Merge pull request #50934 from azat/tests/fix-query_log

Fix tests sanity checks and avoid dropping system.query_log table

.yamllint

@@ -6,8 +6,10 @@ rules:
         level: warning
         indent-sequences: consistent
     line-length:
-        # there are some bash -c "", so this is OK
+        # there are:
-        max: 300
+        # - bash -c "", so this is OK
+        # - yaml in tests
+        max: 1000
         level: warning
     comments:
         min-spaces-from-content: 1

tests/queries/0_stateless/02483_elapsed_time.sh

@@ -25,7 +25,7 @@ QUERY_ID="${CLICKHOUSE_DATABASE}_$(date +%s)_02883_q1"
 ${CLICKHOUSE_CLIENT} -m --query "$EXCEPTION_BEFORE_START_QUERY" --query_id="$QUERY_ID" >/dev/null 2>&1
 
 ${CLICKHOUSE_CLIENT} --query "SYSTEM FLUSH LOGS"
-${CLICKHOUSE_CLIENT} --query "SELECT type == 'ExceptionBeforeStart' as expected_type, query_duration_ms <= 1000 as elapsed_more_than_one_second FROM system.query_log WHERE query_id='$QUERY_ID'"
+${CLICKHOUSE_CLIENT} --query "SELECT type == 'ExceptionBeforeStart' as expected_type, query_duration_ms <= 1000 as elapsed_more_than_one_second FROM system.query_log WHERE current_database = '$CLICKHOUSE_DATABASE' AND query_id='$QUERY_ID'"
 
 # Now we test with a query that will take 1+ seconds. The CLI should show that as part of the output format
 OK_QUERY_JSON="

tests/queries/0_stateless/02494_query_cache_case_agnostic_matching.sql

@@ -1,9 +1,8 @@
 -- Tags: no-parallel
 -- Tag no-parallel: Messes with internal cache
 
--- Start with empty query cache (QC) and query log
+-- Start with empty query cache (QC)
 SYSTEM DROP QUERY CACHE;
-DROP TABLE system.query_log SYNC;
 
 -- Insert an entry into the query cache.
 SELECT 1 SETTINGS use_query_cache = true;
@@ -22,6 +21,7 @@ SYSTEM FLUSH LOGS;
 SELECT ProfileEvents['QueryCacheHits'], ProfileEvents['QueryCacheMisses']
 FROM system.query_log
 WHERE type = 'QueryFinish'
+  AND current_database = currentDatabase()
   AND query = 'select 1 SETTINGS use_query_cache = true;';
 
 SYSTEM DROP QUERY CACHE;

tests/queries/0_stateless/02494_query_cache_events.reference

@@ -3,4 +3,5 @@
 0	1
 ---
 1
+0	1
 1	0

tests/queries/0_stateless/02494_query_cache_events.sql

@@ -1,9 +1,8 @@
 -- Tags: no-parallel
 -- Tag no-parallel: Messes with internal cache
 
--- Start with empty query cache QC and query log
+-- Start with empty query cache QC
 SYSTEM DROP QUERY CACHE;
-DROP TABLE system.query_log SYNC;
 
 -- Run a query with QC on. The first execution is a QC miss.
 SELECT '---';
@@ -13,6 +12,7 @@ SYSTEM FLUSH LOGS;
 SELECT ProfileEvents['QueryCacheHits'], ProfileEvents['QueryCacheMisses']
 FROM system.query_log
 WHERE type = 'QueryFinish'
+  AND current_database = currentDatabase()
   AND query = 'SELECT 1 SETTINGS use_query_cache = true;';
 
 
@@ -20,11 +20,12 @@ WHERE type = 'QueryFinish'
 SELECT '---';
 SELECT 1 SETTINGS use_query_cache = true;
 
-DROP TABLE system.query_log SYNC;
 SYSTEM FLUSH LOGS;
 SELECT ProfileEvents['QueryCacheHits'], ProfileEvents['QueryCacheMisses']
 FROM system.query_log
 WHERE type = 'QueryFinish'
-  AND query = 'SELECT 1 SETTINGS use_query_cache = true;';
+  AND current_database = currentDatabase()
+  AND query = 'SELECT 1 SETTINGS use_query_cache = true;'
+ORDER BY event_time_microseconds;
 
