[
https://issues.apache.org/jira/browse/HBASE-7369?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13543001#comment-13543001
]
Bryan Baugher commented on HBASE-7369:
--------------------------------------
[~zjushch]
I like the idea of moving the logic into close() and will do that shortly.
The reason I created this issue and patch is because I was fighting an issue
where our HConnection would become aborted and no matter how many times we
closed and retrieved new HTables the connection was still aborted (because of
the caching mechanism). With this patch that should no longer be true.
My thought is that HTables should be short lived or it should be understandable
that if an HTable throws an exception one should close it and retrieve a new
one, which would fix the issue.
> HConnectionManager should remove aborted connections
> ----------------------------------------------------
>
> Key: HBASE-7369
> URL: https://issues.apache.org/jira/browse/HBASE-7369
> Project: HBase
> Issue Type: Improvement
> Components: Client
> Affects Versions: 0.94.3
> Reporter: Bryan Baugher
> Priority: Minor
> Attachments: patch2.diff, patch3.diff, patch.diff
>
>
> When an HConnection is abort()'ed (i.e. if numerous services are lost) the
> connection becomes unusable. HConnectionManager cache of HConnections
> currently does not have any logic around removing aborted connections
> automatically. Currently it is up to the consumer to do so using
> HConnectionManager.deleteStaleConnection(HConnection).
--
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
