smengcl opened a new pull request, #11524:
URL: https://github.com/apache/ozone/pull/11524
Generated-by: Claude Code (Opus 5.5)
## What changes were proposed in this pull request?
`TestLeaseRecovery#closeLatestContainer` closed the container with the
highest ID in SCM, assuming it holds the block the test just wrote. Once
earlier test methods have created more containers, the block is often in a
different container, so the test closes an unrelated pipeline. Lease recovery
then finalizes the block through the still open pipeline and never sends
`GetCommittedBlockLength`, which makes `testGetCommittedBlockLengthTimeout` and
`testGetCommittedBlockLengthWithException` fail. They pass on rerun because the
rerun uses a fresh cluster.
In the last 214 master runs, every `integration (flaky)` job needed a rerun
of `TestLeaseRecovery` because of this.
This PR makes the helper close the container of the last block written by
the test's output stream.
The `@Flaky("HDDS-11323")` annotation is kept, since HDDS-11323 tracks a
different failure in `testRecovery`.
## What is the link to the Apache JIRA
https://issues.apache.org/jira/browse/HDDS-16987
## How was this patch tested?
- Locally with `-Ptest-flaky`: `testGetCommittedBlockLengthWithException`
failed on the first attempt before the change; 3 out of 3 runs passed after it.
- Repeated runs:
https://github.com/smengcl/hadoop-ozone/actions/runs/38507887479 (10x10, in
progress)
🤖 Generated with [Claude Code](https://claude.com/claude-code)
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
