hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "rajeshbabu (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-8705) RS holding META when restarted in a single node setup may hang infinitely without META assignment
Date Mon, 22 Jul 2013 06:50:50 GMT

    [ https://issues.apache.org/jira/browse/HBASE-8705?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13714954#comment-13714954
] 

rajeshbabu commented on HBASE-8705:
-----------------------------------

[~ram_krish],[~jxiang]
if we stop all region servers by hbase-deamon.sh stop command then .META. is never online
with 10 seconds delay in starting a region server.
In single node cluster, many times .META. in transition after stopping the RS.
Can we retry meta assignment even after max attempts until we find some online region server.
Otherwise we need to run hbck everytime in this common case.
What do you say?
 


                
> RS holding META when restarted in a single node setup may hang infinitely without META
assignment
> -------------------------------------------------------------------------------------------------
>
>                 Key: HBASE-8705
>                 URL: https://issues.apache.org/jira/browse/HBASE-8705
>             Project: HBase
>          Issue Type: Bug
>    Affects Versions: 0.95.1
>            Reporter: ramkrishna.s.vasudevan
>            Assignee: ramkrishna.s.vasudevan
>            Priority: Minor
>             Fix For: 0.98.0, 0.95.2
>
>         Attachments: HBASE-8705_1.patch, HBASE-8705_2.patch, HBASE-8705.patch
>
>
> This bug may be minor as it likely to happen in a single node setup.
> I restarted the RS holding META. The master tried assigning META using MetaSSH. But tried
this before the new RS came up.
> So as not region plan is found 
> {code}
>  if (plan == null) {
>         LOG.warn("Unable to determine a plan to assign " + region);
>         if (tomActivated){
>           this.timeoutMonitor.setAllRegionServersOffline(true);
>         } else {
>           regionStates.updateRegionState(region, RegionState.State.FAILED_OPEN);
>         }
>         return;
>       }
> {code}
> we just return without assigment.  And this being the META the small cluster just hangs.

--
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

Mime
View raw message