[
https://issues.apache.org/jira/browse/SOLR-5823?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13924528#comment-13924528
]
Mark Miller commented on SOLR-5823:
-----------------------------------
Looks good. I wouldn't be surprised if you have to raise that 10 second timeout
- the jenkins cluster can be brutally slow ;)
We might also want to consider setting alive to false if either Overseer thread
ends. It may not be necessary - in the situations I've thought of so far, close
*should* end up being called relatively quickly after the threads end, but I
think there are some cases where they could end and close is not called until
there is a reconnection to zookeeper. We don't necessarily want a plugin
thinking its in a valid Overseer during that interval probably. But if say, the
state update thread ends, we know right away we are not a functioning Overseer
any longer. It might be safer to lock that knowledge in there.
> Add utility function for internal code to know if it is currently the overseer
> ------------------------------------------------------------------------------
>
> Key: SOLR-5823
> URL: https://issues.apache.org/jira/browse/SOLR-5823
> Project: Solr
> Issue Type: Improvement
> Reporter: Hoss Man
> Attachments: SOLR-5823.patch, SOLR-5823.patch, SOLR-5823.patch
>
>
> It would be useful if there was some Overseer equivalent to
> CloudDescriptor.isLeader() that plugins running in solr could use to know "At
> this moment, am i the leader?"
--
This message was sent by Atlassian JIRA
(v6.2#6252)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
