[ 
https://issues.apache.org/jira/browse/HBASE-5362?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13204823#comment-13204823
 ] 

Ionut Ignatescu commented on HBASE-5362:
----------------------------------------

No,I can't. We have a different version deployed on our Hadoop cluster, so it's 
almost impossible to perform this test.
Anyway, thanks for your suggestion, I'll look into source to check.
                
> ZK connection leak
> ------------------
>
>                 Key: HBASE-5362
>                 URL: https://issues.apache.org/jira/browse/HBASE-5362
>             Project: HBase
>          Issue Type: Bug
>          Components: mapreduce
>    Affects Versions: 0.90.3
>            Reporter: Ionut Ignatescu
>
> I have an application like a scheduler that starts periodically some MR jobs 
> that reads from one HBase table and write in more tables, changing the row 
> key and the columns list. 
> I have encounted a problem with this: after each MR job, I have +1 
> connections open to ZK and after a period of time, I have received 
> IOException. I saw that I am not the first that had had this problem, but I 
> didn't see any fix. No in HBase 0.90.4, nor in HBase 0.92.
> To fix this issue I wrote a custom TableInputFormat that forces to close the 
> connection to ZK after each getSplits call. 
> Also, the initTableMapperJob was overwritten.
> I found an open JIRA issue here: 
> https://issues.apache.org/jira/browse/HBASE-3792
> Regards,
> Ionut I. 

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: 
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

Reply via email to