key_pos < sorting_key_columns.size()) { - if (forbidden_columns.contains(description[i].column_name)) + if (forbidden_columns.contains(description[desc_pos].column_name)) break; - int current_direction = matchSortDescriptionAndKey( - actions[i]->getActions(), description[i], sorting_key_columns[i + fixed_prefix_size]); + int current_direction = matchSortDescriptionAndKey(actions[desc_pos]->getActions(), description[desc_pos], sorting_key_columns[key_pos]); + bool is_matched = current_direction && (desc_pos == 0 || current_direction == read_direction); + + if (!is_matched) + { + /// If one of the sorting columns is constant after filtering, + /// skip it, because it won't affect order anymore. + if (fixed_sorting_columns.contains(sorting_key_columns[key_pos])) + { + ++key_pos; + continue; + } - if (!current_direction || (i > 0 && current_direction != read_direction)) break; + } - if (i == 0) + if (desc_pos == 0) read_direction = current_direction; - order_key_prefix_descr.push_back(required_sort_description[i]); + sort_description_for_merging.push_back(description[desc_pos]); + + ++desc_pos; + ++key_pos; } - if (order_key_prefix_descr.empty()) + if (sort_description_for_merging.empty()) return {}; - SortDescription order_key_fixed_prefix_descr; - order_key_fixed_prefix_descr.reserve(fixed_prefix_size); - for (size_t i = 0; i < fixed_prefix_size; ++i) - order_key_fixed_prefix_descr.emplace_back(sorting_key_columns[i], read_direction); - - return std::make_shared( - std::move(order_key_fixed_prefix_descr), - std::move(order_key_prefix_descr), - read_direction, limit); + return std::make_shared(std::move(sort_description_for_merging), key_pos, read_direction, limit); } InputOrderInfoPtr ReadInOrderOptimizer::getInputOrder( @@ -255,10 +281,10 @@ InputOrderInfoPtr ReadInOrderOptimizer::getInputOrder( aliases_actions[i] = expression_analyzer.getActions(true); } - return getInputOrderImpl(metadata_snapshot, aliases_sort_description, aliases_actions, limit); + return getInputOrderImpl(metadata_snapshot, aliases_sort_description, aliases_actions, context, limit); } - return getInputOrderImpl(metadata_snapshot, required_sort_description, elements_actions, limit); + return getInputOrderImpl(metadata_snapshot, required_sort_description, elements_actions, context, limit); } } diff --git a/src/Storages/ReadInOrderOptimizer.h b/src/Storages/ReadInOrderOptimizer.h index fd8c9187ddb..161c321ae79 100644 --- a/src/Storages/ReadInOrderOptimizer.h +++ b/src/Storages/ReadInOrderOptimizer.h @@ -12,8 +12,6 @@ namespace DB * common prefix, which is needed for * performing reading in order of PK. */ -class Context; - class ReadInOrderOptimizer { public: @@ -30,6 +28,7 @@ private: const StorageMetadataPtr & metadata_snapshot, const SortDescription & description, const ManyExpressionActions & actions, + const ContextPtr & context, UInt64 limit) const; /// Actions for every element of order expression to analyze functions for monotonicity diff --git a/src/Storages/SelectQueryInfo.h b/src/Storages/SelectQueryInfo.h index 758e4736fc1..c41b422199d 100644 --- a/src/Storages/SelectQueryInfo.h +++ b/src/Storages/SelectQueryInfo.h @@ -101,17 +101,33 @@ struct FilterDAGInfo struct InputOrderInfo { - SortDescription order_key_fixed_prefix_descr; - SortDescription order_key_prefix_descr; + /// Sort description for merging of already sorted streams. + /// Always a prefix of ORDER BY or GROUP BY description specified in query. + SortDescription sort_description_for_merging; + + /** Size of prefix of sorting key that is already + * sorted before execution of sorting or aggreagation. + * + * Contains both columns that scpecified in + * ORDER BY or GROUP BY clause of query + * and columns that turned out to be already sorted. + * + * E.g. if we have sorting key ORDER BY (a, b, c, d) + * and query with `WHERE a = 'x' AND b = 'y' ORDER BY c, d` clauses. + * sort_description_for_merging will be equal to (c, d) and + * used_prefix_of_sorting_key_size will be equal to 4. + */ + size_t used_prefix_of_sorting_key_size; + int direction; UInt64 limit; InputOrderInfo( - const SortDescription & order_key_fixed_prefix_descr_, - const SortDescription & order_key_prefix_descr_, + const SortDescription & sort_description_for_merging_, + size_t used_prefix_of_sorting_key_size_, int direction_, UInt64 limit_) - : order_key_fixed_prefix_descr(order_key_fixed_prefix_descr_) - , order_key_prefix_descr(order_key_prefix_descr_) + : sort_description_for_merging(sort_description_for_merging_) + , used_prefix_of_sorting_key_size(used_prefix_of_sorting_key_size_) , direction(direction_), limit(limit_) { } diff --git a/src/Storages/StorageMergeTree.cpp b/src/Storages/StorageMergeTree.cpp index 507047751f3..5ce46ecffc8 100644 --- a/src/Storages/StorageMergeTree.cpp +++ b/src/Storages/StorageMergeTree.cpp @@ -963,7 +963,7 @@ bool StorageMergeTree::merge( if (!merge_mutate_entry) return false; - /// Copying a vector of columns `deduplicate bu columns. + /// Copying a vector of columns `deduplicate by columns. IExecutableTask::TaskResultCallback f = [](bool) {}; auto task = std::make_shared( *this, metadata_snapshot, deduplicate, deduplicate_by_columns, merge_mutate_entry, table_lock_holder, f); diff --git a/src/Storages/StorageURL.cpp b/src/Storages/StorageURL.cpp index c3da392b9cd..8d0d85a39ef 100644 --- a/src/Storages/StorageURL.cpp +++ b/src/Storages/StorageURL.cpp @@ -100,20 +100,19 @@ namespace ReadWriteBufferFromHTTP::HTTPHeaderEntries getHeaders(const ReadWriteBufferFromHTTP::HTTPHeaderEntries & headers_) { ReadWriteBufferFromHTTP::HTTPHeaderEntries headers(headers_.begin(), headers_.end()); - // Propagate OpenTelemetry trace context, if any, downstream. - if (CurrentThread::isInitialized()) - { - const auto & thread_trace_context = CurrentThread::get().thread_trace_context; - if (thread_trace_context.trace_id != UUID()) - { - headers.emplace_back("traceparent", thread_trace_context.composeTraceparentHeader()); - if (!thread_trace_context.tracestate.empty()) - { - headers.emplace_back("tracestate", thread_trace_context.tracestate); - } + // Propagate OpenTelemetry trace context, if any, downstream. + const auto ¤t_trace_context = OpenTelemetry::CurrentContext(); + if (current_trace_context.isTraceEnabled()) + { + headers.emplace_back("traceparent", current_trace_context.composeTraceparentHeader()); + + if (!current_trace_context.tracestate.empty()) + { + headers.emplace_back("tracestate", current_trace_context.tracestate); } } + return headers; } diff --git a/src/Storages/StorageView.cpp b/src/Storages/StorageView.cpp index 3377af685f0..adaf1c4e404 100644 --- a/src/Storages/StorageView.cpp +++ b/src/Storages/StorageView.cpp @@ -179,12 +179,14 @@ void StorageView::replaceWithSubquery(ASTSelectQuery & outer_query, ASTPtr view_ if (!table_expression->database_and_table_name) { - // If it's a view table function, add a fake db.table name. + // If it's a view or merge table function, add a fake db.table name. if (table_expression->table_function) { auto table_function_name = table_expression->table_function->as()->name; - if ((table_function_name == "view") || (table_function_name == "viewIfPermitted")) + if (table_function_name == "view" || table_function_name == "viewIfPermitted") table_expression->database_and_table_name = std::make_shared("__view"); + if (table_function_name == "merge") + table_expression->database_and_table_name = std::make_shared("__merge"); } if (!table_expression->database_and_table_name) throw Exception("Logical error: incorrect table expression", ErrorCodes::LOGICAL_ERROR); diff --git a/src/Storages/System/StorageSystemDistributionQueue.cpp b/src/Storages/System/StorageSystemDistributionQueue.cpp index d8879c3655e..5297c4eb93c 100644 --- a/src/Storages/System/StorageSystemDistributionQueue.cpp +++ b/src/Storages/System/StorageSystemDistributionQueue.cpp @@ -57,7 +57,7 @@ std::string maskDataPath(const std::string & path) size_t user_pw_end = masked_path.find('@', node_pos); if (user_pw_end == std::string::npos) { - /// Likey new format (use_compact_format_in_distributed_parts_names=1) + /// Likely new format (use_compact_format_in_distributed_parts_names=1) return path; } diff --git a/src/Storages/System/StorageSystemFilesystemCache.cpp b/src/Storages/System/StorageSystemFilesystemCache.cpp index ba94cb01b75..cd9324b3253 100644 --- a/src/Storages/System/StorageSystemFilesystemCache.cpp +++ b/src/Storages/System/StorageSystemFilesystemCache.cpp @@ -2,9 +2,9 @@ #include #include #include -#include -#include -#include +#include +#include +#include #include #include diff --git a/src/Storages/System/StorageSystemRemoteDataPaths.cpp b/src/Storages/System/StorageSystemRemoteDataPaths.cpp index 662255d6afd..fe7aaf97970 100644 --- a/src/Storages/System/StorageSystemRemoteDataPaths.cpp +++ b/src/Storages/System/StorageSystemRemoteDataPaths.cpp @@ -1,8 +1,8 @@ #include "StorageSystemRemoteDataPaths.h" #include #include -#include -#include +#include +#include #include #include #include diff --git a/tests/ci/cherry_pick.py b/tests/ci/cherry_pick.py index a0a77a2684e..064a0b3add1 100644 --- a/tests/ci/cherry_pick.py +++ b/tests/ci/cherry_pick.py @@ -206,14 +206,7 @@ Merge it only if you intend to backport changes to the target branch, otherwise ) self.cherrypick_pr.add_to_labels(Labels.LABEL_CHERRYPICK) self.cherrypick_pr.add_to_labels(Labels.LABEL_DO_NOT_TEST) - if self.pr.assignees: - logging.info( - "Assing to assignees of the original PR: %s", - ", ".join(user.login for user in self.pr.assignees), - ) - self.cherrypick_pr.add_to_assignees(*self.pr.assignees) - logging.info("Assign to the author of the original PR: %s", self.pr.user.login) - self.cherrypick_pr.add_to_assignees(self.pr.user) + self._assign_new_pr(self.cherrypick_pr) def create_backport(self): # Checkout the backport branch from the remote and make all changes to @@ -244,14 +237,21 @@ Merge it only if you intend to backport changes to the target branch, otherwise head=self.backport_branch, ) self.backport_pr.add_to_labels(Labels.LABEL_BACKPORT) + self._assign_new_pr(self.backport_pr) + + def _assign_new_pr(self, new_pr: PullRequest): + """Assign `new_pr` to author, merger and assignees of an original PR""" + # It looks there some race when multiple .add_to_assignees are executed, + # so we'll add all at once + assignees = [self.pr.user, self.pr.merged_by] if self.pr.assignees: - logging.info( - "Assing to assignees of the original PR: %s", - ", ".join(user.login for user in self.pr.assignees), - ) - self.cherrypick_pr.add_to_assignees(*self.pr.assignees) - logging.info("Assign to the author of the original PR: %s", self.pr.user.login) - self.backport_pr.add_to_assignees(self.pr.user) + assignees.extend(self.pr.assignees) + logging.info( + "Assing #%s to author and assignees of the original PR: %s", + new_pr.number, + ", ".join(user.login for user in assignees), + ) + new_pr.add_to_assignees(*assignees) @property def backported(self) -> bool: diff --git a/tests/ci/ci_config.py b/tests/ci/ci_config.py index 61dfb07f762..3d0513bca47 100644 --- a/tests/ci/ci_config.py +++ b/tests/ci/ci_config.py @@ -187,13 +187,13 @@ CI_CONFIG = { "tests_config": { # required_build - build name for artifacts # force_tests - force success status for tests - "Stateful tests (address)": { + "Stateful tests (asan)": { "required_build": "package_asan", }, - "Stateful tests (thread)": { + "Stateful tests (tsan)": { "required_build": "package_tsan", }, - "Stateful tests (memory)": { + "Stateful tests (msan)": { "required_build": "package_msan", }, "Stateful