carp84 commented on a change in pull request #9128: [FLINK-13256] Ensure
periodical checkpoint still valid when region failover abort pending checkpoints
URL: https://github.com/apache/flink/pull/9128#discussion_r304048806
##########
File path:
flink-runtime/src/main/java/org/apache/flink/runtime/checkpoint/CheckpointCoordinator.java
##########
@@ -1270,11 +1266,30 @@ public void stopCheckpointScheduler() {
}
}
+ /**
+ * Aborts all the pending checkpoints and check whether {@code
currentPeriodicTrigger} has been assigned as null.
+ * If so, we would create a new trigger as {@code
currentPeriodicTrigger}.
+ *
+ * @param exception The exception.
+ */
+ public void
abortPendingCheckpointsWithTriggerValidation(CheckpointException exception) {
+ synchronized (lock) {
Review comment:
I could see we're synchronizing on `lock` to prevent starting multiple
checkpoint schedulers, not only here but also in the
`start/stopCheckpointScheduler` methods. IMHO this interferes with the comment
on `lock` which indicates it to only safeguard checkpoint updates, or say
manipulation of `pendingCheckpoints`. It's kind of dangerous and using a
separate lock for scheduler may be more reasonable, but it's not necessary to
be improved in this PR.
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
