Merge pull request #52184 from azat/tests/randomize-timezone

Randomize timezone in tests across non-deterministic around 1970 and default

docker/test/stateless/run.sh

@@ -4,6 +4,9 @@
 set -e -x -a
 
 # Choose random timezone for this test run.
+#
+# NOTE: that clickhouse-test will randomize session_timezone by itself as well
+# (it will choose between default server timezone and something specific).
 TZ="$(rg -v '#' /usr/share/zoneinfo/zone.tab  | awk '{print $3}' | shuf | head -n1)"
 echo "Choosen random timezone $TZ"
 ln -snf "/usr/share/zoneinfo/$TZ" /etc/localtime && echo "$TZ" > /etc/timezone

tests/clickhouse-test

@@ -529,6 +529,12 @@ def threshold_generator(always_on_prob, always_off_prob, min_val, max_val):
     return gen
 
 
+# To keep dependency list as short as possible, tzdata is not used here (to
+# avoid try/except block for import)
+def get_localzone():
+    return os.getenv("TZ", "/".join(os.readlink("/etc/localtime").split("/")[-2:]))
+
+
 class SettingsRandomizer:
     settings = {
         "max_insert_threads": lambda: 0
@@ -602,20 +608,33 @@ class SettingsRandomizer:
         "enable_memory_bound_merging_of_aggregation_results": lambda: random.randint(
             0, 1
         ),
+        "session_timezone": lambda: random.choice(
+            [
+                # special non-deterministic around 1970 timezone, see [1].
+                #
+                #   [1]: https://github.com/ClickHouse/ClickHouse/issues/42653
+                "America/Mazatlan",
+                "America/Hermosillo",
+                "Mexico/BajaSur",
+                # server default that is randomized across all timezones
+                # NOTE: due to lots of trickery we cannot use empty timezone here, but this should be the same.
+                get_localzone(),
+            ]
+        ),
     }
 
     @staticmethod
     def get_random_settings(args):
-        random_settings = []
+        random_settings = {}
         is_debug = BuildFlags.DEBUG in args.build_flags
         for setting, generator in SettingsRandomizer.settings.items():
             if (
                 is_debug
                 and setting == "allow_prefetched_read_pool_for_remote_filesystem"
             ):
-                random_settings.append(f"{setting}=0")
+                random_settings[setting] = 0
             else:
-                random_settings.append(f"{setting}={generator()}")
+                random_settings[setting] = generator()
         return random_settings
 
 
@@ -651,10 +670,10 @@ class MergeTreeSettingsRandomizer:
 
     @staticmethod
     def get_random_settings(args):
-        random_settings = []
+        random_settings = {}
         for setting, generator in MergeTreeSettingsRandomizer.settings.items():
             if setting not in args.changed_merge_tree_settings:
-                random_settings.append(f"{setting}={generator()}")
+                random_settings[setting] = generator()
         return random_settings
 
 
@@ -766,7 +785,14 @@ class TestCase:
 
     @staticmethod
     def cli_format_settings(settings_list) -> str:
-        return " ".join([f"--{setting}" for setting in settings_list])
+        out = []
+        for k, v in settings_list.items():
+            out.extend([f"--{k}", str(v)])
+        return " ".join(out)
+
+    @staticmethod
+    def http_format_settings(settings_list) -> str:
+        return urllib.parse.urlencode(settings_list)
 
     def has_show_create_table_in_test(self):
         return not subprocess.call(["grep", "-iq", "show create", self.case_file])
@@ -774,11 +800,12 @@ class TestCase:
     def add_random_settings(self, client_options):
         new_options = ""
         if self.randomize_settings:
+            http_params = self.http_format_settings(self.random_settings)
             if len(self.base_url_params) == 0:
-                os.environ["CLICKHOUSE_URL_PARAMS"] = "&".join(self.random_settings)
+                os.environ["CLICKHOUSE_URL_PARAMS"] = http_params
             else:
                 os.environ["CLICKHOUSE_URL_PARAMS"] = (
-                    self.base_url_params + "&" + "&".join(self.random_settings)
+                    self.base_url_params + "&" + http_params
                 )
 
             new_options += f" {self.cli_format_settings(self.random_settings)}"

tests/queries/0_stateless/00387_use_client_time_zone.sh

@@ -5,4 +5,5 @@ CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
 # shellcheck source=../shell_config.sh
 . "$CURDIR"/../shell_config.sh
 
-env TZ=UTC ${CLICKHOUSE_CLIENT} --use_client_time_zone=1 --query="SELECT toDateTime(1000000000)"
+# NOTE: session_timezone overrides use_client_time_zone, disable it randomization
+env TZ=UTC ${CLICKHOUSE_CLIENT} --session_timezone '' --use_client_time_zone=1 --query="SELECT toDateTime(1000000000)"

tests/queries/0_stateless/00427_alter_primary_key.sh

@@ -7,11 +7,12 @@ CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
 function perform()
 {
     local query=$1
-    TZ=UTC $CLICKHOUSE_CLIENT \
-         --allow_deprecated_syntax_for_merge_tree=1 \
-        --use_client_time_zone=1 \
-        --input_format_values_interpret_expressions=0 \
-        --query "$query" 2>/dev/null
+    local settings=(
+        --allow_deprecated_syntax_for_merge_tree 1
+        --session_timezone UTC
+        --input_format_values_interpret_expressions 0
+    )
+    TZ=UTC $CLICKHOUSE_CLIENT "${settings[@]}" --query "$query" 2>/dev/null
     if [ "$?" -ne 0 ]; then
         echo "query failed"
     fi

tests/queries/0_stateless/00933_ttl_simple.sql

@@ -1,3 +1,15 @@
+-- disable timezone randomization since otherwise TTL may fail at particular datetime, i.e.:
+--
+--     SELECT
+--         now(),
+--         toDate(toTimeZone(now(), 'America/Mazatlan')),
+--         today()
+--
+--     ┌───────────────now()─┬─toDate(toTimeZone(now(), 'America/Mazatlan'))─┬────today()─┐
+--     │ 2023-07-24 06:24:06 │                                    2023-07-23 │ 2023-07-24 │
+--     └─────────────────────┴───────────────────────────────────────────────┴────────────┘
+set session_timezone = '';
+
 drop table if exists ttl_00933_1;
 
 -- Column TTL works only with wide parts, because it's very expensive to apply it for compact parts

tests/queries/0_stateless/01042_system_reload_dictionary_reloads_completely.sh

@@ -7,8 +7,8 @@ CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
 
 set -e -o pipefail
 
-# Run the client.
-$CLICKHOUSE_CLIENT --multiquery <<'EOF'
+# NOTE: dictionaries TTLs works with server timezone, so session_timeout cannot be used
+$CLICKHOUSE_CLIENT --session_timezone '' --multiquery <<'EOF'
 DROP DATABASE IF EXISTS dictdb_01042;
 CREATE DATABASE dictdb_01042;
 CREATE TABLE dictdb_01042.table(x Int64, y Int64, insert_time DateTime) ENGINE = MergeTree ORDER BY tuple();

tests/queries/0_stateless/01070_modify_ttl_recalc_only.sql

@@ -2,6 +2,9 @@
 
 set mutations_sync = 2;
 
+-- system.parts has server default, timezone cannot be randomized
+set session_timezone = '';
+
 drop table if exists ttl;
 
 create table ttl (d Date, a Int) engine = MergeTree order by a partition by toDayOfMonth(d)

tests/queries/0_stateless/02530_dictionaries_update_field.sh

@@ -5,7 +5,8 @@ CUR_DIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
 # shellcheck source=../shell_config.sh
 . "$CUR_DIR"/../shell_config.sh
 
-$CLICKHOUSE_CLIENT -q "
+# NOTE: dictionaries will be updated according to server TZ, not session, so prohibit it's randomization
+$CLICKHOUSE_CLIENT --session_timezone '' -q "
     CREATE TABLE table_for_update_field_dictionary
     (
         key UInt64,