XComp commented on code in PR #21742:
URL: https://github.com/apache/flink/pull/21742#discussion_r1099989601
##########
flink-runtime/src/main/java/org/apache/flink/runtime/leaderelection/DefaultLeaderElectionService.java:
##########
@@ -100,6 +100,7 @@ public final void stop() throws Exception {
if (!running) {
return;
}
+ leaderContender.revokeLeadership();
Review Comment:
Thanks for clarifying things in this matter.
> However, I strongly have a mind that the leader information stored in the
ZNode and ConfigMap needs to be cleaned up if a leader contender is revoked
leadership. Before this change, this is true since the revokeLeadership happens
along with clearConfirmedLeaderInformation.
This could be fixed by adding to the contract of
`DefaultLeaderElectionService` that it calls `onRevokeLeadership()` as part of
the `DefaultLeaderElectionService.stop` call before actually stopping the
service. Essentially, leadership can be revoked by the HA backend or by itself
when calling `stop`. The leader elector still having leadership shouldn't be a
problem because only the contender deregisters himself from the leadership
process.
--
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]
