[ 
https://issues.apache.org/jira/browse/SOLR-6261?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14077508#comment-14077508
 ] 

Ramkumar Aiyengar commented on SOLR-6261:
-----------------------------------------

bq. Ah, the original pull request info in the first JIRA comment has links that 
are all kept up to date as you make changes. Did not realize. I'll commit this 
change shortly.

Yeah, it would be nice if commits to a pull request triggered a JIRA update. 
Worth an INFRA ticket? I could raise one, but wasn't sure if generally only 
committers raised them..

bq. Do you think the default keepAliveTime of 60 seconds is the right choice?

It seemed sensible and I didn't have much of an opinion on a more specific 
value. We certainly do want to reap these threads because the changes don't 
happen all the time, and they would occur within 60s of one another in 
predictable cases I could think of (ephemeral node expiry, overseer queue 
changes due to instances changing state). Other cases like LIR and admin tasks 
done through Overseer are more ad hoc anyway and I doubt we can associate any 
meaningful parameters modelling their use..

> Run ZK watch event callbacks in parallel to the event thread
> ------------------------------------------------------------
>
>                 Key: SOLR-6261
>                 URL: https://issues.apache.org/jira/browse/SOLR-6261
>             Project: Solr
>          Issue Type: Improvement
>          Components: SolrCloud
>    Affects Versions: 4.9
>            Reporter: Ramkumar Aiyengar
>            Assignee: Mark Miller
>            Priority: Minor
>
> Currently checking for leadership (due to the leader's ephemeral node going 
> away) happens in ZK's event thread. If there are many cores and all of them 
> are due leadership, then they would have to serially go through the two-way 
> sync and leadership takeover.
> For tens of cores, this could mean 30-40s without leadership before the last 
> in the list even gets to start the leadership process. If the leadership 
> process happens in a separate thread, then the cores could all take over in 
> parallel.



--
This message was sent by Atlassian JIRA
(v6.2#6252)

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to