[
https://issues.apache.org/jira/browse/IMPALA-14919?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18075887#comment-18075887
]
Quanlong Huang commented on IMPALA-14919:
-----------------------------------------
The test expects the two ALTER_TABLE RENAME events are skipped after
event-processor is paused.
{code:python}
self.execute_query("create table {}.tbl_a(i int)".format(unique_database))
# Wait until the CREATE_DATABASE and CREATE_TABLE events are skipped.
EventProcessorUtils.wait_for_event_processing(self)
self.execute_query(
"alter table {0}.tbl_a rename to {0}.tbl_b".format(unique_database))
self.execute_query(":event_processor('pause')") # <--- The ALTER event
could already be skipped before this.
... run another ALTER RENAME command
events_skipped_before =
EventProcessorUtils.get_int_metric('events-skipped', 0)
self.execute_query(":event_processor('start')")
EventProcessorUtils.wait_for_event_processing(self)
events_skipped_after =
EventProcessorUtils.get_int_metric('events-skipped', 0)
assert events_skipped_after == events_skipped_before + 2{code}
However, the first ALTER event could come and be skipped before the pause, as
shown in the failed logs
{noformat}
I20260422 12:46:24.860530 84928 MetastoreEvents.java:947] EventId: 30364
EventType: ALTER_TABLE Target: test_self_rename_events_43dbc286.tbl_b.
Incremented skipped metric to 29
...
I20260422 12:46:24.901597 97158 JniUtil.java:167] setEventProcessorStatus
request: PAUSE event processor
...
I20260422 12:46:31.869876 84927 MetastoreEvents.java:947] EventId: 30365
EventType: ALTER_TABLE Target: test_self_rename_events_43dbc286.tbl_a.
Incremented skipped metric to 30{noformat}
So the expected number should be 30, not 31 which is due to reading
'events-skipped' too late.
> TestSelfRenameEvent.test_self_rename_events is flaky
> ----------------------------------------------------
>
> Key: IMPALA-14919
> URL: https://issues.apache.org/jira/browse/IMPALA-14919
> Project: IMPALA
> Issue Type: Bug
> Components: Test
> Reporter: Quanlong Huang
> Assignee: Quanlong Huang
> Priority: Major
>
> Saw this failure in a downstream build
> {noformat}
> metadata.test_event_processing.TestSelfRenameEvent.test_self_rename_events
> (from pytest){noformat}
> h3. Stacktrace
> {noformat}
> /data/jenkins/workspace/impala-cdw-master-core-s3-arm-data-cache/repos/Impala/tests/metadata/test_event_processing.py:806:
> in test_self_rename_events
> assert events_skipped_after == events_skipped_before + 2
> E assert 30 == 31
> E +30
> E -31
> alter_client = <tests.common.impala_connection.ImpylaHS2Connection
> object at 0xffff9850ed30>
> alter_handle = <tests.common.impala_connection.OperationHandle object
> at 0xffff956ffdc0>
> catalogd = <CatalogdProcess PID: 84432
> (/data/jenkins/workspace/impala-cdw-master-core-s3-arm-data-cache/repos/Impala/be/build/la...sts
> -kudu_master_hosts 127.0.0.1 -catalog_service_port=26000
> -state_store_subscriber_port=23020 -webserver_port=25020)>
> events_skipped_after = 30
> events_skipped_before = 29
> self =
> <tests.metadata.test_event_processing.TestSelfRenameEvent object at
> 0xffff956ffcd0>
> start_time = 1776887185.0192785
> unique_database = 'test_self_rename_events_43dbc286'
> version_after_create = 2154{noformat}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
