Add query_prefix to query params and switch off grace join in tpcds s10 q72 (#12764)

Author: iddqdex

ydb/apps/ydb/CHANGELOG.md

@@ -1,4 +1,4 @@
-
+* Replaced option `--query-settings` by `--query-prefix` one in `ydb workload <workload> run`.
 * Added new options to `ydb workload topic`: --tx-commit-interval and --tx-commit-messages, allowing you to specify commit interval either in milliseconds or in number of messages written. 
 Also now you can load test YDB topics, using wide transactions that span over all partitions in the topic. This works both in write and in end-to-end workload scenarios. 
 

ydb/public/lib/ydb_cli/commands/ydb_benchmark.cpp

@@ -37,9 +37,11 @@ void TWorkloadCommandBenchmark::Config(TConfig& config) {
     config.Opts->AddLongOption("plan", "Query plans report file name")
         .DefaultValue("")
         .StoreResult(&PlanFileName);
-    config.Opts->AddLongOption("query-settings", "Query settings.\nEvery setting is a line that will be added to the beginning of each query. For multiple settings lines use this option several times.")
-        .DefaultValue("")
+    config.Opts->AddLongOption("query-settings")
+        .AppendTo(&QuerySettings).Hidden();
+    config.Opts->AddLongOption("query-prefix", "Query prefix.\nEvery prefix is a line that will be added to the beginning of each query. For multiple prefixes lines use this option several times.")
         .AppendTo(&QuerySettings);
+    config.Opts->MutuallyExclusive("query-prefix", "query-settings");
     auto fillTestCases = [](TStringBuf line, std::function<void(ui32)>&& op) {
         for (const auto& token : StringSplitter(line).Split(',').SkipEmpty()) {
             TStringBuf part = token.Token();

ydb/tests/olap/lib/ydb_cli.py

@@ -72,7 +72,8 @@ def __init__(self,
                      timeout: float,
                      check_canonical: CheckCanonicalPolicy,
                      query_syntax: str,
-                     scale: Optional[int]):
+                     scale: Optional[int],
+                     query_prefix: Optional[str]):
             def _get_output_path(ext: str) -> str:
                 return yatest.common.test_output_path(f'q{query_num}.{ext}')
 
@@ -85,6 +86,7 @@ def _get_output_path(ext: str) -> str:
             self.check_canonical = check_canonical
             self.query_syntax = query_syntax
             self.scale = scale
+            self.query_prefix = query_prefix
             self._nodes_info: dict[str, dict[str, int]] = {}
             self._plan_path = _get_output_path('plan')
             self._query_output_path = _get_output_path('out')
@@ -210,9 +212,8 @@ def _get_cmd(self) -> list[str]:
                 '--global-timeout', f'{self.timeout}s',
                 '--verbose'
             ]
-            query_preffix = get_external_param('query-prefix', '')
-            if query_preffix:
-                cmd += ['--query-settings', query_preffix]
+            if self.query_prefix:
+                cmd += ['--query-prefix', self.query_prefix]
             if self.check_canonical != CheckCanonicalPolicy.NO:
                 cmd.append('--check-canonical')
             if self.query_syntax:
@@ -248,7 +249,7 @@ def process(self) -> YdbCliHelper.WorkloadRunResult:
     @staticmethod
     def workload_run(workload_type: WorkloadType, path: str, query_num: int, iterations: int = 5,
                      timeout: float = 100., check_canonical: CheckCanonicalPolicy = CheckCanonicalPolicy.NO, query_syntax: str = '',
-                     scale: Optional[int] = None) -> YdbCliHelper.WorkloadRunResult:
+                     scale: Optional[int] = None, query_prefix=None) -> YdbCliHelper.WorkloadRunResult:
         return YdbCliHelper.WorkloadProcessor(
             workload_type,
             path,
@@ -257,5 +258,6 @@ def workload_run(workload_type: WorkloadType, path: str, query_num: int, iterati
             timeout,
             check_canonical,
             query_syntax,
-            scale
+            scale,
+            query_prefix=query_prefix
         ).process()

