chihsuan opened a new pull request, #10620:
URL: https://github.com/apache/ozone/pull/10620

   ## What changes were proposed in this pull request?
   
   Test-only change.
   
   **Problem.** After the earlier fix set `hdds.container.report.interval` to 
1s, `TestContainerReportHandling` (and its HA variant) still timed out 
intermittently on CI. The test marks a non-empty CLOSED container 
DELETING/DELETED and restarts all datanodes, expecting the restart's full 
container report to carry CLOSED replicas so SCM deletes them. During the 
restart, though, a replica is briefly reported as CLOSING. SCM treats a 
non-empty, non-CLOSED replica on a DELETING/DELETED container as a signal to 
resurrect the container back to CLOSED (`AbstractContainerReportHandler`, the 
HDDS-11136 / HDDS-12421 path). Once resurrected, SCM never issues a delete 
command, so the replicas are never removed and the 180s wait times out.
   
   Evidence from a failing CI run: the log shows `Resurrecting container #1 ... 
due to non-empty CLOSING replica`, and that run sent zero delete commands 
versus 9-21 in the passing runs.
   
   **Fix.** The datanode restart was only ever a way to force a timely report 
back when the report interval defaulted to 60 minutes. With the interval now at 
1s, the periodic full container report already delivers the CLOSED replicas and 
triggers deletion, so the restart is unnecessary and is the sole source of the 
transient CLOSING report. Removing it eliminates the race while still 
exercising the intended behavior: a CLOSED replica reported on a 
DELETING/DELETED container gets deleted. The same change is applied to 
`TestContainerReportHandlingWithHA`.
   
   ## What is the link to the Apache JIRA
   
   https://issues.apache.org/jira/browse/HDDS-14774
   
   ## How was this patch tested?
   
   - flaky-test-check workflow: 100 runs (10 splits x 10 iterations) of all 
methods in both classes, all green.
   - Ran both test classes locally across several iterations; all parameters 
pass.
   - `checkstyle` passes; fork `build-branch` CI is green.
   


-- 
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]

Reply via email to