[
https://issues.apache.org/jira/browse/HBASE-11834?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
stack updated HBASE-11834:
--------------------------
Attachment: 11834.txt
If memstore size changed to be non-zero after taking the update lock, we were
skipping out w/o releasing the lock. Remove a try/finally and move the release
of the update lock down so always let go.
Added a utility toString for locks usefully debugging. Added it into the test.
> TestHRegionBusyWait.testParallelAppendWithMemStoreFlush fails sporadically
> --------------------------------------------------------------------------
>
> Key: HBASE-11834
> URL: https://issues.apache.org/jira/browse/HBASE-11834
> Project: HBase
> Issue Type: Bug
> Components: test
> Reporter: stack
> Assignee: stack
> Attachments: 11834.txt
>
>
> See
> https://builds.apache.org/view/H-L/view/HBase/job/HBase-TRUNK/5431/testReport/junit/org.apache.hadoop.hbase.regionserver/TestHRegionBusyWait/testParallelAppendWithMemStoreFlush/
> Fails locally for me regularly. TIming. Turns out that on if we go to flush
> and no edits to flush, we can take the update lock and not release it.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