 SYSTEM DROP QUERY CACHE;

tests/queries/0_stateless/02494_query_cache_normalize_ast.sql

@@ -1,9 +1,8 @@
 -- Tags: no-parallel
 -- Tag no-parallel: Messes with internal cache
 
--- Start with empty query cache (QC) and query log.
+-- Start with empty query cache (QC)
 SYSTEM DROP QUERY CACHE;
-DROP TABLE system.query_log SYNC;
 
 -- Run query whose result gets cached in the query cache.
 -- Besides "use_query_cache", pass two more knobs (one QC-specific knob and one non-QC-specific knob). We just care
@@ -24,6 +23,7 @@ SYSTEM FLUSH LOGS;
 SELECT ProfileEvents['QueryCacheHits'], ProfileEvents['QueryCacheMisses']
 FROM system.query_log
 WHERE type = 'QueryFinish'
+  AND current_database = currentDatabase()
   AND query = 'SELECT 1 SETTINGS use_query_cache = true, enable_writes_to_query_cache = false, max_threads = 16;';
 
 SYSTEM DROP QUERY CACHE;

tests/queries/0_stateless/02494_query_cache_passive_usage.reference

@@ -9,4 +9,5 @@
 -----
 1
 1
+0	1
 1	0

tests/queries/0_stateless/02494_query_cache_passive_usage.sql

@@ -22,10 +22,7 @@ SELECT COUNT(*) FROM system.query_cache;
 
 SELECT '-----';
 
--- Run same query with passive mode again. There must still be one entry in the QC and we must have a QC hit.
+/* Run same query with passive mode again. There must still be one entry in the QC and we must have a QC hit. */
-
--- Get rid of log of previous SELECT
-DROP TABLE system.query_log SYNC;
 
 SELECT 1 SETTINGS use_query_cache = true, enable_writes_to_query_cache = false;
 SELECT COUNT(*) FROM system.query_cache;
@@ -34,6 +31,9 @@ SYSTEM FLUSH LOGS;
 SELECT ProfileEvents['QueryCacheHits'], ProfileEvents['QueryCacheMisses']
 FROM system.query_log
 WHERE type = 'QueryFinish'
-  AND query = 'SELECT 1 SETTINGS use_query_cache = true, enable_writes_to_query_cache = false;';
+  AND current_database = currentDatabase()
+  /* NOTE: client incorrectly join comments from the previous line into query, hence LIKE */
+  AND query LIKE '%\nSELECT 1 SETTINGS use_query_cache = true, enable_writes_to_query_cache = false;'
+ORDER BY event_time_microseconds;
 
 SYSTEM DROP QUERY CACHE;

tests/queries/0_stateless/02499_monotonicity_toUnixTimestamp64.sh

@@ -18,5 +18,5 @@ query_id="${CLICKHOUSE_DATABASE}_02499_$RANDOM$RANDOM"
 $CLICKHOUSE_CLIENT --query_id="$query_id" -q "select ts from t order by toUnixTimestamp64Nano(ts) limit 10 format Null settings max_block_size = $max_block_size, optimize_read_in_order = 1;"
 
 $CLICKHOUSE_CLIENT -q "system flush logs;"
-$CLICKHOUSE_CLIENT --param_query_id="$query_id" -q "select read_rows <= $max_block_size from system.query_log where event_date >= yesterday() and query_id = {query_id:String} and type = 'QueryFinish';"
+$CLICKHOUSE_CLIENT --param_query_id="$query_id" -q "select read_rows <= $max_block_size from system.query_log where event_date >= yesterday() and current_database = '$CLICKHOUSE_DATABASE' and query_id = {query_id:String} and type = 'QueryFinish';"
 

tests/queries/0_stateless/02521_incorrect_dealy_for_insert_bug_44902.sh

@@ -16,7 +16,7 @@ do
     query_id="${CLICKHOUSE_DATABASE}_02521_${i}_$RANDOM$RANDOM"
     $CLICKHOUSE_CLIENT --query_id="$query_id" -q "INSERT INTO test_02521_insert_delay SELECT number, toString(number) FROM numbers(${i}, 1)"
     $CLICKHOUSE_CLIENT -q "SYSTEM FLUSH LOGS"
-    $CLICKHOUSE_CLIENT --param_query_id="$query_id" -q "select ProfileEvents['DelayedInsertsMilliseconds'] as delay from system.query_log where event_date >= yesterday() and query_id = {query_id:String} order by delay desc limit 1"
+    $CLICKHOUSE_CLIENT --param_query_id="$query_id" -q "select ProfileEvents['DelayedInsertsMilliseconds'] as delay from system.query_log where event_date >= yesterday() and current_database = '$CLICKHOUSE_DATABASE' and query_id = {query_id:String} order by delay desc limit 1"
 done
 
