hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Shrijeet Paliwal (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-6660) Meta assignment to a region server caused continuous NPE loop during postOpenDeployTasks
Date Tue, 28 Aug 2012 19:01:08 GMT

    [ https://issues.apache.org/jira/browse/HBASE-6660?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13443397#comment-13443397
] 

Shrijeet Paliwal commented on HBASE-6660:
-----------------------------------------

Sorry for late response. 

Response to Enis's question: DNS issue and hostname mixup is unlikely. DNS is well setup and
has been working reliably since long time. 

Stack, I do not have enough evidence and confidence to label this as blocker. I did nasty
things (killed processes out of order, deleted zookeeper data etc. etc.) It could be a pebkac
issue. 

I am planning to upgrade one more data center today/tomorrow. I will use the same HBase checkout
as this bug report. Upto you guys if you want to wait for my report. 

Thanks.
                
> Meta assignment to a region server caused continuous NPE loop during postOpenDeployTasks
> ----------------------------------------------------------------------------------------
>
>                 Key: HBASE-6660
>                 URL: https://issues.apache.org/jira/browse/HBASE-6660
>             Project: HBase
>          Issue Type: Bug
>    Affects Versions: 0.92.2
>         Environment: CentOS release 5.7
>            Reporter: Shrijeet Paliwal
>            Priority: Blocker
>
> Recently I upgraded three data centers to our own checkout of 0.92.2, last commit :
> {noformat}
> commit 5accb6a1be4776630126ac21d07adb652b74df95
> Author: Zhihong Yu <tedyu@apache.org>
> Date:   Mon Aug 20 18:19:45 2012 +0000
> HBASE-6608 Fix for HBASE-6160, META entries from daughters can be deleted before parent
entries, shouldn't compare HRegionInfo's (Enis)
> {noformat}
> Two upgrades went fine, upgrade to one data center failed. Failed in the sense that ROOT
and META assignment took forever. Panic struck I restarted master and all region servers.
I may have deleted zookeeper node /hbase/root-region-server as well, dont ask me why :-( 
> After this I managed to get ROOT assigned. But META assignment got stuck again. 
> The log is here : https://raw.github.com/gist/3455435/adebd118b47aa3d715201010aa09e5eb8930033c/npe_rs_0.92.2.log
> Notice how region server was stuck in a loop of NPE (grep processBatchCallback). There
is one more NPE related to zookeeper constructor. 

--
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