[jira] Commented: (HBASE-500) Regionserver stuck on exit

["Jim Kellerman (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HBASE-500?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12589270#action_12589270 ] 

Jim Kellerman commented on HBASE-500:
-------------------------------------

In the thread dump, the cacheFlusher thread is not present, so
HRegionServer.run has interrupted it.

HRegionServer.run is now waiting to enter the synchronized block
around logRollerLock which is held by the logRoller thread. This means
that HRegionServer.run has interrupted the compactSplit thread (which
does not protect itself using compactSplitLock as it should while it
is working.

The logRoller, HRegionServer$Worker and compactSplit threads are
waiting to acquire HLog.cacheFlushLock It is not clear why one of the
threads cannot proceed as it appears that there is no thread that
could currently be holding that lock.

Certainly, interrupting the compactSplit thread while it is working is
an error. Creating a patch that will fix this error and see if this
lock up occurs after it is committed.

> Regionserver stuck on exit
> --------------------------
>
>                 Key: HBASE-500
>                 URL: https://issues.apache.org/jira/browse/HBASE-500
>             Project: Hadoop HBase
>          Issue Type: Bug
>          Components: regionserver
>    Affects Versions: 0.16.0, 0.2.0, 0.1.0
>            Reporter: stack
>            Assignee: Jim Kellerman
>         Attachments: stacktrace.txt
>
>
> Found in 0.16.0 cluster.  We're rolling a log and trying to split it too.  Looks like hung up locks.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.