[
https://issues.apache.org/jira/browse/HBASE-7838?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13579081#comment-13579081
]
nkeywal commented on HBASE-7838:
--------------------------------
Imho, there are 4 distinct points.
- If seems reasonable to be able to kill 9 automatically the process if it
can't stop by itself. That's also what we do if memory consumption goes crazy.
That's this JIRA.
- As JM pointed it out, there is some code duplication in the scripts, and it
would be better to have something cleaner. Imho, except if 0.94 backport is not
a concern, it's better to have this in another JIRA, if someone wants to fix it.
- The root issue (why we can't stop) deserves a JIRA as well.
- Lastly, we should have unit tests for the various stop scenarios (during a
split, with .meta. regionserver holder shutdown first, ...). This could make it
in another JIRA as well.
I think it's reasonable :-), so if nobody disagrees, I will fix the point
mentioned by [~rajesh23] to close this JIRA for trunk.
I will also look at the unit tests part to see where we stand on trunk, and I
will create some JIRAs if I find something. It will be in my low priority stuff
list, so if someone else wants to do it, it's fine for me :-).
> HBase regionserver never stops when running `hbase-daemon.sh stop
> regionserver`
> -------------------------------------------------------------------------------
>
> Key: HBASE-7838
> URL: https://issues.apache.org/jira/browse/HBASE-7838
> Project: HBase
> Issue Type: Bug
> Components: regionserver, scripts, shell
> Affects Versions: 0.92.1, 0.96.0, 0.94.4
> Environment: Using CDH4.1.2
> Reporter: Damien Hardy
> Assignee: Damien Hardy
> Labels: newbie
> Fix For: 0.96.0
>
> Attachments: HBASE-7838-0.94.patch, HBASE-7838-trunk.patch
>
>
> Currently running full stack (Hadoop + Hbase + zookeeper) on a VM I want to
> reboot, every services stopped but HBase regionserver. For 2 days now, it's
> aligning dots on the console ...
> I propose a timeout (default 600 overidable by $HBASE_STOP_TIMEOUT) to force
> kill -9 on the process.
--
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
