[
https://issues.apache.org/jira/browse/HBASE-2458?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12857819#action_12857819
]
Todd Lipcon commented on HBASE-2458:
------------------------------------
In particular, HConnectionManager.locateRegionInMeta calls getCachedLocation
outside of the regionLockObject. Although this appears to be "read only", the
SoftValueSortedMap calls through to treemap.remove and rq.poll, both of which
are mutative.
Possible fixes: 1) add synchronization around getCachedLocation in
HCM.locateRegionInMeta, or 2) Add synchronization internal to
SoftValueSortedMap so that it becomes threadsafe.
[this appears to have been introduced by HBASE-2034]
> RS stuck in TreeMap,remove
> --------------------------
>
> Key: HBASE-2458
> URL: https://issues.apache.org/jira/browse/HBASE-2458
> Project: Hadoop HBase
> Issue Type: Bug
> Components: regionserver
> Affects Versions: 0.20.4
> Reporter: Todd Lipcon
> Priority: Blocker
> Attachments: stack.txt
>
>
> Testing 0.20_pre_durabil...@934691 one of my region servers got permanently
> stuck with one thread looping inside TreeMap.remove. See attached stack.
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
https://issues.apache.org/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira
