CDCSDKConsistentStreamTest#TestCDCSDKConsistentStreamWithTabletSplit fails intermittently #20778
Assignees
Labels
Comments
yugabyte-ci
added
area/cdcsdk
devansh-ism
pushed a commit
that referenced
this issue
Mar 7, 2024
Summary: There were several instances of test runs failures with FlushTables client call timing out caused by https://yugabyte.atlassian.net/browse/DB-9925. The fix for the root issue has been landed as part of https://phorge.dev.yugabyte.com/D32215. This diff will add retry function as an wrapper over FlushTables call to retry on specific error. Jira: DB-9776 Test Plan: Jenkins: test regex: .*CDCSDK.* Reviewers: skumar, stiwary, siddharth.shah Reviewed By: skumar Subscribers: ycdcxcluster Differential Revision: https://phorge.dev.yugabyte.com/D32781
jasonyb
pushed a commit
that referenced
this issue
Apr 23, 2024
…ter-merge Merge YB master commit 1f64b6e titled [#20778] CDCSDK: Add retry logic over FlushTables calls in test and committed 2024-03-07T13:26:11+05:30 into YB pg15. - TestPgAlterTableChangePrimaryKey.java: YB master d5d7363 deletes this file, claiming relevant tests were moved to regress test yb_alter_table_rewrite. YB pg15 branch made minor adjustments such as message changes. Delete it. - TestPgReplicationSlot.java: YB master ddbe411 takes test replicationConnectionConsumption and turns it into helper testReplicationConnectionConsumption which is called by replicationConnectionConsumption and replicationConnectionConsumptionMultipleBatches later in the file. YB pg15 417e9b3 ignores test replicationConnectionConsumption, so do the same ignore to both tests. - parallel.c: YB master cfda8c7 changes spacing for enumblacklistlen, but that declaration was removed in YB pg15, so ignore it. - tablecmds.c: - AlteredTableInfo: trivial conflict on adding extra fields at the bottom with YB pg15 and YB master d5d7363. - function declarations: YB master d5d7363 adds extra parameter yb_wqueue to two functions. YB pg15 has several changes in the area. Trivial resolution. - ATRewriteTables: trivial conflict between YB pg15 merge 5627af5 and YB master d5d7363 regarding both sides adding params to function make_new_heap. - YbATSetPKRewriteChildPartitions: YB master d5d7363 adds heap_open/heap_close, but upstream PG switched to table_open/table_close. - slot.c: conflict between YB pg15 merge 417e9b3 and YB master e5dbd2e. Resolution appears to be to change "slot" to "s" and take YB master's change. - slotfuncs.c: - pg_create_logical_replication_slot: adjacent lines conflict between YB master e5dbd2e and YB pg15 merge ed96733. - pg_get_replication_slots: merge YB master e5dbd2e and YB pg15 merge ed96733. - ipci.c: upstream PG 0bd305ee1d427ef29f5fa4fa20567e3b3f5ff792 moves code into CalculateShmemSize. YB bda4da7 touches code there. Move that code to the new location. - pg_dump.c: YB master 5659b73 adds use_roles_sql logic while upstream PG changes the area. Merge. - event_trigger.h: YB master d5d7363 adds YB_AT_REWRITE_ALTER_PRIMARY_KEY as 0x16 while upstream PG 578b229718e8f15fa779e20f086c4b6bb3776106 and b0483263dda0824cc49e3f8a022dab07e1cdf9a7 modify definitions for 0x08. Trivial merge. - pgstat.h/wait_event.h: YB master bda4da7 touches yb_pgstat_report_wait_start, but YB pg15 merge 5627af5 moves the function to wait_event.h. Apply the modification there, and also move the associated yb_ash.h include. - pgstat.c/wait_event.c: YB master bda4da7 touches functions formerly in pgstat.c now moved to wait_event.c. Apply the same changes to the new location using sed -i 's/YBEnableAsh()/yb_ash_enable_infra/' src/postgres/src/backend/utils/activity/wait_event.c - guc.c: YB master bda4da7 adds ASH GUCs with config groups RESOURCES and STATS_COLLECTOR that upstream PG removes. Choose config group STATS_MONITORING for both GUCs. - pg15_tests/passing_tests.tsv: - remove org.yb.pgsql.TestPgAlterTableChangePrimaryKey from the list as that test was deleted by YB master d5d7363. - remove org.yb.pgsql.TestPgRegressReplicationSlot and CDCSDKYsqlTest.TestReplicationSlotDropWithActiveInvalid from the list because these tests start failing due to actual bugs. The source of the bugs is unclear, but in order to push forward merge progress, leave deeper investigation for later. - pg15_tests/test_regress_table.sh: YB master d5d7363 adds new regress test which suffers the same serial type bug as yb_alter_table. Add it to the list of tests expected to fail.
