[
https://issues.apache.org/jira/browse/HBASE-12630?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14240494#comment-14240494
]
Manukranth Kolloju commented on HBASE-12630:
--------------------------------------------
This problem wasn't observed in any prod system of ours. [~tedyu] suggested
that it would be nice to have to handle the cases where the observers might hit
an exception and stay in that case. This was a speculative case, but the idea
wasn't bad. But, the priority of the Jira should have not been MAJOR/MINOR. I
took a look at the patch, I don't really think that we should be storing this
extra map for the sake of a speculative case.
> Provide capability for dropping malfunctioning ConfigurationObserver
> automatically
> ----------------------------------------------------------------------------------
>
> Key: HBASE-12630
> URL: https://issues.apache.org/jira/browse/HBASE-12630
> Project: HBase
> Issue Type: Improvement
> Reporter: Ted Yu
> Assignee: Ashish Singhi
> Priority: Minor
> Fix For: 1.0.0, 2.0.0
>
> Attachments: HBASE-12630-v2.patch, HBASE-12630-v3.patch,
> HBASE-12630.patch
>
>
> Chatting with Manukranth, online config update feature should handle the case
> where certain ConfigurationObserver(s) constantly produce exception when
> notified of config update.
> We can handle such ConfigurationObserver by sidelining it after configurable
> number of exceptions seen from the ConfigurationObserver.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