tests (ubsan)": { @@ -214,13 +214,13 @@ CI_CONFIG = { "Stateful tests (release, DatabaseReplicated)": { "required_build": "package_release", }, - "Stateless tests (address)": { + "Stateless tests (asan)": { "required_build": "package_asan", }, - "Stateless tests (thread)": { + "Stateless tests (tsan)": { "required_build": "package_tsan", }, - "Stateless tests (memory)": { + "Stateless tests (msan)": { "required_build": "package_msan", }, "Stateless tests (ubsan)": { @@ -247,16 +247,16 @@ CI_CONFIG = { "Stateless tests (release, s3 storage)": { "required_build": "package_release", }, - "Stress test (address)": { + "Stress test (asan)": { "required_build": "package_asan", }, - "Stress test (thread)": { + "Stress test (tsan)": { "required_build": "package_tsan", }, - "Stress test (undefined)": { + "Stress test (ubsan)": { "required_build": "package_ubsan", }, - "Stress test (memory)": { + "Stress test (msan)": { "required_build": "package_msan", }, "Stress test (debug)": { @@ -265,13 +265,13 @@ CI_CONFIG = { "Integration tests (asan)": { "required_build": "package_asan", }, - "Integration tests (thread)": { + "Integration tests (tsan)": { "required_build": "package_tsan", }, "Integration tests (release)": { "required_build": "package_release", }, - "Integration tests (memory)": { + "Integration tests (msan)": { "required_build": "package_msan", }, "Integration tests flaky check (asan)": { @@ -301,19 +301,19 @@ CI_CONFIG = { "AST fuzzer (debug)": { "required_build": "package_debug", }, - "AST fuzzer (ASan)": { + "AST fuzzer (asan)": { "required_build": "package_asan", }, - "AST fuzzer (MSan)": { + "AST fuzzer (msan)": { "required_build": "package_msan", }, - "AST fuzzer (TSan)": { + "AST fuzzer (tsan)": { "required_build": "package_tsan", }, - "AST fuzzer (UBSan)": { + "AST fuzzer (ubsan)": { "required_build": "package_ubsan", }, - "Stateless tests flaky check (address)": { + "Stateless tests flaky check (asan)": { "required_build": "package_asan", }, "ClickHouse Keeper Jepsen": { diff --git a/tests/ci/run_check.py b/tests/ci/run_check.py index 5e6542f6e4c..0a269c07642 100644 --- a/tests/ci/run_check.py +++ b/tests/ci/run_check.py @@ -20,8 +20,6 @@ from workflow_approve_rerun_lambda.app import TRUSTED_CONTRIBUTORS NAME = "Run Check" TRUSTED_ORG_IDS = { - 7409213, # yandex - 28471076, # altinity 54801242, # clickhouse } diff --git a/tests/ci/workflow_approve_rerun_lambda/app.py b/tests/ci/workflow_approve_rerun_lambda/app.py index 29f3271a34c..39bd9cfb283 100644 --- a/tests/ci/workflow_approve_rerun_lambda/app.py +++ b/tests/ci/workflow_approve_rerun_lambda/app.py @@ -50,8 +50,6 @@ WorkflowDescription = namedtuple( # See https://api.github.com/orgs/{name} TRUSTED_ORG_IDS = { - 7409213, # yandex - 28471076, # altinity 54801242, # clickhouse } @@ -104,8 +102,6 @@ TRUSTED_CONTRIBUTORS = { "kreuzerkrieg", "lehasm", # DOCSUP "michon470", # DOCSUP - "MyroTk", # Tester in Altinity - "myrrc", # Michael Kot, Altinity "nikvas0", "nvartolomei", "olgarev", # DOCSUP diff --git a/tests/clickhouse-test b/tests/clickhouse-test index f40c93c6f5d..14cf4d0674a 100755 --- a/tests/clickhouse-test +++ b/tests/clickhouse-test @@ -533,7 +533,7 @@ class TestCase: else: # If --database is not specified, we will create temporary database with # unique name and we will recreate and drop it for each test - def random_str(length=6): + def random_str(length=8): alphabet = string.ascii_lowercase + string.digits # NOTE: it is important not to use default random generator, since it shares state. return "".join( @@ -995,9 +995,12 @@ class TestCase: args.timeout - (datetime.now() - start_time).total_seconds(), 20 ) try: + drop_database_query = "DROP DATABASE " + database + if args.replicated_database: + drop_database_query += " ON CLUSTER test_cluster_database_replicated" clickhouse_execute( args, - "DROP DATABASE " + database, + drop_database_query, timeout=seconds_left, settings={ "log_comment": args.testcase_basename, diff --git a/tests/instructions/clang-tidy.txt b/tests/instructions/clang-tidy.txt deleted file mode 100644 index 84145564bf0..00000000000 --- a/tests/instructions/clang-tidy.txt +++ /dev/null @@ -1,2 +0,0 @@ -# clang-tidy has been integrated into CMake: -# --> Build ClickHouse with -DENABLE_CLANG_TIDY=1 and see cmake/clang_tidy.cmake for details diff --git a/tests/integration/test_backup_restore_new/test.py b/tests/integration/test_backup_restore_new/test.py index 2fe3bb99e45..ca0d6a632a0 100644 --- a/tests/integration/test_backup_restore_new/test.py +++ b/tests/integration/test_backup_restore_new/test.py @@ -224,6 +224,89 @@ def test_incremental_backup_after_renaming_table(): assert instance.query("SELECT count(), sum(x) FROM test.table2") == "100\t4950\n" +def test_incremental_backup_for_log_family(): + backup_name = new_backup_name() + create_and_fill_table(engine="Log") + + assert instance.query("SELECT count(), sum(x) FROM test.table") == "100\t4950\n" + instance.query(f"BACKUP TABLE test.table TO {backup_name}") + + instance.query("INSERT INTO test.table VALUES (65, 'a'), (66, 'b')") + + assert instance.query("SELECT count(), sum(x) FROM test.table") == "102\t5081\n" + + backup_name2 = new_backup_name() + instance.query(f"BACKUP TABLE test.table TO {backup_name2}") + + backup_name_inc = new_backup_name() + instance.query( + f"BACKUP TABLE test.table TO {backup_name_inc} SETTINGS base_backup = {backup_name}" + ) + + metadata_path = os.path.join( + get_path_to_backup(backup_name), "metadata/test/table.sql" + ) + + metadata_path2 = os.path.join( + get_path_to_backup(backup_name2), "metadata/test/table.sql" + ) + + metadata_path_inc = os.path.join( + get_path_to_backup(backup_name_inc), "metadata/test/table.sql" + ) + + assert os.path.isfile(metadata_path) + assert os.path.isfile(metadata_path2) + assert not os.path.isfile(metadata_path_inc) + assert os.path.getsize(metadata_path) > 0 + assert os.path.getsize(metadata_path) == os.path.getsize(metadata_path2) + + x_bin_path = os.path.join(get_path_to_backup(backup_name), "data/test/table/x.bin") + y_bin_path = os.path.join(get_path_to_backup(backup_name), "data/test/table/y.bin") + + x_bin_path2 = os.path.join( + get_path_to_backup(backup_name2), "data/test/table/x.bin" + ) + y_bin_path2 = os.path.join( + get_path_to_backup(backup_name2), "data/test/table/y.bin" + ) + + x_bin_path_inc = os.path.join( + get_path_to_backup(backup_name_inc), "data/test/table/x.bin" + ) + + y_bin_path_inc = os.path.join( + get_path_to_backup(backup_name_inc), "data/test/table/y.bin" + ) + + assert os.path.isfile(x_bin_path) + assert os.path.isfile(y_bin_path) + assert os.path.isfile(x_bin_path2) + assert os.path.isfile(y_bin_path2) + assert os.path.isfile(x_bin_path_inc) + assert os.path.isfile(y_bin_path_inc) + + x_bin_size = os.path.getsize(x_bin_path) + y_bin_size = os.path.getsize(y_bin_path) + x_bin_size2 = os.path.getsize(x_bin_path2) + y_bin_size2 = os.path.getsize(y_bin_path2) + x_bin_size_inc = os.path.getsize(x_bin_path_inc) + y_bin_size_inc = os.path.getsize(y_bin_path_inc) + + assert x_bin_size > 0 + assert y_bin_size > 0 + assert x_bin_size2 > 0 + assert y_bin_size2 > 0 + assert x_bin_size_inc > 0 + assert y_bin_size_inc > 0 + assert x_bin_size2 == x_bin_size + x_bin_size_inc + assert y_bin_size2 == y_bin_size + y_bin_size_inc + + instance.query(f"RESTORE TABLE test.table AS test.table2 FROM {backup_name_inc}") + + assert instance.query("SELECT count(), sum(x) FROM test.table2") == "102\t5081\n" + + def test_backup_not_found_or_already_exists(): backup_name = new_backup_name() diff --git a/tests/integration/test_detached_parts_metrics/__init__.py b/tests/integration/test_detached_parts_metrics/__init__.py new file mode 100644 index 00000000000..e69de29bb2d diff --git a/tests/integration/test_detached_parts_metrics/configs/asynchronous_metrics_update_period_s.xml b/tests/integration/test_detached_parts_metrics/configs/asynchronous_metrics_update_period_s.xml new file mode 100644 index 00000000000..0a56d734805 --- /dev/null +++ b/tests/integration/test_detached_parts_metrics/configs/asynchronous_metrics_update_period_s.xml @@ -0,0 +1,4 @@ + + 1 + 1 + diff --git a/tests/integration/test_detached_parts_metrics/test.py b/tests/integration/test_detached_parts_metrics/test.py new file mode 100644 index 00000000000..62b70ebd430 --- /dev/null +++ b/tests/integration/test_detached_parts_metrics/test.py @@ -0,0 +1,133 @@ +import time +import pytest +from helpers.cluster import ClickHouseCluster +from helpers.test_tools import assert_eq_with_retry + + +cluster = ClickHouseCluster(__file__) +node1 = cluster.add_instance( + "node1", + main_configs=["configs/asynchronous_metrics_update_period_s.xml"], +) + + +@pytest.fixture(scope="module") +def started_cluster(): + try: + cluster.start() + yield cluster + finally: + cluster.shutdown() + + +def test_event_time_microseconds_field(started_cluster): + cluster.start() + query_create = """ + CREATE TABLE t + ( + id Int64, + event_time Date + ) + Engine=MergeTree() + PARTITION BY toYYYYMMDD(event_time) + ORDER BY id; + """ + node1.query(query_create) + + # gives us 2 partitions with 3 parts in total + node1.query("INSERT INTO t VALUES (1, toDate('2022-09-01'));") + node1.query("INSERT INTO t VALUES (2, toDate('2022-08-29'));") + node1.query("INSERT INTO t VALUES (3, toDate('2022-09-01'));") + + query_number_detached_parts_in_async_metric = """ + SELECT value + FROM system.asynchronous_metrics + WHERE metric LIKE 'NumberOfDetachedParts'; + """ + query_number_detached_by_user_parts_in_async_metric = """ + SELECT value + FROM system.asynchronous_metrics + WHERE metric LIKE 'NumberOfDetachedByUserParts'; + """ + query_count_active_parts = """ + SELECT count(*) FROM system.parts WHERE table = 't' AND active + """ + query_count_detached_parts = """ + SELECT count(*) FROM system.detached_parts WHERE table = 't' + """ + + query_one_partition_name = """ + SELECT name FROM system.parts WHERE table = 't' AND active AND partition = '20220829' + """ + partition_name = node1.query(query_one_partition_name).strip() + + assert 0 == int(node1.query(query_count_detached_parts)) + assert 3 == int(node1.query(query_count_active_parts)) + assert 0 == int(node1.query(query_number_detached_parts_in_async_metric)) + assert 0 == int(node1.query(query_number_detached_by_user_parts_in_async_metric)) + + # detach some parts and wait until asynchronous metrics notice it + node1.query("ALTER TABLE t DETACH PARTITION '20220901';") + + assert 2 == int(node1.query(query_count_detached_parts)) + assert 1 == int(node1.query(query_count_active_parts)) + + assert_eq_with_retry( + node1, + query_number_detached_parts_in_async_metric, + "2\n", + ) + assert 2 == int(node1.query(query_number_detached_by_user_parts_in_async_metric)) + + # detach the rest parts and wait until asynchronous metrics notice it + node1.query("ALTER TABLE t DETACH PARTITION ALL") + + assert 3 == int(node1.query(query_count_detached_parts)) + assert 0 == int(node1.query(query_count_active_parts)) + + assert_eq_with_retry( + node1, + query_number_detached_parts_in_async_metric, + "3\n", + ) + assert 3 == int(node1.query(query_number_detached_by_user_parts_in_async_metric)) + + # inject some data directly and wait until asynchronous metrics notice it + node1.exec_in_container( + [ + "bash", + "-c", + "mkdir /var/lib/clickhouse/data/default/t/detached/unexpected_all_0_0_0", + ] + ) + + assert 4 == int(node1.query(query_count_detached_parts)) + assert 0 == int(node1.query(query_count_active_parts)) + + assert_eq_with_retry( + node1, + query_number_detached_parts_in_async_metric, + "4\n", + ) + assert 3 == int(node1.query(query_number_detached_by_user_parts_in_async_metric)) + + # drop some data directly and wait asynchronous metrics notice it + node1.exec_in_container( + [ + "bash", + "-c", + "rm -rf /var/lib/clickhouse/data/default/t/detached/{}".format( + partition_name + ), + ] + ) + + assert 3 == int(node1.query(query_count_detached_parts)) + assert 0 == int(node1.query(query_count_active_parts)) + + assert_eq_with_retry( + node1, + query_number_detached_parts_in_async_metric, + "3\n", + ) + assert 2 == int(node1.query(query_number_detached_by_user_parts_in_async_metric)) diff --git a/tests/integration/test_grpc_protocol/test.py b/tests/integration/test_grpc_protocol/test.py index 469113cd68e..52c583973d0 100644 --- a/tests/integration/test_grpc_protocol/test.py +++ b/tests/integration/test_grpc_protocol/test.py @@ -744,7 +744,7 @@ def test_opentelemetry_context_propagation(): assert ( node.query( f"SELECT attribute['db.statement'], attribute['clickhouse.tracestate'] FROM system.opentelemetry_span_log " - f"WHERE trace_id='{trace_id}' AND parent_span_id={parent_span_id}" + f"WHERE trace_id='{trace_id}' AND operation_name='query'" ) == "SELECT 1\tsome custom state\n" ) diff --git a/tests/integration/test_ttl_move/test.py b/tests/integration/test_ttl_move/test.py index 49d7ab4f2fc..99978cbf6dc 100644 --- a/tests/integration/test_ttl_move/test.py +++ b/tests/integration/test_ttl_move/test.py @@ -1284,19 +1284,6 @@ def test_materialize_ttl_in_partition(started_cluster, name, engine): def test_alter_multiple_ttls(started_cluster, name, engine, positive): name = unique_table_name(name) - """Copyright 2019, Altinity LTD - -Licensed under the Apache License, Version 2.0 (the "License"); -you may not use this file except in compliance with the License. -You may obtain a copy of the License at - - http://www.apache.org/licenses/LICENSE-2.0 - -Unless required by applicable law or agreed to in writing, software -distributed under the License is distributed on an "AS IS" BASIS, -WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -See the License for the specific language governing permissions and -limitations under the License.""" """Check that when multiple TTL expressions are set and before any parts are inserted the TTL expressions are changed with ALTER command then all old @@ -1664,16 +1651,6 @@ def test_double_move_while_select(started_cluster, name, positive): def test_alter_with_merge_work(started_cluster, name, engine, positive): name = unique_table_name(name) - """Copyright 2019, Altinity LTD -Licensed under the Apache License, Version 2.0 (the "License"); -you may not use this file except in compliance with the License. -You may obtain a copy of the License at - http://www.apache.org/licenses/LICENSE-2.0 -Unless required by applicable law or agreed to in writing, software -distributed under the License is distributed on an "AS IS" BASIS, -WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -See the License for the specific language governing permissions and -limitations under the License.""" """Check that TTL expressions are re-evaluated for existing parts after ALTER command changes TTL expressions and parts are merged. diff --git a/tests/integration/test_ttl_replicated/test.py b/tests/integration/test_ttl_replicated/test.py index bcdb2d25912..cacd9ef0c78 100644 --- a/tests/integration/test_ttl_replicated/test.py +++ b/tests/integration/test_ttl_replicated/test.py @@ -342,16 +342,6 @@ def optimize_with_retry(node, table_name, retry=20): ], ) def test_ttl_alter_delete(started_cluster, name, engine): - """Copyright 2019, Altinity LTD - Licensed under the Apache License, Version 2.0 (the "License"); - you may not use this file except in compliance with the License. - You may obtain a copy of the License at - http://www.apache.org/licenses/LICENSE-2.0 - Unless required by applicable law or agreed to in writing, software - distributed under the License is distributed on an "AS IS" BASIS, - WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - See the License for the specific language governing permissions and - limitations under the License.""" """Check compatibility with old TTL delete expressions to make sure that: * alter modify of column's TTL delete expression works diff --git a/tests/performance/join_set_filter.xml b/tests/performance/join_set_filter.xml new file mode 100644 index 00000000000..7f7804853fc --- /dev/null +++ b/tests/performance/join_set_filter.xml @@ -0,0 +1,45 @@ + + + + table_size + + 100000000 + + + + + + full_sorting_merge + + + + CREATE TABLE t1 (x UInt64, y UInt64) ENGINE = MergeTree ORDER BY y + AS SELECT + sipHash64(number, 't1_x') % {table_size} AS x, + sipHash64(number, 't1_y') % {table_size} AS y + FROM numbers({table_size}) + + + + CREATE TABLE t2 (x UInt64, y UInt64) ENGINE = MergeTree ORDER BY y + AS SELECT + sipHash64(number, 't2_x') % {table_size} AS x, + sipHash64(number, 't2_y') % {table_size} AS y + FROM numbers({table_size}) + + + SELECT * FROM t1 JOIN t2 ON t1.x = t2.x WHERE less(t1.y, 10000) + SELECT * FROM t2 JOIN t1 ON t1.x = t2.x WHERE less(t1.y, 10000) + + SELECT * FROM t1 JOIN t2 ON t1.x = t2.x WHERE greater(t1.y, {table_size} - 10000) + SELECT * FROM t2 JOIN t1 ON t1.x = t2.x WHERE greater(t1.y, {table_size} - 10000) + + SELECT * FROM t1 JOIN t2 ON t1.x = t2.x WHERE t1.y % 100 = 0 + SELECT * FROM t2 JOIN t1 ON t1.x = t2.x WHERE t1.y % 100 = 0 + + SELECT * FROM t1 JOIN t2 ON t1.x = t2.x WHERE t1.y % 1000 = 0 + SELECT * FROM t2 JOIN t1 ON t1.x = t2.x WHERE t1.y % 1000 = 0 + + DROP TABLE IF EXISTS t1 + DROP TABLE IF EXISTS t2 + diff --git a/tests/performance/queries_over_aggregation.xml b/tests/performance/queries_over_aggregation.xml index 2a92ea26819..ceaed61c5bb 100644 --- a/tests/performance/queries_over_aggregation.xml +++ b/tests/performance/queries_over_aggregation.xml @@ -1,4 +1,8 @@ + select sipHash64(number) from numbers(1e7) group by number format Null + select * from (select * from numbers(1e7) group by number) group by number format Null + select * from (select * from numbers(1e7) group by number) order by number format Null + select * from (select * from numbers_mt(1e7) group by number) group by number format Null select * from (select * from numbers_mt(1e7) group by number) order by number format Null select * from (select * from numbers_mt(1e7) group by number) group by number format Null settings max_bytes_before_external_group_by = 1 diff --git a/tests/queries/0_stateless/01064_incremental_streaming_from_2_src_with_feedback.sql b/tests/queries/0_stateless/01064_incremental_streaming_from_2_src_with_feedback.sql index 0bc5fcd1db8..9a439180265 100644 --- a/tests/queries/0_stateless/01064_incremental_streaming_from_2_src_with_feedback.sql +++ b/tests/queries/0_stateless/01064_incremental_streaming_from_2_src_with_feedback.sql @@ -1,4 +1,5 @@ SET joined_subquery_requires_alias = 0; +SET max_threads = 1; -- incremental streaming usecase -- that has sense only if data filling order has guarantees of chronological order diff --git a/tests/queries/0_stateless/01091_num_threads.sql b/tests/queries/0_stateless/01091_num_threads.sql index faeceb0e6d6..0d2a66a8c2e 100644 --- a/tests/queries/0_stateless/01091_num_threads.sql +++ b/tests/queries/0_stateless/01091_num_threads.sql @@ -28,7 +28,7 @@ WITH ORDER BY event_time DESC LIMIT 1 ) AS id -SELECT uniqExact(thread_id) +SELECT uniqExact(thread_id) > 2 FROM system.query_thread_log WHERE (event_date >= (today() - 1)) AND (query_id = id) AND (thread_id != master_thread_id); diff --git a/tests/queries/0_stateless/01281_group_by_limit_memory_tracking.sh b/tests/queries/0_stateless/01281_group_by_limit_memory_tracking.sh index 2f4164ee0d1..7f353b84a80 100755 --- a/tests/queries/0_stateless/01281_group_by_limit_memory_tracking.sh +++ b/tests/queries/0_stateless/01281_group_by_limit_memory_tracking.sh @@ -1,5 +1,5 @@ #!/usr/bin/env bash -# Tags: no-replicated-database, no-parallel, no-fasttest, no-tsan, no-asan, no-random-settings, no-s3-storage +# Tags: no-replicated-database, no-parallel, no-fasttest, no-tsan, no-asan, no-random-settings, no-s3-storage, no-msan # Tag no-fasttest: max_memory_usage_for_user can interfere another queries running concurrently # Regression for MemoryTracker that had been incorrectly accounted diff --git a/tests/queries/0_stateless/01455_opentelemetry_distributed.reference b/tests/queries/0_stateless/01455_opentelemetry_distributed.reference index e70506599ec..d48b3738bc2 100644 --- a/tests/queries/0_stateless/01455_opentelemetry_distributed.reference +++ b/tests/queries/0_stateless/01455_opentelemetry_distributed.reference @@ -15,7 +15,7 @@ {"query":"select 1 format Null\n","query_status":"QueryFinish","tracestate":"another custom state","sorted_by_finish_time":1} {"query":"select 1 format Null\n","query_status":"QueryFinish","tracestate":"another custom state","sorted_by_finish_time":1} {"query":"select * from url('http:\/\/127.0.0.2:8123\/?query=select%201%20format%20Null', CSV, 'a int')","query_status":"QueryFinish","tracestate":"another custom state","sorted_by_finish_time":1} -{"total spans":"3","unique spans":"3","unique non-zero parent spans":"2"} +{"total spans":"3","unique spans":"3","unique non-zero parent spans":"3"} {"initial query spans with proper parent":"1"} {"unique non-empty tracestate values":"1"} ===sampled=== diff --git a/tests/queries/0_stateless/01455_opentelemetry_distributed.sh b/tests/queries/0_stateless/01455_opentelemetry_distributed.sh index 95d99449837..b2b5ae89105 100755 --- a/tests/queries/0_stateless/01455_opentelemetry_distributed.sh +++ b/tests/queries/0_stateless/01455_opentelemetry_distributed.sh @@ -48,21 +48,16 @@ select count(*) "'"'"total spans"'"'", ; -- Also check that the initial query span in ClickHouse has proper parent span. +-- the first span should be child of input trace context +-- the 2nd span should be the 'query' span select count(*) "'"'"initial query spans with proper parent"'"'" - from - (select *, attribute_name, attribute_value - from system.opentelemetry_span_log - array join mapKeys(attribute) as attribute_name, - mapValues(attribute) as attribute_value) o - join system.query_log on query_id = o.attribute_value + from system.opentelemetry_span_log where trace_id = UUIDNumToString(toFixedString(unhex('$trace_id'), 16)) - and current_database = currentDatabase() and operation_name = 'query' - and parent_span_id = reinterpretAsUInt64(unhex('73')) - and o.attribute_name = 'clickhouse.query_id' - and is_initial_query - and type = 'QueryFinish' + and parent_span_id in ( + select span_id from system.opentelemetry_span_log where trace_id = UUIDNumToString(toFixedString(unhex('$trace_id'), 16)) and parent_span_id = reinterpretAsUInt64(unhex('73')) + ) ; -- Check that the tracestate header was propagated. It must have exactly the @@ -136,7 +131,6 @@ ${CLICKHOUSE_CLIENT} -q " select if(2 <= count() and count() <= 18, 'OK', 'Fail') from system.opentelemetry_span_log where operation_name = 'query' - and parent_span_id = 0 -- only account for the initial queries and attribute['clickhouse.query_id'] like '$query_id-%' ; " diff --git a/tests/queries/0_stateless/01524_do_not_merge_across_partitions_select_final.reference b/tests/queries/0_stateless/01524_do_not_merge_across_partitions_select_final.reference index a3f2106cd5f..540137d4887 100644 --- a/tests/queries/0_stateless/01524_do_not_merge_across_partitions_select_final.reference +++ b/tests/queries/0_stateless/01524_do_not_merge_across_partitions_select_final.reference @@ -6,4 +6,4 @@ 2020-01-01 00:00:00 2 1 499999 -5 +18 diff --git a/tests/queries/0_stateless/01524_do_not_merge_across_partitions_select_final.sql b/tests/queries/0_stateless/01524_do_not_merge_across_partitions_select_final.sql index a5423d1a3ff..23678c1abd9 100644 --- a/tests/queries/0_stateless/01524_do_not_merge_across_partitions_select_final.sql +++ b/tests/queries/0_stateless/01524_do_not_merge_across_partitions_select_final.sql @@ -1,7 +1,7 @@ DROP TABLE IF EXISTS select_final; SET do_not_merge_across_partitions_select_final = 1; -SET max_threads = 0; +SET max_threads = 16; CREATE TABLE select_final (t DateTime, x Int32, string String) ENGINE = ReplacingMergeTree() PARTITION BY toYYYYMM(t) ORDER BY (x, t); diff --git a/tests/queries/0_stateless/01655_plan_optimizations_optimize_read_in_window_order.reference b/tests/queries/0_stateless/01655_plan_optimizations_optimize_read_in_window_order.reference index 7fcd29b5faf..00eb03bd5f0 100644 --- a/tests/queries/0_stateless/01655_plan_optimizations_optimize_read_in_window_order.reference +++ b/tests/queries/0_stateless/01655_plan_optimizations_optimize_read_in_window_order.reference @@ -10,3 +10,12 @@ No sorting plan optimize_read_in_window_order=1 Prefix sort description: n ASC, x ASC Result sort description: n ASC, x ASC +Complex ORDER BY + optimize_read_in_window_order=0 +3 3 1 +4 5 2 +5 7 3 + optimize_read_in_window_order=1 +3 3 1 +4 5 2 +5 7 3 diff --git a/tests/queries/0_stateless/01655_plan_optimizations_optimize_read_in_window_order.sh b/tests/queries/0_stateless/01655_plan_optimizations_optimize_read_in_window_order.sh index 418baea8113..328d181fadd 100755 --- a/tests/queries/0_stateless/01655_plan_optimizations_optimize_read_in_window_order.sh +++ b/tests/queries/0_stateless/01655_plan_optimizations_optimize_read_in_window_order.sh @@ -31,6 +31,15 @@ $CLICKHOUSE_CLIENT -q "explain plan actions=1, description=1 select n, sum(x) OV echo ' optimize_read_in_window_order=1' $CLICKHOUSE_CLIENT -q "explain plan actions=1, description=1 select n, sum(x) OVER (ORDER BY n, x ROWS BETWEEN 100 PRECEDING AND CURRENT ROW) from ${name}_n_x SETTINGS optimize_read_in_window_order=1" | grep -i "sort description" +echo 'Complex ORDER BY' +$CLICKHOUSE_CLIENT -q "CREATE TABLE ${name}_complex (unique1 Int32, unique2 Int32, ten Int32) ENGINE=MergeTree ORDER BY tuple() SETTINGS index_granularity = 8192" +$CLICKHOUSE_CLIENT -q "INSERT INTO ${name}_complex VALUES (1, 2, 3), (2, 3, 4), (3, 4, 5)" +echo ' optimize_read_in_window_order=0' +$CLICKHOUSE_CLIENT -q "SELECT ten, sum(unique1) + sum(unique2) AS res, rank() OVER (ORDER BY sum(unique1) + sum(unique2) ASC) AS rank FROM ${name}_complex GROUP BY ten ORDER BY ten ASC SETTINGS optimize_read_in_window_order=0" +echo ' optimize_read_in_window_order=1' +$CLICKHOUSE_CLIENT -q "SELECT ten, sum(unique1) + sum(unique2) AS res, rank() OVER (ORDER BY sum(unique1) + sum(unique2) ASC) AS rank FROM ${name}_complex GROUP BY ten ORDER BY ten ASC SETTINGS optimize_read_in_window_order=1" + $CLICKHOUSE_CLIENT -q "drop table ${name}" $CLICKHOUSE_CLIENT -q "drop table ${name}_n" $CLICKHOUSE_CLIENT -q "drop table ${name}_n_x" +$CLICKHOUSE_CLIENT -q "drop table ${name}_complex" diff --git a/tests/queries/0_stateless/01671_aggregate_function_group_bitmap_data.reference b/tests/queries/0_stateless/01671_aggregate_function_group_bitmap_data.reference index 161f4a6372f..8c3288df670 100644 --- a/tests/queries/0_stateless/01671_aggregate_function_group_bitmap_data.reference +++ b/tests/queries/0_stateless/01671_aggregate_function_group_bitmap_data.reference @@ -1,27 +1,12 @@ 1 50 50 1 0 49 1 50 50 1 0 49 1 50 50 1 0 49 -1 50 51 0 1 51 -1 50 50 1 0 49 -1 50 51 0 1 51 -1 50 50 1 0 49 -1 50 50 1 0 49 -1 50 50 1 0 49 1 50 50 1 0 49 1 50 50 1 0 49 -1 50 51 0 1 51 -1 50 50 1 0 49 -1 50 51 0 1 51 -1 50 50 1 0 49 -1 50 51 0 1 51 -1 50 50 1 0 49 -1 50 51 0 1 51 1 50 50 1 0 49 1 50 50 1 0 49 1 50 50 1 0 49 -1 50 51 0 1 51 1 50 50 1 0 49 -1 50 51 0 1 51 1 50 50 1 0 49 1 50 50 1 0 49 1 50 50 1 0 49 @@ -29,32 +14,47 @@ 1 50 50 1 0 49 1 50 50 1 0 49 1 50 50 1 0 49 -1 50 51 0 1 51 +1 50 50 1 0 49 +1 50 50 1 0 49 +1 50 50 1 0 49 +1 50 50 1 0 49 +1 50 50 1 0 49 +1 50 50 1 0 49 +1 50 50 1 0 49 +1 50 50 1 0 49 +1 50 50 1 0 49 +1 50 50 1 0 49 +1 50 50 1 0 49 +1 50 50 1 0 49 +1 50 50 1 0 49 +1 50 50 1 0 49 +1 50 50 1 0 49 +1 50 50 1 0 49 +1 50 50 1 0 49 +1 50 50 1 0 49 +1 50 50 1 0 49 +1 50 50 1 0 49 +1 50 50 1 0 49 +1 50 50 1 0 49 +1 50 50 1 0 49 1 50 50 1 0 49 1 50 51 0 1 51 -1 50 50 1 0 49 1 50 51 0 1 51 1 50 51 0 1 51 -1 50 50 1 0 49 -1 50 50 1 0 49 -1 50 50 1 0 49 -1 50 51 0 1 51 -1 50 50 1 0 49 -1 50 50 1 0 49 1 50 51 0 1 51 1 50 51 0 1 51 -1 50 50 1 0 49 -1 50 51 0 1 51 -1 50 50 1 0 49 -1 50 50 1 0 49 -1 50 50 1 0 49 -1 50 50 1 0 49 1 50 51 0 1 51 1 50 51 0 1 51 -1 50 50 1 0 49 -1 50 50 1 0 49 1 50 51 0 1 51 -1 50 50 1 0 49 -1 50 50 1 0 49 -1 50 50 1 0 49 +1 50 51 0 1 51 +1 50 51 0 1 51 +1 50 51 0 1 51 +1 50 51 0 1 51 +1 50 51 0 1 51 +1 50 51 0 1 51 +1 50 51 0 1 51 +1 50 51 0 1 51 +1 50 51 0 1 51 +1 50 51 0 1 51 +1 50 51 0 1 51 1 50 51 0 1 51 diff --git a/tests/queries/0_stateless/01671_aggregate_function_group_bitmap_data.sql b/tests/queries/0_stateless/01671_aggregate_function_group_bitmap_data.sql index d70665655ca..a04f40058fd 100644 --- a/tests/queries/0_stateless/01671_aggregate_function_group_bitmap_data.sql +++ b/tests/queries/0_stateless/01671_aggregate_function_group_bitmap_data.sql @@ -52,6 +52,7 @@ ALL LEFT JOIN FROM group_bitmap_data_test WHERE pickup_date = '2019-01-01' GROUP BY city_id -) AS js2 USING (city_id); +) AS js2 USING (city_id) +ORDER BY today_users, before_users, ll_users, old_users, new_users, diff_users; DROP TABLE IF EXISTS group_bitmap_data_test; diff --git a/tests/queries/0_stateless/01710_aggregate_projection_with_grouping_set.reference b/tests/queries/0_stateless/01710_aggregate_projection_with_grouping_set.reference new file mode 100644 index 00000000000..b233507ce6d --- /dev/null +++ b/tests/queries/0_stateless/01710_aggregate_projection_with_grouping_set.reference @@ -0,0 +1,28 @@ +a 2 +a x 1 +a y 1 +b 2 +b x 1 +b y 1 + 4 +a 2 +a x 1 +a y 1 +b 2 +b x 1 +b y 1 + 4 + x 2 + y 2 +a 2 +a x 1 +a y 1 +b 2 +b x 1 +b y 1 +a x 1 +a y 1 +b x 1 +b y 1 + + 4 diff --git a/tests/queries/0_stateless/01710_aggregate_projection_with_grouping_set.sql b/tests/queries/0_stateless/01710_aggregate_projection_with_grouping_set.sql new file mode 100644 index 00000000000..652ce786b5d --- /dev/null +++ b/tests/queries/0_stateless/01710_aggregate_projection_with_grouping_set.sql @@ -0,0 +1,15 @@ +drop table if exists test; + +create table test(dim1 String, dim2 String, projection p1 (select dim1, dim2, count() group by dim1, dim2)) engine MergeTree order by dim1; + +insert into test values ('a', 'x') ('a', 'y') ('b', 'x') ('b', 'y'); + +select dim1, dim2, count() from test group by grouping sets ((dim1, dim2), dim1) order by dim1, dim2, count(); + +select dim1, dim2, count() from test group by dim1, dim2 with rollup order by dim1, dim2, count(); + +select dim1, dim2, count() from test group by dim1, dim2 with cube order by dim1, dim2, count(); + +select dim1, dim2, count() from test group by dim1, dim2 with totals order by dim1, dim2, count(); + +drop table test; diff --git a/tests/queries/0_stateless/01730_distributed_group_by_no_merge_order_by_long.sql b/tests/queries/0_stateless/01730_distributed_group_by_no_merge_order_by_long.sql index 62b578c21d6..3d6a25fe799 100644 --- a/tests/queries/0_stateless/01730_distributed_group_by_no_merge_order_by_long.sql +++ b/tests/queries/0_stateless/01730_distributed_group_by_no_merge_order_by_long.sql @@ -12,7 +12,7 @@ select * from remote('127.{2..11}', view(select * from numbers(1e6))) group by n -- and the query with GROUP BY on remote servers will first do GROUP BY and then send the block, -- so the initiator will first receive all blocks from remotes and only after start merging, -- and will hit the memory limit. -select * from remote('127.{2..11}', view(select * from numbers(1e6))) group by number order by number limit 1e6 settings distributed_group_by_no_merge=2, max_memory_usage='100Mi'; -- { serverError 241 } +select * from remote('127.{2..11}', view(select * from numbers(1e6))) group by number order by number limit 1e6 settings distributed_group_by_no_merge=2, max_memory_usage='100Mi', max_block_size=1e12; -- { serverError 241 } -- with optimize_aggregation_in_order=1 remote servers will produce blocks more frequently, -- since they don't need to wait until the aggregation will be finished, diff --git a/tests/queries/0_stateless/01825_type_json_order_by.reference b/tests/queries/0_stateless/01825_type_json_order_by.reference new file mode 100644 index 00000000000..611d2835127 --- /dev/null +++ b/tests/queries/0_stateless/01825_type_json_order_by.reference @@ -0,0 +1,6 @@ +0 +0 +{"k":"v"} + +{"k":"v"} +{"k":"v"} diff --git a/tests/queries/0_stateless/01825_type_json_order_by.sql b/tests/queries/0_stateless/01825_type_json_order_by.sql new file mode 100644 index 00000000000..1d5768fe0b4 --- /dev/null +++ b/tests/queries/0_stateless/01825_type_json_order_by.sql @@ -0,0 +1,5 @@ +-- Tags: no-fasttest + +SELECT dummy FROM system.one ORDER BY materialize('{"k":"v"}'::JSON); +SELECT dummy FROM system.one ORDER BY materialize('{"k":"v"}'::JSON), dummy; +SELECT materialize('{"k":"v"}'::JSON) SETTINGS extremes = 1; diff --git a/tests/queries/0_stateless/02004_intersect_except_const_column.reference b/tests/queries/0_stateless/02004_intersect_except_const_column.reference new file mode 100644 index 00000000000..6e3081b017d --- /dev/null +++ b/tests/queries/0_stateless/02004_intersect_except_const_column.reference @@ -0,0 +1,85 @@ +-- { echo } +-- Test: crash the server +SELECT 'fooooo' INTERSECT DISTINCT SELECT 'fooooo'; +fooooo +SELECT 'fooooo' EXCEPT ALL SELECT 'fooooo'; +-- Test: intersect return incorrect result for const column +SELECT 1 FROM numbers(10) INTERSECT SELECT 1 FROM numbers(10); +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +SELECT toString(1) FROM numbers(10) INTERSECT SELECT toString(1) FROM numbers(10); +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +SELECT '1' FROM numbers(10) INTERSECT SELECT '1' FROM numbers(10); +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +SELECT 1 FROM numbers(10) INTERSECT DISTINCT SELECT 1 FROM numbers(10); +1 +SELECT toString(1) FROM numbers(10) INTERSECT DISTINCT SELECT toString(1) FROM numbers(10); +1 +SELECT '1' FROM numbers(10) INTERSECT DISTINCT SELECT '1' FROM numbers(10); +1 +-- Test: except return incorrect result for const column +SELECT 2 FROM numbers(10) EXCEPT SELECT 1 FROM numbers(5); +2 +2 +2 +2 +2 +2 +2 +2 +2 +2 +SELECT toString(2) FROM numbers(10) EXCEPT SELECT toString(1) FROM numbers(5); +2 +2 +2 +2 +2 +2 +2 +2 +2 +2 +SELECT '2' FROM numbers(10) EXCEPT SELECT '1' FROM numbers(5); +2 +2 +2 +2 +2 +2 +2 +2 +2 +2 +SELECT 2 FROM numbers(10) EXCEPT DISTINCT SELECT 1 FROM numbers(5); +2 +SELECT toString(2) FROM numbers(10) EXCEPT DISTINCT SELECT toString(1) FROM numbers(5); +2 +SELECT '2' FROM numbers(10) EXCEPT DISTINCT SELECT '1' FROM numbers(5); +2 diff --git a/tests/queries/0_stateless/02004_intersect_except_const_column.sql b/tests/queries/0_stateless/02004_intersect_except_const_column.sql new file mode 100644 index 00000000000..6fabf34d38a --- /dev/null +++ b/tests/queries/0_stateless/02004_intersect_except_const_column.sql @@ -0,0 +1,20 @@ +-- { echo } +-- Test: crash the server +SELECT 'fooooo' INTERSECT DISTINCT SELECT 'fooooo'; +SELECT 'fooooo' EXCEPT ALL SELECT 'fooooo'; + +-- Test: intersect return incorrect result for const column +SELECT 1 FROM numbers(10) INTERSECT SELECT 1 FROM numbers(10); +SELECT toString(1) FROM numbers(10) INTERSECT SELECT toString(1) FROM numbers(10); +SELECT '1' FROM numbers(10) INTERSECT SELECT '1' FROM numbers(10); +SELECT 1 FROM numbers(10) INTERSECT DISTINCT SELECT 1 FROM numbers(10); +SELECT toString(1) FROM numbers(10) INTERSECT DISTINCT SELECT toString(1) FROM numbers(10); +SELECT '1' FROM numbers(10) INTERSECT DISTINCT SELECT '1' FROM numbers(10); + +-- Test: except return incorrect result for const column +SELECT 2 FROM numbers(10) EXCEPT SELECT 1 FROM numbers(5); +SELECT toString(2) FROM numbers(10) EXCEPT SELECT toString(1) FROM numbers(5); +SELECT '2' FROM numbers(10) EXCEPT SELECT '1' FROM numbers(5); +SELECT 2 FROM numbers(10) EXCEPT DISTINCT SELECT 1 FROM numbers(5); +SELECT toString(2) FROM numbers(10) EXCEPT DISTINCT SELECT toString(1) FROM numbers(5); +SELECT '2' FROM numbers(10) EXCEPT DISTINCT SELECT '1' FROM numbers(5); \ No newline at end of file diff --git a/tests/queries/0_stateless/02004_intersect_except_distinct_operators.reference b/tests/queries/0_stateless/02004_intersect_except_distinct_operators.reference new file mode 100644 index 00000000000..49529aa7683 --- /dev/null +++ b/tests/queries/0_stateless/02004_intersect_except_distinct_operators.reference @@ -0,0 +1,133 @@ +-- { echo } + +set intersect_default_mode = 'DISTINCT'; +set except_default_mode = 'DISTINCT'; +select 1 intersect select 1; +1 +select 2 intersect select 1; +select 1 except select 1; +select 2 except select 1; +2 +select 5 from numbers(20) intersect select number from numbers(5, 5); +5 +select number from numbers(10) except select number from numbers(5); +5 +6 +7 +8 +9 +select number, number+10 from numbers(12) except select number+5, number+15 from numbers(10); +0 10 +1 11 +2 12 +3 13 +4 14 +select 1 except select 2 intersect select 1; +1 +select 1 except select 2 intersect select 2; +1 +select 1 intersect select 1 except select 2; +1 +select 1 intersect select 1 except select 1; +select 1 intersect select 1 except select 2 intersect select 1 except select 3 intersect select 1; +1 +select 1 intersect select 1 except select 2 intersect select 1 except select 3 intersect select 2; +1 +select 1 intersect select 1 except select 2 intersect select 1 except select 3 intersect select 2 except select 1; +select number%3 from numbers(10) except select 1; +0 +2 +select number from numbers(100) intersect select number from numbers(20, 60) except select number from numbers(30, 20) except select number from numbers(60, 20); +20 +21 +22 +23 +24 +25 +26 +27 +28 +29 +50 +51 +52 +53 +54 +55 +56 +57 +58 +59 +select * from (select 1 intersect select 1); +1 +with (select number from numbers(10) intersect select 5) as a select a * 10; +50 +with (select 5 except select 1) as a select a except select 5; +with (select number from numbers(10) intersect select 5) as a select a intersect select 1; +with (select number from numbers(10) intersect select 5) as a select a except select 1; +5 +select count() from (select number from numbers(10) except select 5); +9 +select count() from (select number from numbers(1000000) intersect select number from numbers(200000, 600000)); +600000 +select count() from (select number from numbers(100) intersect select number from numbers(20, 60) except select number from numbers(30, 20) except select number from numbers(60, 20)); +20 +select count() from (select number from numbers(100) intersect select number from numbers(20, 60) except select number from numbers(30, 20) except select number from numbers(60, 20) union all select number from numbers(100, 10)); +30 +select count() from (select number from numbers(1000000) intersect select number from numbers(200000, 600000) except select number from numbers(300000, 200000) except select number from numbers(600000, 200000)); +200000 +select count() from (select 1 intersect select 1) limit 100; +1 +select count() from (select 1 except select 2) limit 100; +1 +with (select count() from (select 1 union distinct select 2 except select 1)) as max +select count() from (select 1 union all select max) limit 100; +2 +select 1 union all select 1 intersect select 1; +1 +1 +select 1 union all select 1 intersect select 2; +1 +select * from (select 1 union all select 2 union all select 3 union all select 4 except select 3 union all select 5) order by 1; +1 +2 +4 +5 +select * from (select 1 union all select 2 union all select 3 union all select 4 intersect select 3 union all select 5) order by 1; +1 +2 +3 +5 +select * from (select 1 union all select 2 union all select 3 union all select 4 intersect select 3 union all select 5 except select 1) order by 1; +2 +3 +5 +select 1 intersect (select 1 except select 2); +1 +select 1 union all select 2 except (select 2 except select 1 union all select 1) except select 4; +select 1 intersect select count() from (select 1 except select 2 intersect select 2 union all select 1); +explain syntax select 1 intersect select 1; +SELECT 1 +INTERSECT DISTINCT +SELECT 1 +explain syntax select 1 except select 1; +SELECT 1 +EXCEPT DISTINCT +SELECT 1 +explain syntax select 1 union all select 2 except (select 2 except select 1 union all select 1) except select 4; +SELECT 1 +UNION ALL +SELECT 2 +EXCEPT DISTINCT +SELECT 2 +EXCEPT DISTINCT +SELECT 1 +UNION ALL +SELECT 1 +EXCEPT DISTINCT +SELECT 4 +set limit=1; +select 1 intersect select 1; +1 +(((select 1) intersect select 1)); +1 diff --git a/tests/queries/0_stateless/02004_intersect_except_distinct_operators.sql b/tests/queries/0_stateless/02004_intersect_except_distinct_operators.sql new file mode 100644 index 00000000000..efb48e59f3d --- /dev/null +++ b/tests/queries/0_stateless/02004_intersect_except_distinct_operators.sql @@ -0,0 +1,58 @@ +-- { echo } + +set intersect_default_mode = 'DISTINCT'; +set except_default_mode = 'DISTINCT'; + +select 1 intersect select 1; +select 2 intersect select 1; +select 1 except select 1; +select 2 except select 1; + +select 5 from numbers(20) intersect select number from numbers(5, 5); +select number from numbers(10) except select number from numbers(5); +select number, number+10 from numbers(12) except select number+5, number+15 from numbers(10); + +select 1 except select 2 intersect select 1; +select 1 except select 2 intersect select 2; +select 1 intersect select 1 except select 2; +select 1 intersect select 1 except select 1; +select 1 intersect select 1 except select 2 intersect select 1 except select 3 intersect select 1; +select 1 intersect select 1 except select 2 intersect select 1 except select 3 intersect select 2; +select 1 intersect select 1 except select 2 intersect select 1 except select 3 intersect select 2 except select 1; + +select number%3 from numbers(10) except select 1; +select number from numbers(100) intersect select number from numbers(20, 60) except select number from numbers(30, 20) except select number from numbers(60, 20); + +select * from (select 1 intersect select 1); +with (select number from numbers(10) intersect select 5) as a select a * 10; +with (select 5 except select 1) as a select a except select 5; +with (select number from numbers(10) intersect select 5) as a select a intersect select 1; +with (select number from numbers(10) intersect select 5) as a select a except select 1; +select count() from (select number from numbers(10) except select 5); +select count() from (select number from numbers(1000000) intersect select number from numbers(200000, 600000)); +select count() from (select number from numbers(100) intersect select number from numbers(20, 60) except select number from numbers(30, 20) except select number from numbers(60, 20)); +select count() from (select number from numbers(100) intersect select number from numbers(20, 60) except select number from numbers(30, 20) except select number from numbers(60, 20) union all select number from numbers(100, 10)); +select count() from (select number from numbers(1000000) intersect select number from numbers(200000, 600000) except select number from numbers(300000, 200000) except select number from numbers(600000, 200000)); + +select count() from (select 1 intersect select 1) limit 100; +select count() from (select 1 except select 2) limit 100; +with (select count() from (select 1 union distinct select 2 except select 1)) as max +select count() from (select 1 union all select max) limit 100; + +select 1 union all select 1 intersect select 1; +select 1 union all select 1 intersect select 2; +select * from (select 1 union all select 2 union all select 3 union all select 4 except select 3 union all select 5) order by 1; +select * from (select 1 union all select 2 union all select 3 union all select 4 intersect select 3 union all select 5) order by 1; +select * from (select 1 union all select 2 union all select 3 union all select 4 intersect select 3 union all select 5 except select 1) order by 1; + +select 1 intersect (select 1 except select 2); +select 1 union all select 2 except (select 2 except select 1 union all select 1) except select 4; +select 1 intersect select count() from (select 1 except select 2 intersect select 2 union all select 1); + +explain syntax select 1 intersect select 1; +explain syntax select 1 except select 1; +explain syntax select 1 union all select 2 except (select 2 except select 1 union all select 1) except select 4; + +set limit=1; +select 1 intersect select 1; +(((select 1) intersect select 1)); diff --git a/tests/queries/0_stateless/02004_intersect_except_operators.reference b/tests/queries/0_stateless/02004_intersect_except_operators.reference index a96a6bc7264..0acb550c1e8 100644 --- a/tests/queries/0_stateless/02004_intersect_except_operators.reference +++ b/tests/queries/0_stateless/02004_intersect_except_operators.reference @@ -116,23 +116,23 @@ select 1 union all select 2 except (select 2 except select 1 union all select 1 select 1 intersect select count() from (select 1 except select 2 intersect select 2 union all select 1); explain syntax select 1 intersect select 1; SELECT 1 -INTERSECT +INTERSECT ALL SELECT 1 explain syntax select 1 except select 1; SELECT 1 -EXCEPT +EXCEPT ALL SELECT 1 explain syntax select 1 union all select 2 except (select 2 except select 1 union all select 1) except select 4; SELECT 1 UNION ALL SELECT 2 -EXCEPT +EXCEPT ALL SELECT 2 -EXCEPT +EXCEPT ALL SELECT 1 UNION ALL SELECT 1 -EXCEPT +EXCEPT ALL SELECT 4 set limit=1; select 1 intersect select 1; diff --git a/tests/queries/0_stateless/02131_remove_columns_in_subquery.reference b/tests/queries/0_stateless/02131_remove_columns_in_subquery.reference index d00491fd7e5..6ed281c757a 100644 --- a/tests/queries/0_stateless/02131_remove_columns_in_subquery.reference +++ b/tests/queries/0_stateless/02131_remove_columns_in_subquery.reference @@ -1 +1,2 @@ 1 +1 diff --git a/tests/queries/0_stateless/02131_remove_columns_in_subquery.sql b/tests/queries/0_stateless/02131_remove_columns_in_subquery.sql index f9ca2269aad..c765d989d1d 100644 --- a/tests/queries/0_stateless/02131_remove_columns_in_subquery.sql +++ b/tests/queries/0_stateless/02131_remove_columns_in_subquery.sql @@ -1 +1,2 @@ -select count(1) from (SELECT 1 AS a, count(1) FROM numbers(5)) +select count(1) from (SELECT 1 AS a, count(1) FROM numbers(5)); +select count(1) from (SELECT 1 AS a, count(1) + 1 FROM numbers(5)); \ No