[
https://issues.apache.org/jira/browse/HBASE-12480?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14213320#comment-14213320
]
Virag Kothari commented on HBASE-12480:
---------------------------------------
Thanks for the quick review Jimmy on a Friday evening!
bq. If admin fixes the problem causing failed open/close, they can re-assign
the region from shell
In our case, we see regions going in failed open due to mainly two reasons
1) sometimes we have datanodes behaving badly causing regionservers to be
unable to open regions and lot of regions end up in FAILED_OPEN
2) Sometimes master cannot find a region plan after many attempts and makes the
region failed open (This is more common for us due to custom group stuff as
sometimes all region servers in the group are dead)
The admin could reassign from shell but the our admin likes restarting master
as a first try when it comes to any assignment issues. I think it would be good
to maintain that 0.94 behaviour.
bq. no need to change serverManager.isServerOnline(regionState.getServerName())
I think
isServerOnline(ServerName) will return false when serverName is null (It will
be null in case 2 above). This will cause
(!serverManager.isServerOnline(regionState.getServerName()) to be true causing
the method to return and will not proceed to the switch case.
bq. Are you sure this is an issue for 2.0.0?
No. But the test fails without the patch and passes with it. But the test could
be wrong. Can you do a check? Thanks!
> Regions in FAILED_OPEN/FAILED_CLOSE should be processed on master failover
> ---------------------------------------------------------------------------
>
> Key: HBASE-12480
> URL: https://issues.apache.org/jira/browse/HBASE-12480
> Project: HBase
> Issue Type: Bug
> Components: Region Assignment
> Reporter: Virag Kothari
> Assignee: Virag Kothari
> Fix For: 2.0.0, 0.98.9, 0.99.2
>
> Attachments: HBASE-12480.patch
>
>
> For zk assignment, we used to process this regions. For zk less assignment,
> we should do the same
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