ydb/tests/olap/load/lib/conftest.py

@@ -6,6 +6,7 @@
 from ydb.tests.olap.lib.ydb_cluster import YdbCluster
 from ydb.tests.olap.lib.allure_utils import allure_test_description
 from ydb.tests.olap.lib.results_processor import ResultsProcessor
+from ydb.tests.olap.lib.utils import get_external_param
 from ydb.tests.olap.scenario.helpers.scenario_tests_helper import ScenarioTestHelper
 from time import time
 from typing import Optional
@@ -15,9 +16,10 @@
 
 class LoadSuiteBase:
     class QuerySettings:
-        def __init__(self, iterations: Optional[int] = None, timeout: Optional[float] = None) -> None:
+        def __init__(self, iterations: Optional[int] = None, timeout: Optional[float] = None, query_prefix: Optional[str] = None) -> None:
             self.iterations = iterations
             self.timeout = timeout
+            self.query_prefix = query_prefix
 
     iterations: int = 5
     workload_type: WorkloadType = None
@@ -27,6 +29,7 @@ def __init__(self, iterations: Optional[int] = None, timeout: Optional[float] =
     query_syntax: str = ''
     query_settings: dict[int, LoadSuiteBase.QuerySettings] = {}
     scale: Optional[int] = None
+    query_prefix: str = get_external_param('query-prefix', '')
 
     @classmethod
     def suite(cls) -> str:
@@ -36,14 +39,20 @@ def suite(cls) -> str:
         return result
 
     @classmethod
-    def _get_iterations(cls, query_num: int) -> int:
-        q = cls.query_settings.get(query_num, None)
-        return q.iterations if q is not None and q.iterations is not None else cls.iterations
-
-    @classmethod
-    def _get_timeout(cls, query_num: int) -> float:
-        q = cls.query_settings.get(query_num, None)
-        return q.timeout if q is not None and q.timeout is not None else cls.timeout
+    def _get_query_settings(cls, query_num: int) -> QuerySettings:
+        result = LoadSuiteBase.QuerySettings(
+            iterations=cls.iterations,
+            timeout=cls.timeout,
+            query_prefix=cls.query_prefix
+        )
+        q = cls.query_settings.get(query_num, LoadSuiteBase.QuerySettings())
+        if q.iterations is not None:
+            result.iterations = q.iterations
+        if q.timeout is not None:
+            result.timeout = q.timeout
+        if q.query_prefix is not None:
+            result.query_prefix = q.query_prefix
+        return result
 
     @classmethod
     def _test_name(cls, query_num: int) -> str:
@@ -204,16 +213,18 @@ def run_workload_test(self, path: str, query_num: int) -> None:
                     for param in allure_test_result.parameters:
                         if param.name == 'query_num':
                             param.mode = allure.parameter_mode.HIDDEN.value
+        qparams = self._get_query_settings(query_num)
         start_time = time()
         result = YdbCliHelper.workload_run(
             path=path,
             query_num=query_num,
-            iterations=self._get_iterations(query_num),
+            iterations=qparams.iterations,
             workload_type=self.workload_type,
-            timeout=self._get_timeout(query_num),
+            timeout=qparams.timeout,
             check_canonical=self.check_canonical,
             query_syntax=self.query_syntax,
-            scale=self.scale
+            scale=self.scale,
+            query_prefix=qparams.query_prefix
         )
         allure_test_description(self.suite(), self._test_name(query_num), refference_set=self.refference, start_time=start_time, end_time=time())
-        self.process_query_result(result, query_num, self._get_iterations(query_num), True)
+        self.process_query_result(result, query_num, qparams.iterations, True)

ydb/tests/olap/load/lib/tpcds.py

@@ -73,6 +73,10 @@ class TestTpcds10(TpcdsSuiteBase):
     scale: int = 10
     check_canonical: bool = CheckCanonicalPolicy.WARNING
     timeout = max(TpcdsSuiteBase.timeout, 300.)
+    query_settings = {
+        # temporary, https://github.com/ydb-platform/ydb/issues/11767#issuecomment-2553353146
+        72: LoadSuiteBase.QuerySettings(query_prefix='pragma ydb.UseGraceJoinCoreForMap = "false";'),
+    }
     tables_size: dict[str, int] = {
         'call_center': 24,
         'catalog_page': 12000,