 $CLICKHOUSE_CLIENT -q "INSERT INTO test_02521_insert_delay VALUES(0, 'This query throws error')" 2>&1 | grep -o 'TOO_MANY_PARTS' | head -n 1

tests/queries/0_stateless/02578_parameterized_rename_queries.sql

@@ -24,6 +24,7 @@ SET param_new_tbl_name = 02661_t1;
 CREATE TABLE {new_db_name:Identifier}.{old_tbl_name:Identifier} (a UInt64) ENGINE = MergeTree ORDER BY tuple();
 RENAME TABLE {new_db_name:Identifier}.{old_tbl_name:Identifier} TO {new_db_name:Identifier}.{new_tbl_name:Identifier};
 
+-- NOTE: no 'database = currentDatabase()' on purpose
 SELECT name FROM system.tables WHERE name = {new_tbl_name:String};
 
 -- Case 3: RENAME DICTIONARY

tests/queries/0_stateless/02585_query_status_deadlock.sh

@@ -1,4 +1,5 @@
 #!/usr/bin/env bash
+# Tags: long
 
 CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
 # shellcheck source=../shell_config.sh
@@ -14,7 +15,7 @@ $CLICKHOUSE_CLIENT -q "SYSTEM FLUSH LOGS"
 
 while true
 do
-    res=$($CLICKHOUSE_CLIENT -q "select query, event_time from system.query_log where query_id = '$QUERY_ID' and query like 'select%' limit 1")
+    res=$($CLICKHOUSE_CLIENT -q "select query, event_time from system.query_log where query_id = '$QUERY_ID' and current_database = '$CLICKHOUSE_DATABASE' and query like 'select%' limit 1")
     if [ -n "$res" ]; then
         break
     fi

tests/queries/0_stateless/02681_undrop_query.sql

@@ -21,7 +21,7 @@ detach table 02681_undrop_detach;
 undrop table 02681_undrop_detach; -- { serverError 57 }
 attach table 02681_undrop_detach;
 alter table 02681_undrop_detach update num = 2 where id = 1;
-select command from system.mutations where table='02681_undrop_detach' limit 1;
+select command from system.mutations where table='02681_undrop_detach' and database=currentDatabase() limit 1;
 drop table 02681_undrop_detach sync;
 
 select 'test MergeTree with cluster';

tests/queries/0_stateless/02761_ddl_initial_query_id.sh

@@ -21,4 +21,6 @@ $CLICKHOUSE_CLIENT -q "SYSTEM FLUSH LOGS"
 # - replace() is required to avoid non deterministic behaviour of
 #   normalizeQuery() that replaces the identifier with "?" only if it has more
 #   then two numbers.
+#
+# NOTE: no current_database = '$CLICKHOUSE_DATABASE' filter on purpose (since ON CLUSTER queries does not have current_database passed)
 $CLICKHOUSE_CLIENT -q "SELECT normalizeQuery(replace(query, currentDatabase(), 'default')) FROM system.query_log WHERE initial_query_id = '$query_id' AND type != 'QueryStart' ORDER BY event_time_microseconds"

tests/queries/0_stateless/02783_parallel_replicas_trivial_count_optimization.sh

@@ -12,7 +12,7 @@ function has_used_parallel_replicas () {
             sumIf(read_rows, is_initial_query) as read_rows,
             sumIf(read_bytes, is_initial_query) as read_bytes
         FROM system.query_log
-    WHERE event_date >= yesterday() and initial_query_id LIKE '$1%'
+    WHERE event_date >= yesterday() and initial_query_id LIKE '$1%' AND current_database = '$CLICKHOUSE_DATABASE'
     GROUP BY initial_query_id
     ORDER BY min(event_time_microseconds) ASC
     FORMAT TSV"