jasonyb
pushed a commit
that referenced
this issue
Apr 23, 2024
…dd3f76d' into pg15 Summary: Merge YB master commit 1f64b6e titled [#20778] CDCSDK: Add retry logic over FlushTables calls in test and committed 2024-03-07T13:26:11+05:30 into YB pg15. - TestPgAlterTableChangePrimaryKey.java: YB master d5d7363 deletes this file, claiming relevant tests were moved to regress test yb_alter_table_rewrite. YB pg15 branch made minor adjustments such as message changes. Delete it. - TestPgReplicationSlot.java: YB master ddbe411 takes test replicationConnectionConsumption and turns it into helper testReplicationConnectionConsumption which is called by replicationConnectionConsumption and replicationConnectionConsumptionMultipleBatches later in the file. YB pg15 417e9b3 ignores test replicationConnectionConsumption, so do the same ignore to both tests. - parallel.c: YB master cfda8c7 changes spacing for enumblacklistlen, but that declaration was removed in YB pg15, so ignore it. - tablecmds.c: - AlteredTableInfo: trivial conflict on adding extra fields at the bottom with YB pg15 and YB master d5d7363. - function declarations: YB master d5d7363 adds extra parameter yb_wqueue to two functions. YB pg15 has several changes in the area. Trivial resolution. - ATRewriteTables: trivial conflict between YB pg15 merge 5627af5 and YB master d5d7363 regarding both sides adding params to function make_new_heap. - YbATSetPKRewriteChildPartitions: YB master d5d7363 adds heap_open/heap_close, but upstream PG switched to table_open/table_close. - slot.c: conflict between YB pg15 merge 417e9b3 and YB master e5dbd2e. Resolution appears to be to change "slot" to "s" and take YB master's change. - slotfuncs.c: - pg_create_logical_replication_slot: adjacent lines conflict between YB master e5dbd2e and YB pg15 merge ed96733. - pg_get_replication_slots: merge YB master e5dbd2e and YB pg15 merge ed96733. - ipci.c: upstream PG 0bd305ee1d427ef29f5fa4fa20567e3b3f5ff792 moves code into CalculateShmemSize. YB bda4da7 touches code there. Move that code to the new location. - pg_dump.c: YB master 5659b73 adds use_roles_sql logic while upstream PG changes the area. Merge. - event_trigger.h: YB master d5d7363 adds YB_AT_REWRITE_ALTER_PRIMARY_KEY as 0x16 while upstream PG 578b229718e8f15fa779e20f086c4b6bb3776106 and b0483263dda0824cc49e3f8a022dab07e1cdf9a7 modify definitions for 0x08. Trivial merge. - pgstat.h/wait_event.h: YB master bda4da7 touches yb_pgstat_report_wait_start, but YB pg15 merge 5627af5 moves the function to wait_event.h. Apply the modification there, and also move the associated yb_ash.h include. - pgstat.c/wait_event.c: YB master bda4da7 touches functions formerly in pgstat.c now moved to wait_event.c. Apply the same changes to the new location using sed -i 's/YBEnableAsh()/yb_ash_enable_infra/' src/postgres/src/backend/utils/activity/wait_event.c - guc.c: YB master bda4da7 adds ASH GUCs with config groups RESOURCES and STATS_COLLECTOR that upstream PG removes. Choose config group STATS_MONITORING for both GUCs. - pg15_tests/passing_tests.tsv: - remove org.yb.pgsql.TestPgAlterTableChangePrimaryKey from the list as that test was deleted by YB master d5d7363. - remove org.yb.pgsql.TestPgRegressReplicationSlot and CDCSDKYsqlTest.TestReplicationSlotDropWithActiveInvalid from the list because these tests start failing due to actual bugs. The source of the bugs is unclear, but in order to push forward merge progress, leave deeper investigation for later. - pg15_tests/test_regress_table.sh: YB master d5d7363 adds new regress test which suffers the same serial type bug as yb_alter_table. Add it to the list of tests expected to fail. Test Plan: Almalinux 8: #!