newline at end of file diff --git a/tests/queries/0_stateless/02277_full_sort_join_misc.sql b/tests/queries/0_stateless/02277_full_sort_join_misc.sql index b4e3882edaf..4297f532b98 100644 --- a/tests/queries/0_stateless/02277_full_sort_join_misc.sql +++ b/tests/queries/0_stateless/02277_full_sort_join_misc.sql @@ -1,24 +1,24 @@ SET join_algorithm = 'full_sorting_merge'; -SELECT * FROM (SELECT 1 as key) AS t1 JOIN (SELECT 1 as key) t2 ON t1.key = t2.key; +SELECT * FROM (SELECT 1 as key) AS t1 JOIN (SELECT 1 as key) t2 ON t1.key = t2.key ORDER BY key; -SELECT * FROM (SELECT 1 as key) AS t1 JOIN (SELECT 1 as key) t2 USING key; +SELECT * FROM (SELECT 1 as key) AS t1 JOIN (SELECT 1 as key) t2 USING key ORDER BY key; -SELECT * FROM (SELECT 1 :: UInt32 as key) AS t1 FULL JOIN (SELECT 1 :: Nullable(UInt32) as key) t2 USING (key); +SELECT * FROM (SELECT 1 :: UInt32 as key) AS t1 FULL JOIN (SELECT 1 :: Nullable(UInt32) as key) t2 USING (key) ORDER BY key; -SELECT * FROM (SELECT 1 :: UInt32 as key) AS t1 FULL JOIN (SELECT NULL :: Nullable(UInt32) as key) t2 USING (key); +SELECT * FROM (SELECT 1 :: UInt32 as key) AS t1 FULL JOIN (SELECT NULL :: Nullable(UInt32) as key) t2 USING (key) ORDER BY key; -SELECT * FROM (SELECT 1 :: Int32 as key) AS t1 JOIN (SELECT 1 :: UInt32 as key) t2 ON t1.key = t2.key; +SELECT * FROM (SELECT 1 :: Int32 as key) AS t1 JOIN (SELECT 1 :: UInt32 as key) t2 ON t1.key = t2.key ORDER BY key; -SELECT * FROM (SELECT -1 :: Nullable(Int32) as key) AS t1 FULL JOIN (SELECT 4294967295 :: UInt32 as key) t2 ON t1.key = t2.key; +SELECT * FROM (SELECT -1 :: Nullable(Int32) as key) AS t1 FULL JOIN (SELECT 4294967295 :: UInt32 as key) t2 ON t1.key = t2.key ORDER BY key; -SELECT * FROM (SELECT 'a' :: LowCardinality(String) AS key) AS t1 JOIN (SELECT 'a' :: String AS key) AS t2 ON t1.key = t2.key; +SELECT * FROM (SELECT 'a' :: LowCardinality(String) AS key) AS t1 JOIN (SELECT 'a' :: String AS key) AS t2 ON t1.key = t2.key ORDER BY key; -SELECT * FROM (SELECT 'a' :: LowCardinality(Nullable(String)) AS key) AS t1 JOIN (SELECT 'a' :: String AS key) AS t2 ON t1.key = t2.key; +SELECT * FROM (SELECT 'a' :: LowCardinality(Nullable(String)) AS key) AS t1 JOIN (SELECT 'a' :: String AS key) AS t2 ON t1.key = t2.key ORDER BY key; -SELECT * FROM (SELECT 'a' :: LowCardinality(Nullable(String)) AS key) AS t1 JOIN (SELECT 'a' :: Nullable(String) AS key) AS t2 ON t1.key = t2.key; +SELECT * FROM (SELECT 'a' :: LowCardinality(Nullable(String)) AS key) AS t1 JOIN (SELECT 'a' :: Nullable(String) AS key) AS t2 ON t1.key = t2.key ORDER BY key; -SELECT * FROM (SELECT 'a' :: LowCardinality(String) AS key) AS t1 JOIN (SELECT 'a' :: LowCardinality(String) AS key) AS t2 ON t1.key = t2.key; +SELECT * FROM (SELECT 'a' :: LowCardinality(String) AS key) AS t1 JOIN (SELECT 'a' :: LowCardinality(String) AS key) AS t2 ON t1.key = t2.key ORDER BY key; -SELECT 5 == count() FROM (SELECT number as a from numbers(5)) as t1 LEFT JOIN (SELECT number as b from numbers(5) WHERE number > 100) as t2 ON t1.a = t2.b; -SELECT 5 == count() FROM (SELECT number as a from numbers(5) WHERE number > 100) as t1 RIGHT JOIN (SELECT number as b from numbers(5)) as t2 ON t1.a = t2.b; +SELECT 5 == count() FROM (SELECT number as a from numbers(5)) as t1 LEFT JOIN (SELECT number as b from numbers(5) WHERE number > 100) as t2 ON t1.a = t2.b ORDER BY 1; +SELECT 5 == count() FROM (SELECT number as a from numbers(5) WHERE number > 100) as t1 RIGHT JOIN (SELECT number as b from numbers(5)) as t2 ON t1.a = t2.b ORDER BY 1; diff --git a/tests/queries/0_stateless/02337_join_analyze_stuck.reference b/tests/queries/0_stateless/02337_join_analyze_stuck.reference new file mode 100644 index 00000000000..4c8ac5b69c5 --- /dev/null +++ b/tests/queries/0_stateless/02337_join_analyze_stuck.reference @@ -0,0 +1,26 @@ +WITH + x AS + ( + SELECT number + FROM numbers(10) + ), + cross_sales AS + ( + SELECT 1 AS xx + FROM + x, + x AS d1, + x AS d2, + x AS d3, + x AS d4, + x AS d5, + x AS d6, + x AS d7, + x AS d8, + x AS d9 + WHERE x.number = d9.number + ) +SELECT xx +FROM +cross_sales +WHERE xx = 2000 diff --git a/tests/queries/0_stateless/02337_join_analyze_stuck.sql b/tests/queries/0_stateless/02337_join_analyze_stuck.sql new file mode 100644 index 00000000000..62dd0888673 --- /dev/null +++ b/tests/queries/0_stateless/02337_join_analyze_stuck.sql @@ -0,0 +1,13 @@ +-- Tags: long + +-- https://github.com/ClickHouse/ClickHouse/issues/21557 + +EXPLAIN SYNTAX +WITH + x AS ( SELECT number FROM numbers(10) ), + cross_sales AS ( + SELECT 1 AS xx + FROM x, x AS d1, x AS d2, x AS d3, x AS d4, x AS d5, x AS d6, x AS d7, x AS d8, x AS d9 + WHERE x.number = d9.number + ) +SELECT xx FROM cross_sales WHERE xx = 2000; diff --git a/tests/queries/0_stateless/02343_aggregation_pipeline.reference b/tests/queries/0_stateless/02343_aggregation_pipeline.reference index 67bd9c414ba..ec9a394d05d 100644 --- a/tests/queries/0_stateless/02343_aggregation_pipeline.reference +++ b/tests/queries/0_stateless/02343_aggregation_pipeline.reference @@ -1,5 +1,22 @@ -- { echoOn } +explain pipeline select * from (select * from numbers(1e8) group by number) group by number; +(Expression) +ExpressionTransform × 16 + (Aggregating) + Resize 16 → 16 + AggregatingTransform × 16 + StrictResize 16 → 16 + (Expression) + ExpressionTransform × 16 + (Aggregating) + Resize 1 → 16 + AggregatingTransform + (Expression) + ExpressionTransform + (ReadFromStorage) + Limit + Numbers 0 → 1 explain pipeline select * from (select * from numbers_mt(1e8) group by number) group by number; (Expression) ExpressionTransform × 16 diff --git a/tests/queries/0_stateless/02343_aggregation_pipeline.sql b/tests/queries/0_stateless/02343_aggregation_pipeline.sql index d259889b042..85e9fd1be1e 100644 --- a/tests/queries/0_stateless/02343_aggregation_pipeline.sql +++ b/tests/queries/0_stateless/02343_aggregation_pipeline.sql @@ -1,9 +1,12 @@ set max_threads = 16; set prefer_localhost_replica = 1; set optimize_aggregation_in_order = 0; +set max_block_size = 65505; -- { echoOn } +explain pipeline select * from (select * from numbers(1e8) group by number) group by number; + explain pipeline select * from (select * from numbers_mt(1e8) group by number) group by number; explain pipeline select * from (select * from numbers_mt(1e8) group by number) order by number; diff --git a/tests/queries/0_stateless/02346_aggregation_in_order_fixed_prefix.reference b/tests/queries/0_stateless/02346_aggregation_in_order_fixed_prefix.reference new file mode 100644 index 00000000000..fecc2acec7e --- /dev/null +++ b/tests/queries/0_stateless/02346_aggregation_in_order_fixed_prefix.reference @@ -0,0 +1,8 @@ +1 100000 1 +1 100001 1 +1 100002 1 +ReadType: InOrder +100000 1 +100001 1 +100002 1 +ReadType: InOrder diff --git a/tests/queries/0_stateless/02346_aggregation_in_order_fixed_prefix.sh b/tests/queries/0_stateless/02346_aggregation_in_order_fixed_prefix.sh new file mode 100755 index 00000000000..265c7894852 --- /dev/null +++ b/tests/queries/0_stateless/02346_aggregation_in_order_fixed_prefix.sh @@ -0,0 +1,18 @@ +#!/usr/bin/env bash + +CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd) +# shellcheck source=../shell_config.sh +. "$CURDIR"/../shell_config.sh + +$CLICKHOUSE_CLIENT -q "DROP TABLE IF EXISTS test_agg" + +$CLICKHOUSE_CLIENT -q "CREATE TABLE test_agg ( A Int64, B Int64 ) Engine=MergeTree() ORDER BY (A, B)" +$CLICKHOUSE_CLIENT -q "INSERT INTO test_agg SELECT intDiv(number, 1e5), number FROM numbers(1e6)" + +$CLICKHOUSE_CLIENT --optimize_aggregation_in_order 1 -q "SELECT A, B, count() FROM test_agg where A = 1 GROUP BY A, B ORDER BY A, B LIMIT 3" +$CLICKHOUSE_CLIENT --optimize_aggregation_in_order 1 -q "EXPLAIN actions = 1 SELECT A, B, count() FROM test_agg where A = 1 GROUP BY A, B ORDER BY A, B LIMIT 3" | grep -o "ReadType: InOrder" + +$CLICKHOUSE_CLIENT --optimize_aggregation_in_order 1 -q "SELECT B, count() FROM test_agg where A = 1 GROUP BY B ORDER BY B LIMIT 3" +$CLICKHOUSE_CLIENT --optimize_aggregation_in_order 1 -q "EXPLAIN actions = 1 SELECT B, count() FROM test_agg where A = 1 GROUP BY B ORDER BY B LIMIT 3" | grep -o "ReadType: InOrder" + +$CLICKHOUSE_CLIENT -q "DROP TABLE IF EXISTS test_agg" diff --git a/tests/queries/0_stateless/02346_read_in_order_fixed_prefix.python b/tests/queries/0_stateless/02346_read_in_order_fixed_prefix.python new file mode 100644 index 00000000000..399533480a9 --- /dev/null +++ b/tests/queries/0_stateless/02346_read_in_order_fixed_prefix.python @@ -0,0 +1,132 @@ +#!/usr/bin/env python3 +import os +import sys +from itertools import chain, combinations, permutations + +CURDIR = os.path.dirname(os.path.realpath(__file__)) +sys.path.insert(0, os.path.join(CURDIR, "helpers")) + +from pure_http_client import ClickHouseClient + +client = ClickHouseClient() + + +def powerset(iterable): + s = list(iterable) + return chain.from_iterable(combinations(s, r) for r in range(len(s) + 1)) + + +queries = [ + {"optimize": True, "where": [], "order_by": ["a"]}, + {"optimize": True, "where": [], "order_by": ["a", "b"]}, + {"optimize": True, "where": [], "order_by": ["a", "b", "c"]}, + {"optimize": True, "where": [], "order_by": ["a", "b", "c", "d"]}, + {"optimize": True, "where": ["a"], "order_by": ["a"]}, + {"optimize": True, "where": ["a"], "order_by": ["a", "b"]}, + {"optimize": True, "where": ["a"], "order_by": ["b"]}, + {"optimize": True, "where": ["a"], "order_by": ["b", "c"]}, + {"optimize": True, "where": ["b"], "order_by": ["a"]}, + {"optimize": True, "where": ["b"], "order_by": ["a", "c"]}, + {"optimize": False, "where": ["b"], "order_by": ["b", "c"]}, + {"optimize": True, "where": ["c"], "order_by": ["a"]}, + {"optimize": True, "where": ["c"], "order_by": ["a", "b"]}, + {"optimize": True, "where": ["a", "b"], "order_by": ["a"]}, + {"optimize": True, "where": ["a", "b"], "order_by": ["a", "b"]}, + {"optimize": True, "where": ["a", "b"], "order_by": ["a", "c"]}, + {"optimize": True, "where": ["a", "b"], "order_by": ["a", "b", "c"]}, + {"optimize": True, "where": ["a", "b"], "order_by": ["a", "b", "c", "d"]}, + {"optimize": True, "where": ["a", "b"], "order_by": ["b", "c"]}, + {"optimize": True, "where": ["a", "b"], "order_by": ["c", "d"]}, + {"optimize": True, "where": ["a", "c"], "order_by": ["a"]}, + {"optimize": True, "where": ["a", "c"], "order_by": ["a", "b"]}, + {"optimize": True, "where": ["a", "c"], "order_by": ["b", "d"]}, + {"optimize": True, "where": ["a", "c"], "order_by": ["a", "b", "c"]}, + {"optimize": True, "where": ["a", "c"], "order_by": ["b", "c", "d"]}, + {"optimize": True, "where": ["a", "c"], "order_by": ["a", "b", "c", "d"]}, + {"optimize": False, "where": [], "order_by": ["b"]}, + {"optimize": False, "where": [], "order_by": ["b", "a"]}, + {"optimize": False, "where": [], "order_by": ["b", "c"]}, + {"optimize": False, "where": ["a"], "order_by": ["c"]}, + {"optimize": False, "where": ["a"], "order_by": ["c", "b"]}, + {"optimize": False, "where": ["a"], "order_by": ["c", "d"]}, + {"optimize": False, "where": ["c"], "order_by": ["c", "d"]}, + {"optimize": False, "where": ["c"], "order_by": ["b", "c"]}, +] + + +client.query("DROP TABLE IF EXISTS t_fixed_prefix") +client.query( + """ +CREATE TABLE t_fixed_prefix (a UInt32, b UInt32, c UInt32, d UInt32, e UInt32) +ENGINE = MergeTree ORDER BY (a, b, c, d)""" +) + +client.query("SYSTEM STOP MERGES t_fixed_prefix") + +# create several parts +for _ in range(4): + client.query( + "INSERT INTO t_fixed_prefix SELECT number % 2, number % 10, number % 100, number % 1000, number FROM