[
https://issues.apache.org/jira/browse/SOLR-16453?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17614089#comment-17614089
]
Justin Sweeney commented on SOLR-16453:
---------------------------------------
This is a bug in ZkStateWriter that can occur when using both PRS and non-PRS
collections in the same cluster. The issue here is that PRS updates through
overseer are not batched, but non-PRS updates are batched. When using both PRS
and non-PRS in the same cluster, a PRS collection update through overseer will
be immediately written to ZK and will reset the counters that control when to
flush batched updates without actually flushing the batched updates. The result
is that updates that were batched can be skipped entirely.
I am opening a PR now that will fix this by not resetting the batch counters
when immediately writing a PRS update.
> overseer doesn't handle prs messages properly
> ---------------------------------------------
>
> Key: SOLR-16453
> URL: https://issues.apache.org/jira/browse/SOLR-16453
> Project: Solr
> Issue Type: Bug
> Security Level: Public(Default Security Level. Issues are Public)
> Components: SolrCloud
> Affects Versions: 9.1
> Reporter: Hitesh Khamesra
> Priority: Blocker
>
> Overseer thread doesn't process prs message properly. It gives preference to
> prs message, but then it drops other messages at that time.
>
> More detail needs to add [~jsweene2]
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
