[
https://issues.apache.org/jira/browse/PHOENIX-7506?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17912342#comment-17912342
]
Istvan Toth commented on PHOENIX-7506:
--------------------------------------
In the master logs, I see the balancer trying to close the region, then
aborting it five minutes later.
What is the timeout property used by master here ?
Maybe we should use the same timeout ?
If we crash the RS anyway, perhaps it's better let it crash because of the
original deadlock, and not from the preCommit hook ?
> Crash in UngroupedAggregateRegionObserver.waitForScansToFinish on region move
> -----------------------------------------------------------------------------
>
> Key: PHOENIX-7506
> URL: https://issues.apache.org/jira/browse/PHOENIX-7506
> Project: Phoenix
> Issue Type: Bug
> Components: core
> Reporter: Istvan Toth
> Priority: Major
>
> Crash in UngroupedAggregateRegionObserver.waitForScansToFinish causes
> crashes on region stop.
> # I am not sure about using a client property for coprocessor functionality
> like this.
> # Even ignoring that, if a scan takes longer that timeout, a crash will
> still happen.
> I am not sure if there is an easy for for this, but forcing a crash in this
> case is not a good solution.
> Does the coporcessor accept new calls after preClose is called ?
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