numbers(25000)" + ) + + +def check_query( + fixed_columns, order_by_columns, should_be_optimized, should_use_finish_sorting +): + where_clause = " AND ".join([f"{c} = 1" for c in fixed_columns]) + order_by_clause = ", ".join(order_by_columns) + + query = "SELECT {} FROM t_fixed_prefix".format(order_by_clause) + if len(where_clause) != 0: + query += " WHERE " + where_clause + + if len(order_by_clause) != 0: + query += " ORDER BY " + order_by_clause + + query += " SETTINGS optimize_read_in_order = {}" + + res_optimized = client.query(query.format(1)) + res_not_optimized = client.query(query.format(0)) + + if res_optimized != res_not_optimized: + print("Failed query {}. Result of queries mismatched".format(query)) + exit(1) + + res_explain = client.query("EXPLAIN PIPELINE {}".format(query.format(1))) + + is_optimized = "MergeSortingTransform" not in res_explain + uses_finish_sorting = "FinishSortingTransform" in res_explain + + if ( + is_optimized != should_be_optimized + or uses_finish_sorting != should_use_finish_sorting + ): + print( + """ +Wrong query pipeline is built for query {}: +{} +Should be optimized: {}. +Is optimized: {}. +Should use FinishSortingTransform: {}. +Uses FinishSortingTransform: {} +""".format( + query.format(1), + res_explain, + should_be_optimized, + is_optimized, + should_use_finish_sorting, + uses_finish_sorting, + ) + ) + exit(1) + + +for query in queries: + check_query(query["where"], query["order_by"], query["optimize"], False) + check_query(query["where"], query["order_by"] + ["e"], query["optimize"], query["optimize"]) + + where_columns = [f"bitNot({col})" for col in query["where"]] + check_query(where_columns, query["order_by"], query["optimize"], False) + check_query(where_columns, query["order_by"] + ["e"], query["optimize"], query["optimize"]) + +print("OK") diff --git a/tests/queries/0_stateless/02346_read_in_order_fixed_prefix.reference b/tests/queries/0_stateless/02346_read_in_order_fixed_prefix.reference new file mode 100644 index 00000000000..d86bac9de59 --- /dev/null +++ b/tests/queries/0_stateless/02346_read_in_order_fixed_prefix.reference @@ -0,0 +1 @@ +OK diff --git a/tests/queries/0_stateless/02346_read_in_order_fixed_prefix.sh b/tests/queries/0_stateless/02346_read_in_order_fixed_prefix.sh new file mode 100755 index 00000000000..e1a393af0c5 --- /dev/null +++ b/tests/queries/0_stateless/02346_read_in_order_fixed_prefix.sh @@ -0,0 +1,10 @@ +#!/usr/bin/env bash +# Tags: no-fasttest, long +# Tag no-fasttest: Require python libraries like scipy, pandas and numpy + +CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd) +# shellcheck source=../shell_config.sh +. "$CURDIR"/../shell_config.sh + +# We should have correct env vars from shell_config.sh to run this test +python3 "$CURDIR"/02346_read_in_order_fixed_prefix.python diff --git a/tests/queries/0_stateless/02355_control_block_size_in_aggregator.reference b/tests/queries/0_stateless/02355_control_block_size_in_aggregator.reference new file mode 100644 index 00000000000..d00491fd7e5 --- /dev/null +++ b/tests/queries/0_stateless/02355_control_block_size_in_aggregator.reference @@ -0,0 +1 @@ +1 diff --git a/tests/queries/0_stateless/02355_control_block_size_in_aggregator.sql b/tests/queries/0_stateless/02355_control_block_size_in_aggregator.sql new file mode 100644 index 00000000000..b4754c6d6fe --- /dev/null +++ b/tests/queries/0_stateless/02355_control_block_size_in_aggregator.sql @@ -0,0 +1,9 @@ +SET max_block_size = 4213; + +SELECT DISTINCT (blockSize() <= 4213) +FROM +( + SELECT number + FROM numbers(100000) + GROUP BY number +); diff --git a/tests/queries/0_stateless/02377_extend_protocol_with_query_parameters.reference b/tests/queries/0_stateless/02377_extend_protocol_with_query_parameters.reference index f46cdb6e5e3..1da5cd0b7b3 100644 --- a/tests/queries/0_stateless/02377_extend_protocol_with_query_parameters.reference +++ b/tests/queries/0_stateless/02377_extend_protocol_with_query_parameters.reference @@ -7,3 +7,10 @@ UInt64 String DateTime Map(UUID, Array(Float32)) 13 str 2022-08-04 18:30:53 {'10':[11,12],'13':[14,15]} 1 1 +_CAST(42, \'Int64\') Int64 +_CAST([1, 2, 3], \'Array(UInt8)\') Array(UInt8) +_CAST(((\'abc\', 22), (\'def\', 33)), \'Map(String, UInt8)\') Map(String, UInt8) +_CAST([[4, 5, 6], [7], [8, 9]], \'Array(Array(UInt8))\') Array(Array(UInt8)) +_CAST(((10, [11, 12]), (13, [14, 15])), \'Map(UInt8, Array(UInt8))\') Map(UInt8, Array(UInt8)) +_CAST(((\'ghj\', ((\'klm\', [16, 17]))), (\'nop\', ((\'rst\', [18])))), \'Map(String, Map(String, Array(UInt8)))\') Map(String, Map(String, Array(UInt8))) +a Int8 diff --git a/tests/queries/0_stateless/02377_extend_protocol_with_query_parameters.sh b/tests/queries/0_stateless/02377_extend_protocol_with_query_parameters.sh index 335af1bb6e6..e61dc337d2a 100755 --- a/tests/queries/0_stateless/02377_extend_protocol_with_query_parameters.sh +++ b/tests/queries/0_stateless/02377_extend_protocol_with_query_parameters.sh @@ -68,13 +68,27 @@ $CLICKHOUSE_CLIENT -n -q "select {n: UInt8} -- { serverError 456 }" $CLICKHOUSE_CLIENT -n -q "set param_n = 12; set param_n = 13; select {n: UInt8}" -# but multiple different parameters could be defined within each session +# multiple different parameters could be defined within each session $CLICKHOUSE_CLIENT -n -q " set param_a = 13, param_b = 'str'; set param_c = '2022-08-04 18:30:53'; set param_d = '{\'10\': [11, 12], \'13\': [14, 15]}'; select {a: UInt32}, {b: String}, {c: DateTime}, {d: Map(String, Array(UInt8))}" + # empty parameter name is not allowed $CLICKHOUSE_CLIENT --param_="" -q "select 1" 2>&1 | grep -c 'Code: 36' $CLICKHOUSE_CLIENT -q "set param_ = ''" 2>&1 | grep -c 'Code: 36' + + +# parameters are also supported for DESCRIBE TABLE queries +$CLICKHOUSE_CLIENT \ + --param_id="42" \ + --param_arr="[1, 2, 3]" \ + --param_map="{'abc': 22, 'def': 33}" \ + --param_mul_arr="[[4, 5, 6], [7], [8, 9]]" \ + --param_map_arr="{10: [11, 12], 13: [14, 15]}" \ + --param_map_map_arr="{'ghj': {'klm': [16, 17]}, 'nop': {'rst': [18]}}" \ + -q "describe table(select {id: Int64}, {arr: Array(UInt8)}, {map: Map(String, UInt8)}, {mul_arr: Array(Array(UInt8))}, {map_arr: Map(UInt8, Array(UInt8))}, {map_map_arr: Map(String, Map(String, Array(UInt8)))})" + +$CLICKHOUSE_CLIENT --param_p=42 -q "describe table (select * from (select {p:Int8} as a group by a) order by a)" diff --git a/tests/queries/0_stateless/02381_intersect_except_const_column.reference b/tests/queries/0_stateless/02381_intersect_except_const_column.reference deleted file mode 100644 index 290835b412e..00000000000 --- a/tests/queries/0_stateless/02381_intersect_except_const_column.reference +++ /dev/null @@ -1,61 +0,0 @@ -fooooo -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2 diff --git a/tests/queries/0_stateless/02381_intersect_except_const_column.sql b/tests/queries/0_stateless/02381_intersect_except_const_column.sql deleted file mode 100644 index b10f913dd1e..00000000000 --- a/tests/queries/0_stateless/02381_intersect_except_const_column.sql +++ /dev/null @@ -1,13 +0,0 @@ --- Test: crash the server -SELECT 'fooooo' INTERSECT SELECT 'fooooo'; -SELECT 'fooooo' EXCEPT SELECT 'fooooo'; - --- Test: intersect return incorrect result for const column -SELECT 1 FROM numbers(10) INTERSECT SELECT 1 FROM numbers(10); -SELECT toString(1) FROM numbers(10) INTERSECT SELECT toString(1) FROM numbers(10); -SELECT '1' FROM numbers(10) INTERSECT SELECT '1' FROM numbers(10); - --- Test: except return incorrect result for const column -SELECT 2 FROM numbers(10) EXCEPT SELECT 1 FROM numbers(5); -SELECT toString(2) FROM numbers(10) EXCEPT SELECT toString(1) FROM numbers(5); -SELECT '2' FROM numbers(10) EXCEPT SELECT '1' FROM numbers(5); \ No newline at end of file diff --git a/tests/queries/0_stateless/02381_join_dup_columns_in_plan.sql b/tests/queries/0_stateless/02381_join_dup_columns_in_plan.sql index a418a06803f..4ed6d965292 100644 --- a/tests/queries/0_stateless/02381_join_dup_columns_in_plan.sql +++ b/tests/queries/0_stateless/02381_join_dup_columns_in_plan.sql @@ -7,6 +7,8 @@ USING (key); SET join_algorithm = 'full_sorting_merge'; +SET max_rows_in_set_to_optimize_join = 0; + EXPLAIN actions=0, description=0, header=1 SELECT * FROM ( SELECT 'key2' AS key ) AS s1 JOIN ( SELECT 'key1' AS key, '1' AS value UNION ALL SELECT 'key2' AS key, '1' AS value ) AS s2 diff --git a/tests/queries/0_stateless/02382_join_and_filtering_set.reference b/tests/queries/0_stateless/02382_join_and_filtering_set.reference new file mode 100644 index 00000000000..58c8ccca6a2 --- /dev/null +++ b/tests/queries/0_stateless/02382_join_and_filtering_set.reference @@ -0,0 +1,7 @@ +106 +46 +42 +51 +42 +24 +10 diff --git a/tests/queries/0_stateless/02382_join_and_filtering_set.sql b/tests/queries/0_stateless/02382_join_and_filtering_set.sql new file mode 100644 index 00000000000..4b425f22c87 --- /dev/null +++ b/tests/queries/0_stateless/02382_join_and_filtering_set.sql @@ -0,0 +1,20 @@ +DROP TABLE IF EXISTS t1; +DROP TABLE IF EXISTS t2; + +CREATE TABLE t1 (x UInt64, y UInt64) ENGINE = MergeTree ORDER BY y +AS SELECT sipHash64(number, 't1_x') % 100 AS x, sipHash64(number, 't1_y') % 100 AS y FROM numbers(100); + +CREATE TABLE t2 (x UInt64, y UInt64) ENGINE = MergeTree ORDER BY y +AS SELECT sipHash64(number, 't2_x') % 100 AS x, sipHash64(number, 't2_y') % 100 AS y FROM numbers(100); + +SET max_rows_in_set_to_optimize_join = 1000; +SET join_algorithm = 'full_sorting_merge'; + +-- different combinations of conditions on key/attribute columns for the left/right tables +SELECT count() FROM t1 JOIN t2 ON t1.x = t2.x; +SELECT count() FROM t1 JOIN t2 ON t1.x = t2.x WHERE t1.y % 2 == 0; +SELECT count() FROM t1 JOIN t2 ON t1.x = t2.x WHERE t1.x % 2 == 0; +SELECT count() FROM t1 JOIN t2 ON t1.x = t2.x WHERE t2.y % 2 == 0; +SELECT count() FROM t1 JOIN t2 ON t1.x = t2.x WHERE t2.x % 2 == 0; +SELECT count() FROM t1 JOIN t2 ON t1.x = t2.x WHERE t1.y % 2 == 0 AND t2.y % 2 == 0; +SELECT count() FROM t1 JOIN t2 ON t1.x = t2.x WHERE t1.x % 2 == 0 AND t2.x % 2 == 0 AND t1.y % 2 == 0 AND t2.y % 2 == 0; diff --git a/tests/queries/0_stateless/02383_join_and_filtering_set.reference b/tests/queries/0_stateless/02383_join_and_filtering_set.reference new file mode 100644 index 00000000000..2ad282ca07f --- /dev/null +++ b/tests/queries/0_stateless/02383_join_and_filtering_set.reference @@ -0,0 +1,10 @@ +Ok +Ok +Ok +Ok +Ok +Ok +Ok +Ok +Ok +Ok diff --git a/tests/queries/0_stateless/02383_join_and_filtering_set.sh b/tests/queries/0_stateless/02383_join_and_filtering_set.sh new file mode 100755 index 00000000000..3356be58ff7 --- /dev/null +++ b/tests/queries/0_stateless/02383_join_and_filtering_set.sh @@ -0,0 +1,55 @@ +#!