[
https://issues.apache.org/jira/browse/IMPALA-13701?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17922215#comment-17922215
]
ASF subversion and git services commented on IMPALA-13701:
----------------------------------------------------------
Commit b47955df1fa6cf89a8a987821ff5ebcf38970aff in impala's branch
refs/heads/master from Riza Suminto
[ https://gitbox.apache.org/repos/asf?p=impala.git;h=b47955df1 ]
IMPALA-13701: Rewrite test_invalidate_stale_partition_on_reload
Test test_invalidate_stale_partition_on_reload continues to be flaky. It
is suspected that the test order is not deterministic enough to ensure
expected Coordinator log lines are printed consistently.
This patch rewrite the test and make it more deterministic by:
1. Set shorter hms_event_polling_interval_s and
statestore_heartbeat_frequency_ms.
2. Running the "select *" query twice: before, and after the Hive insert
query.
3. Wait for a full statestore_heartbeat_frequency_ms before inspecting
logs and run the second query.
4. Validate that the second 'select *' query use higher table version
number than the first one.
Testing:
- Loop the test 100 times and pass in local machine.
Change-Id: Id578de1d6c9f809178d0b612068c449cf6c5f653
Reviewed-on: http://gerrit.cloudera.org:8080/22400
Reviewed-by: Impala Public Jenkins <[email protected]>
Tested-by: Impala Public Jenkins <[email protected]>
> Make test_invalidate_stale_partition_on_reload more deterministic
> -----------------------------------------------------------------
>
> Key: IMPALA-13701
> URL: https://issues.apache.org/jira/browse/IMPALA-13701
> Project: IMPALA
> Issue Type: Bug
> Components: Test
> Reporter: Riza Suminto
> Assignee: Riza Suminto
> Priority: Major
> Labels: broken-build
> Fix For: Impala 4.5.0
>
>
> Test test_invalidate_stale_partition_on_reload continues to be flaky.
> I suspect that the test order is not deterministic enough to ensure expected
> Coordinator log lines are printed consistently. We can make this test case
> more deterministic by:
> # Running the "select *" query twice: before, and after the Hive insert query.
> # Set shorter hms_event_polling_interval_s and
> statestore_heartbeat_frequency_ms.
> # Wait for a full statestore_heartbeat_frequency_ms before inspecting logs
> and run the second query.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
