[
https://issues.apache.org/jira/browse/HIVE-17905?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16269849#comment-16269849
]
Prasanth Jayachandran commented on HIVE-17905:
----------------------------------------------
looks good to me, +1
Although the core change went in earlier, we can improve the latency of cluster
change notification.
Looks like we are polling for changes every 60s? This seems like a lot where
for about a minute we have stale info. Is that right?
Any reason not to use registry notifications for changes to llap cluster size
and then compute executor counts on every ZK notification?
> propagate background LLAP cluster changes to WM
> -----------------------------------------------
>
> Key: HIVE-17905
> URL: https://issues.apache.org/jira/browse/HIVE-17905
> Project: Hive
> Issue Type: Sub-task
> Reporter: Sergey Shelukhin
> Assignee: Sergey Shelukhin
> Attachments: HIVE-17905.patch
>
>
> HIVE-17841 already adds a method, it just needs to be called when there are
> relevant cluster changes that HS2 can detect
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