/usr/bin/env bash set -eu ./yb_build.sh fastdebug --gcc11 pg15_tests/run_tests.sh Almalinux 8, fastdebug, gcc11, get the following results: 0 2024-04-21T10:58:52-07:00 pgwrapper_pg_get_lock_status-test PgGetLockStatusTest.TestGetLockStatusLimitNumTxnLocks 0 2024-04-21T10:59:04-07:00 pgwrapper_pg_wait_on_conflict-test PgWaitQueuesTest.TestDDLsNotBlockedOnWaiters 0 2024-04-21T10:59:21-07:00 integration-tests_cdcsdk_ysql-test CDCSDKYsqlTest.TestCDCStateEntryForReplicationSlot 0 2024-04-21T11:00:33-07:00 pgwrapper_pg_wrapper-test PgWrapperFlagsTest.* 1 2024-04-21T11:30:42-07:00 JAVA org.yb.pgsql.TestPgEncryption#testSslWithAuth[0] 1 2024-04-21T12:02:00-07:00 JAVA org.yb.pgsql.TestPgEncryption 0 2024-04-21T12:02:11-07:00 master_clone_state_manager-test CloneStateManagerTest.* 0 2024-04-21T12:02:25-07:00 pgwrapper_pg_fkey-test PgFKeyTest.SameTableReference 1 2024-04-21T12:03:27-07:00 JAVA org.yb.pgsql.TestPgRegressParallel#testPgRegressParallel 1 2024-04-21T12:05:19-07:00 JAVA org.yb.pgsql.TestPgParallelReadIsolation 0 2024-04-21T12:05:50-07:00 integration-tests_wait_states-itest WaitStateITest.* 0 2024-04-21T12:10:29-07:00 integration-tests_wait_states-itest */AshTestVerifyOccurrence.VerifyWaitStateEntered/* 0 2024-04-21T12:11:20-07:00 JAVA org.yb.pgsql.TestYbBackup#testBackupRestoreRoles 0 2024-04-21T12:12:11-07:00 JAVA org.yb.pgsql.TestYbBackup#testBackupRolesWithoutUseRoles 0 2024-04-21T12:13:01-07:00 JAVA org.yb.pgsql.TestYbBackup#testBackupRolesWithoutRestoreRoles 1 2024-04-21T12:14:44-07:00 JAVA org.yb.pgsql.TestYsqlDump#ysqlDumpWithYbMetadata 1 2024-04-21T12:15:24-07:00 JAVA org.yb.pgsql.TestYsqlDump#ysqlDumpWithoutYbMetadata 1 2024-04-21T12:16:21-07:00 JAVA org.yb.pgsql.TestYsqlDump#ysqlDumpAllWithoutYbMetadata 1 2024-04-21T12:17:18-07:00 JAVA org.yb.pgsql.TestYsqlDump#ysqlDumpAllWithYbMetadata 1 2024-04-21T12:18:19-07:00 JAVA org.yb.pgsql.TestYsqlDump#ysqlDumpLegacyColocatedDB 1 2024-04-21T12:19:09-07:00 JAVA org.yb.pgsql.TestYsqlDump#ysqlDumpColocatedDB 1 2024-04-21T12:19:29-07:00 JAVA org.yb.pgsql.TestYbBackup#testExtensionBackupUsingTestExtension 0 2024-04-21T12:20:42-07:00 JAVA org.yb.pgsql.TestYbBackup#testIgnoreExistingTablespaces 0 2024-04-21T12:20:59-07:00 pgwrapper_pg_wait_on_conflict-test PgWaitQueuesTest.TestMultipleRequestsPerTxn 0 2024-04-21T12:21:36-07:00 integration-tests_cdcsdk_ysql-test CDCSDKYsqlTest.TestAtomicDDLDropColumn 0 2024-04-21T12:22:04-07:00 integration-tests_xcluster_ysql-test XClusterYsqlTest.DmlOperationsBlockedOnStandbyCluster 0 2024-04-21T12:27:51-07:00 integration-tests_wait_states-itest * 0 2024-04-21T12:28:19-07:00 pgwrapper_pg_ash-test PgAshTest.* 0 2024-04-21T12:39:33-07:00 integration-tests_cdcsdk_consumption_consistent_changes-test CDCSDKConsumptionConsistentChangesTest.* 1 2024-04-21T12:45:31-07:00 JAVA org.yb.pgsql.TestPgRegressPlanner 0 2024-04-21T12:45:46-07:00 client_snapshot-schedule-test CloneFromScheduleTest.Clone 0 2024-04-21T12:46:21-07:00 integration-tests_minicluster-snapshot-test PgCloneTest.Clone 8 2024-04-21T13:07:35-07:00 pgwrapper_pg_catalog_version-test * Almalinux 8, release, clang17, get the following results: 0 2024-04-21T13:26:49-07:00 pgwrapper_pg_index_backfill-test PgIndexBackfillTest.GinStress Jenkins: rebase: pg15 Reviewers: aagrawal, tfoucher, xCluster, hsunder Reviewed By: tfoucher Subscribers: ybase, ycdcxcluster, yql Differential Revision: https://phorge.dev.yugabyte.com/D34352
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Jira Link: DB-9776
The text was updated successfully, but these errors were encountered: