MartijnVisser commented on code in PR #29675:
URL: https://github.com/apache/flink/pull/29675#discussion_r4426549412
##########
flink-tests/src/test/java/org/apache/flink/test/state/ChangelogCompatibilityITCase.java:
##########
@@ -307,6 +311,15 @@ public void before() throws Exception {
Configuration config = new Configuration();
config.set(CHECKPOINTS_DIRECTORY, pathToString(checkpointDir));
config.set(SAVEPOINT_DIRECTORY, pathToString(savepointDir));
+ // The completed checkpoint store reads this from the cluster config,
not the job config.
+ // With the default of 1, a checkpoint completing while the job is
being cancelled subsumes
Review Comment:
In the [10-08
nightly](https://dev.azure.com/apache-flink/apache-flink/_build/results?buildId=79931&view=logs&j=2c3cbe13-dee0-5837-cf47-3053da9a8a78),
chk-2 completed at 02:57:34,143, before the cancel at 34,148, and the restore
still used chk-1. I think that is a stale
`CheckpointStatsTracker#createSnapshot`; can this comment and the commit
message describe that race instead?
--
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]
