dawidwys commented on a change in pull request #19198:
URL: https://github.com/apache/flink/pull/19198#discussion_r833995464
##########
File path:
flink-runtime/src/main/java/org/apache/flink/runtime/checkpoint/CheckpointCoordinator.java
##########
@@ -1243,7 +1243,7 @@ private void completePendingCheckpoint(PendingCheckpoint
pendingCheckpoint)
// the pending checkpoint must be discarded after the finalization
Preconditions.checkState(pendingCheckpoint.isDisposed() &&
completedCheckpoint != null);
- if (!props.isSavepoint()) {
+ if (!props.isSavepoint() || props.isSynchronous()) {
lastSubsumed =
addCompletedCheckpointToStoreAndSubsumeOldest(
Review comment:
1. For the subsumption, we do have a logic that will keep at least one
other checkpoint (not-savepoint), so I think that's fine. (see:
`CheckpointSubsumeHelper#subsume`).
2. As for the ownership, there are two aspects here. First, we can say
savepoint (stop-with-savepoint) is under Flink's control until the job has
finished. Secondly, this reverts to the behaviour of <=1.14, the better
handling we discussed should be implemented in
https://issues.apache.org/jira/browse/FLINK-26683
--
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]
