hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "stack (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-7724) [0.94] Just OPENED regions are ignored by ServerShutdownHandler and go unassigned for ever
Date Fri, 01 Feb 2013 01:20:11 GMT

    [ https://issues.apache.org/jira/browse/HBASE-7724?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13568337#comment-13568337

stack commented on HBASE-7724:

[~ram_krish] let me review hbase-7521 (though going by Enis's testing, it looks like the patch
has issues)...Thanks for taking a look.
> [0.94] Just OPENED regions are ignored by ServerShutdownHandler and go unassigned for
> ------------------------------------------------------------------------------------------
>                 Key: HBASE-7724
>                 URL: https://issues.apache.org/jira/browse/HBASE-7724
>             Project: HBase
>          Issue Type: Bug
>          Components: Region Assignment
>            Reporter: stack
> Visiting a user today, I came across following interesting case (0.94.2 HBase).
> A server was added to cluster.  It was assigned regions by the balancer.  A bunch opened
on the regionserver and just after the open, the regionserver was manually shutdown.  There
was a lag processing the zk region open events in the master (because clean shutdown, there
was no zk activity when the regions were closed on the shutdown regionserver).  Processing
the server shutdown, we do this for a good few of the regions that had just been opened on
the regionserver:
> 2013-01-30 02:41:19,917 INFO org.apache.hadoop.hbase.master.handler.ServerShutdownHandler:
Skip assigning region OBFUSCATED_TABLE,OBFUSCATED_STARTROW,1344723216908.55e9cb551edeea0b52bb91af7c2de199.
state=OPEN, ts=1359513674715, server=XX.XX.18.40,10304,1359513445136
> Seems like outright bug that'd we'd skip a region that is in transition that is in the
OPEN state.
> More detail to follow.

This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

View raw message