tests/queries/1_stateful/00177_memory_bound_merging.sh

@@ -12,6 +12,8 @@ CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
 check_replicas_read_in_order() {
     # to check this we actually look for at least one log message from MergeTreeInOrderSelectProcessor.
     # hopefully logger's names are a bit more stable than log messages itself
+    #
+    # NOTE: lack of "current_database = '$CLICKHOUSE_DATABASE'" filter is made on purpose
     $CLICKHOUSE_CLIENT -nq "
         SYSTEM FLUSH LOGS;
 

utils/check-style/check-style

@@ -13,7 +13,7 @@
 #  and then to run formatter only for the specified files.
 
 ROOT_PATH=$(git rev-parse --show-toplevel)
-EXCLUDE_DIRS='build/|integration/|widechar_width/|glibc-compatibility/|poco/|memcpy/|consistent-hashing|benchmark|tests/|utils/keeper-bench/example.yaml'
+EXCLUDE_DIRS='build/|integration/|widechar_width/|glibc-compatibility/|poco/|memcpy/|consistent-hashing|benchmark|tests/.*.cpp|utils/keeper-bench/example.yaml'
 
 # From [1]:
 #     But since array_to_string_internal() in array.c still loops over array
@@ -163,14 +163,12 @@ find $ROOT_PATH -not -path $ROOT_PATH'/contrib*' \( -name '*.yaml' -or -name '*.
 
 # Tests should not be named with "fail" in their names. It makes looking at the results less convenient.
 find $ROOT_PATH/tests/queries -iname '*fail*' |
-    grep -vP $EXCLUDE_DIRS |
     grep . && echo 'Tests should not be named with "fail" in their names. It makes looking at the results less convenient when you search for "fail" substring in browser.'
 
 # Queries to system.query_log/system.query_thread_log should have current_database = currentDatabase() condition
 # NOTE: it is not that accurate, but at least something.
 tests_with_query_log=( $(
     find $ROOT_PATH/tests/queries -iname '*.sql' -or -iname '*.sh' -or -iname '*.py' -or -iname '*.j2' |
-        grep -vP $EXCLUDE_DIRS |
         xargs grep --with-filename -e system.query_log -e system.query_thread_log | cut -d: -f1 | sort -u
 ) )
 for test_case in "${tests_with_query_log[@]}"; do
@@ -205,7 +203,6 @@ tables_with_database_column=(
 # NOTE: it is not that accuate, but at least something.
 tests_with_database_column=( $(
     find $ROOT_PATH/tests/queries -iname '*.sql' -or -iname '*.sh' -or -iname '*.py' -or -iname '*.j2' |
-        grep -vP $EXCLUDE_DIRS |
         xargs grep --with-filename $(printf -- "-e %s " "${tables_with_database_column[@]}") |
         grep -v -e ':--' -e ':#' |
         cut -d: -f1 | sort -u
@@ -225,7 +222,6 @@ done
 # NOTE: it is not that accuate, but at least something.
 tests_with_replicated_merge_tree=( $(
     find $ROOT_PATH/tests/queries -iname '*.sql' -or -iname '*.sh' -or -iname '*.py' -or -iname '*.j2' |
-        grep -vP $EXCLUDE_DIRS |
         xargs grep --with-filename -e "Replicated.*MergeTree[ ]*(.*" | cut -d: -f1 | sort -u
 ) )
 for test_case in "${tests_with_replicated_merge_tree[@]}"; do
@@ -233,7 +229,7 @@ for test_case in "${tests_with_replicated_merge_tree[@]}"; do
         *.gen.*)
             ;;
         *.sh)
-            test_case_zk_prefix="\$CLICKHOUSE_TEST_ZOOKEEPER_PREFIX"
+            test_case_zk_prefix="\(\$CLICKHOUSE_TEST_ZOOKEEPER_PREFIX\|{database}\)"
             grep -q -e "Replicated.*MergeTree[ ]*(.*$test_case_zk_prefix" "$test_case" || echo "Replicated.*MergeTree should contain '$test_case_zk_prefix' in zookeeper path to avoid overlaps ($test_case)"
             ;;
         *.sql|*.sql.j2)