/usr/bin/env bash +# Tags: no-asan,no-msan,no-tsan,no-ubsan +# +# Test doesn't run complex queries, just test the logic of setting, so no need to run with different builds. +# Also, we run similar queries in 02382_join_and_filtering_set.sql which is enabled for these builds. +# + +CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd) +# shellcheck source=../shell_config.sh +. "$CURDIR"/../shell_config.sh + +$CLICKHOUSE_CLIENT -mn -q """ +CREATE TABLE t1 (x UInt64, y UInt64) ENGINE = MergeTree ORDER BY y +AS SELECT sipHash64(number, 't1_x') % 100 AS x, sipHash64(number, 't1_y') % 100 AS y FROM numbers(100); + +CREATE TABLE t2 (x UInt64, y UInt64) ENGINE = MergeTree ORDER BY y +AS SELECT sipHash64(number, 't2_x') % 100 AS x, sipHash64(number, 't2_y') % 100 AS y FROM numbers(100); +""" + +# Arguments: +# - value of max_rows_in_set_to_optimize_join +# - join kind +# - expected number of steps in plan +# - expected number of steps in pipeline +function test() { + +PARAM_VALUE=$1 +JOIN_KIND=${2:-} + +EXPECTED_PLAN_STEPS=$3 +RES=$( + $CLICKHOUSE_CLIENT --max_rows_in_set_to_optimize_join=${PARAM_VALUE} --join_algorithm='full_sorting_merge' \ + -q "EXPLAIN PLAN SELECT count() FROM t1 ${JOIN_KIND} JOIN t2 ON t1.x = t2.x" | grep -o 'CreateSetAndFilterOnTheFlyStep' | wc -l +) +[ "$RES" -eq "$EXPECTED_PLAN_STEPS" ] && echo "Ok" || echo "Fail: $RES != $EXPECTED_PLAN_STEPS" + +EXPECTED_PIPELINE_STEPS=$4 +RES=$( + $CLICKHOUSE_CLIENT --max_rows_in_set_to_optimize_join=${PARAM_VALUE} --join_algorithm='full_sorting_merge' \ + -q "EXPLAIN PIPELINE SELECT count() FROM t1 ${JOIN_KIND} JOIN t2 ON t1.x = t2.x" \ + | grep -o -e ReadHeadBalancedProcessor -e FilterBySetOnTheFlyTransform -e CreatingSetsOnTheFlyTransform | wc -l +) +[ "$RES" -eq "$EXPECTED_PIPELINE_STEPS" ] && echo "Ok" || echo "Fail: $RES != $EXPECTED_PIPELINE_STEPS" + +} + +test 1000 '' 2 6 + +# no filtering for left/right side +test 1000 'LEFT' 2 5 +test 1000 'RIGHT' 2 5 + +# when disabled no extra steps should be created +test 1000 'FULL' 0 0 +test 0 '' 0 0 diff --git a/tests/queries/0_stateless/02400_memory_accounting_on_error.sql b/tests/queries/0_stateless/02400_memory_accounting_on_error.sql index 32046e854dd..f80c372b81c 100644 --- a/tests/queries/0_stateless/02400_memory_accounting_on_error.sql +++ b/tests/queries/0_stateless/02400_memory_accounting_on_error.sql @@ -1,4 +1,4 @@ -- max_block_size to avoid randomization --- SELECT * FROM generateRandom('i Array(Int8)', 1, 1, 1048577) LIMIT 65536 SETTINGS max_memory_usage='1Gi', max_block_size=65505, log_queries=1; -- { serverError MEMORY_LIMIT_EXCEEDED } --- SYSTEM FLUSH LOGS; --- SELECT * FROM system.query_log WHERE event_date >= yesterday() AND current_database = currentDatabase() AND memory_usage > 100e6; +SELECT * FROM generateRandom('i Array(Int8)', 1, 1, 1048577) LIMIT 65536 SETTINGS max_memory_usage='1Gi', max_block_size=65505, log_queries=1; -- { serverError MEMORY_LIMIT_EXCEEDED } +SYSTEM FLUSH LOGS; +SELECT * FROM system.query_log WHERE event_date >= yesterday() AND current_database = currentDatabase() AND memory_usage > 100e6 FORMAT JSONEachRow; diff --git a/tests/queries/0_stateless/02402_merge_engine_with_view.reference b/tests/queries/0_stateless/02402_merge_engine_with_view.reference index 4a869286f82..6564c244b87 100644 --- a/tests/queries/0_stateless/02402_merge_engine_with_view.reference +++ b/tests/queries/0_stateless/02402_merge_engine_with_view.reference @@ -4,3 +4,4 @@ 3 4 4 +1 diff --git a/tests/queries/0_stateless/02402_merge_engine_with_view.sql b/tests/queries/0_stateless/02402_merge_engine_with_view.sql index 613f76e24c9..64822784845 100644 --- a/tests/queries/0_stateless/02402_merge_engine_with_view.sql +++ b/tests/queries/0_stateless/02402_merge_engine_with_view.sql @@ -1,4 +1,4 @@ - +-- #40014 CREATE TABLE m0 (id UInt64) ENGINE=MergeTree ORDER BY id SETTINGS index_granularity = 1; INSERT INTO m0 SELECT number FROM numbers(10); CREATE TABLE m1 (id UInt64, s String) ENGINE=MergeTree ORDER BY id SETTINGS index_granularity = 1; @@ -7,4 +7,8 @@ CREATE VIEW m1v AS SELECT id FROM m1; CREATE TABLE m2 (id UInt64) ENGINE=Merge(currentDatabase(),'m0|m1v'); -SELECT * FROM m2 WHERE id > 1 AND id < 5 ORDER BY id SETTINGS force_primary_key=1, max_bytes_to_read=64; \ No newline at end of file +SELECT * FROM m2 WHERE id > 1 AND id < 5 ORDER BY id SETTINGS force_primary_key=1, max_bytes_to_read=64; + +-- #40706 +CREATE VIEW v AS SELECT 1; +SELECT 1 FROM merge(currentDatabase(), '^v$'); \ No newline at end of file diff --git a/tests/queries/0_stateless/02416_in_set_same_ast_diff_columns.reference b/tests/queries/0_stateless/02416_in_set_same_ast_diff_columns.reference new file mode 100644 index 00000000000..d00491fd7e5 --- /dev/null +++ b/tests/queries/0_stateless/02416_in_set_same_ast_diff_columns.reference @@ -0,0 +1 @@ +1 diff --git a/tests/queries/0_stateless/02416_in_set_same_ast_diff_columns.sql b/tests/queries/0_stateless/02416_in_set_same_ast_diff_columns.sql new file mode 100644 index 00000000000..c3475f37e7e --- /dev/null +++ b/tests/queries/0_stateless/02416_in_set_same_ast_diff_columns.sql @@ -0,0 +1,3 @@ +CREATE TABLE set_crash (key1 Int32, id1 Int64, c1 Int64) ENGINE = MergeTree PARTITION BY id1 ORDER BY key1; +INSERT INTO set_crash VALUES (-1, 1, 0); +SELECT 1 in (-1, 1) FROM set_crash WHERE (key1, id1) in (-1, 1); diff --git a/tests/queries/0_stateless/02416_json_object_inference.reference b/tests/queries/0_stateless/02416_json_object_inference.reference new file mode 100644 index 00000000000..01537523906 --- /dev/null +++ b/tests/queries/0_stateless/02416_json_object_inference.reference @@ -0,0 +1 @@ +a Object(Nullable(\'json\')) diff --git a/tests/queries/0_stateless/02416_json_object_inference.sql b/tests/queries/0_stateless/02416_json_object_inference.sql new file mode 100644 index 00000000000..b861468a08a --- /dev/null +++ b/tests/queries/0_stateless/02416_json_object_inference.sql @@ -0,0 +1,2 @@ +-- Tags: no-fasttest +desc format(JSONEachRow, '{"a" : {"b" : {"c" : 1, "d" : "str"}}}'); diff --git a/tests/queries/0_stateless/02417_repeat_input_commands.expect b/tests/queries/0_stateless/02417_repeat_input_commands.expect new file mode 100755 index 00000000000..119aac68645 --- /dev/null +++ b/tests/queries/0_stateless/02417_repeat_input_commands.expect @@ -0,0 +1,81 @@ +#!/usr/bin/expect -f + +set basedir [file dirname $argv0] +set basename [file tail $argv0] +exp_internal -f $env(CLICKHOUSE_TMP)/$basename.debuglog 0 + +log_user 0 +set timeout 10 +match_max 100000 + +expect_after { + # Do not ignore eof from expect + eof { exp_continue } + # A default timeout action is to do nothing, change it to fail + timeout { exit 1 } +} + +spawn bash -c "source $basedir/../shell_config.sh ; \$CLICKHOUSE_CLIENT_BINARY \$CLICKHOUSE_CLIENT_OPT --disable_suggestion" +expect ":) " + +# ----------------------------------------- +# test . and / commands prior to the first query + +send -- ".\r" +expect "Empty query" +expect ":) " + +send -- "/\r" +expect "Empty query" +expect ":) " + +# ----------------------------------------- +# test . and / commands after first query + +send -- "SELECT 123\r" +expect "│ 123 │" +expect "1 row in set." +expect ":) " + +send -- ".\r" +expect "│ 123 │" +expect "1 row in set." +expect ":) " + +# test input of . more than once in a row +send -- ".\r" +expect "│ 123 │" +expect "1 row in set." +expect ":) " + +send -- "/\r" +expect "│ 123 │" +expect "1 row in set." +expect ":) " + +# test input of / more than once in a row +send -- "/\r" +expect "│ 123 │" +expect "1 row in set." +expect ":) " + +# ----------------------------------------- +# test . and / commands after another query + +send -- "SELECT 321\r" +expect "│ 321 │" +expect "1 row in set." +expect ":) " + +send -- ".\r" +expect "│ 321 │" +expect "1 row in set." +expect ":) " + +send -- "/\r" +expect "│ 321 │" +expect "1 row in set." +expect ":) " + +send -- "quit\r" +expect eof diff --git a/tests/queries/0_stateless/02417_repeat_input_commands.reference b/tests/queries/0_stateless/02417_repeat_input_commands.reference new file mode 100644 index 00000000000..e69de29bb2d diff --git a/utils/check-style/check-typos b/utils/check-style/check-typos index 3819b6785d7..9194a9464a7 100755 --- a/utils/check-style/check-typos +++ b/utils/check-style/check-typos @@ -5,7 +5,7 @@ ROOT_PATH=$(git rev-parse --show-toplevel) codespell \ - --skip "*generated*,*gperf*,*.bin,*.mrk*,*.idx,checksums.txt,*.dat,*.pyc,*.kate-swp,*obfuscateQueries.cpp,d3-*.js,*.min.js,${ROOT_PATH}/utils/check-style/aspell-ignore" \ + --skip "*generated*,*gperf*,*.bin,*.mrk*,*.idx,checksums.txt,*.dat,*.pyc,*.kate-swp,*obfuscateQueries.cpp,d3-*.js,*.min.js,*.sum,${ROOT_PATH}/utils/check-style/aspell-ignore" \ --ignore-words "${ROOT_PATH}/utils/check-style/codespell-ignore-words.list" \ --exclude-file "${ROOT_PATH}/utils/check-style/codespell-ignore-lines.list" \ --quiet-level 2 \ diff --git a/utils/check-style/codespell-ignore-words.list b/utils/check-style/codespell-ignore-words.list index 062e8a1622b..fc021920041 100644 --- a/utils/check-style/codespell-ignore-words.list +++ b/utils/check-style/codespell-ignore-words.list @@ -16,3 +16,8 @@ ot te fo ba +ro +rightt +iiterator +hastable +nam diff --git a/utils/self-extracting-executable/decompressor.cpp b/utils/self-extracting-executable/decompressor.cpp index 8d8d137a2ac..ad4cd5b2f63 100644 --- a/utils/self-extracting-executable/decompressor.cpp +++ b/utils/self-extracting-executable/decompressor.cpp @@ -13,6 +13,8 @@ #include #include #include +#include +#include #if (defined(OS_DARWIN) || defined(OS_FREEBSD)) && defined(__GNUC__) # include @@ -359,6 +361,31 @@ int decompressFiles(int input_fd, char * path, char * name, bool & have_compress #endif +uint32_t getInode(const char * self) +{ + std::ifstream maps("/proc/self/maps"); + if (maps.fail()) + { + perror("open maps"); + return 0; + } + + /// Record example for /proc/self/maps: + /// address perms offset device inode pathname + /// 561a247de000-561a247e0000 r--p 00000000 103:01 1564 /usr/bin/cat + /// see "man 5 proc" + for (std::string line; std::getline(maps, line);) + { + std::stringstream ss(line); // STYLE_CHECK_ALLOW_STD_STRING_STREAM + std::string addr, mode, offset, id, path; + uint32_t inode = 0; + if (ss >> addr >> mode >> offset >> id >> inode >> path && path == self) + return inode; + } + + return 0; +} + int main(int/* argc*/, char* argv[]) { char self[4096] = {0}; @@ -382,6 +409,58 @@ int main(int/* argc*/, char* argv[]) else name = file_path; + /// get inode of this executable + uint32_t inode = getInode(self); + if (inode == 0) + { + std::cerr << "Unable to obtain inode." << std::endl; + return 1; + } + + std::stringstream lock_path; // STYLE_CHECK_ALLOW_STD_STRING_STREAM + lock_path << "/tmp/" << name << ".decompression." << inode << ".lock"; + int lock = open(lock_path.str().c_str(), O_CREAT | O_RDWR, 0666); + if (lock < 0) + { + perror("lock open"); + return 1; + } + + /// lock file should be closed on exec call + fcntl(lock, F_SETFD, FD_CLOEXEC); + + if (lockf(lock, F_LOCK, 0)) + { + perror("lockf"); + return 1; + } + + struct stat input_info; + if (0 != stat(self, &input_info)) + { + perror("stat"); + return 1; + } + + /// if decompression was performed by another process since this copy was started + /// then file referred by path "self" is already pointing to different inode + if (input_info.st_ino != inode) + { + struct stat lock_info; + if (0 != fstat(lock, &lock_info)) + { + perror("fstat lock"); + return 1; + } + + /// size 1 of lock file indicates that another decompressor has found active executable + if (lock_info.st_size == 1) + execv(self, argv); + + printf("No target executable - decompression only was performed.\n"); + return 0; + } + int input_fd = open(self, O_RDONLY); if (input_fd == -1) { @@ -443,6 +522,10 @@ int main(int/* argc*/, char* argv[]) if (has_exec) { + /// write one byte to the lock in case other copies of compressed are running to indicate that + /// execution should be performed + write(lock, "1", 1); + execv(self, argv); /// This part of code will be reached only if error happened @@ -450,6 +533,9 @@ int main(int/* argc*/, char* argv[]) return 1; } + /// since inodes can be reused - it's a precaution if lock file already exists and have size of 1 + ftruncate(lock, 0); + printf("No target executable - decompression only was performed.\n"); } }