[
https://issues.apache.org/jira/browse/GEODE-10287?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17537840#comment-17537840
]
Darrel Schneider commented on GEODE-10287:
------------------------------------------
I'm okay with not fixing this for 1.15.
But something is wrong. At most all waitForCurrentOperations will do is call
forceNewMembershipVersion and waitForCurrentOperations on the closed advisor.
But both of these methods will do nothing if the advisor is closed because the
underlying operationMonitor was also closed.
So if the current code is correct then we should not even bother calling
waitForCurrentOperations in this method since it is ALWAYS a noop. But I don't
know if we actually need to call a functional waitForCurrentOperations here. If
we do then the hangs you found would need to be addressed.
> DistributedRegion.distributedRegionCleanup logic looks wrong
> ------------------------------------------------------------
>
> Key: GEODE-10287
> URL: https://issues.apache.org/jira/browse/GEODE-10287
> Project: Geode
> Issue Type: Bug
> Components: core
> Reporter: Darrel Schneider
> Assignee: Jinmei Liao
> Priority: Major
> Labels: needsTriage
>
> DistributedRegion.distributedRegionCleanup does this: distAdvisor.close().
> Then a few lines later it calls "waitForCurrentOperations()". But
> waitForCurrentOperations uses the closed distAdvisor. Maybe it is okay to
> uses a closed distAdvisor but it seems better to call wait first and then
> close distAdvisor.
--
This message was sent by Atlassian Jira
(v8.20.7#820007)
