[
https://issues.apache.org/jira/browse/HBASE-3403?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12976168#action_12976168
]
Todd Lipcon commented on HBASE-3403:
------------------------------------
I think the RS crashed somewhere between offlining the parent and opening one
of the new daughters, where this comment is:
{code}
// The is the point of no return. We are committed to the split now. We
// have still the daughter regions to open but meta has been changed.
// If we fail from here on out, we can not rollback so, we'll just abort.
// The meta has been changed though so there will need to be a fixup run
// during processing of the crashed server by master (TODO: Verify this in
place).
{code}
When the master processed the RS failure, the daughter region wasn't yet in
META, so it didn't count them as unassigned during processing the server
shutdown.
Perhaps MetaReader.getServerUserRegions() needs to take into account offline
parents. If it finds an offline parent, it should add the two daughters to a
list of potentially mid-split regions. Then it can issue a get() for each of
those regions, and if it doesn't find a row, it may need to be counted as
unassigned... seems a bit messy :)
> Region got undeployed
> ---------------------
>
> Key: HBASE-3403
> URL: https://issues.apache.org/jira/browse/HBASE-3403
> Project: HBase
> Issue Type: Bug
> Affects Versions: 0.90.0
> Reporter: Todd Lipcon
> Fix For: 0.90.0
>
> Attachments: broken-split.txt
>
>
> ERROR: Region
> hdfs://haus01.sf.cloudera.com:11020/hbase-normal/usertable/2ad8df700eea55f70e02ea89178a65a2
> on HDFS, but not listed in META or deployed on any region server.
> ERROR: Found inconsistency in table usertable
> Not sure how I got into this state, will look through logs.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
