[ 
https://issues.apache.org/jira/browse/SOLR-17077?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17793333#comment-17793333
 ] 

ASF subversion and git services commented on SOLR-17077:
--------------------------------------------------------

Commit 5f7c68514416ffdc52b164799f378b3a16fe7f1f in solr's branch 
refs/heads/main from Pierre Salagnac
[ https://gitbox.apache.org/repos/asf?p=solr.git;h=5f7c6851441 ]

SOLR-17077: Don't cancel twice when rejoining shard leader election (#2084)

When a shard rejoins leader election, leave previous election only once to save 
unneeded calls to Zookeeper.  The prior behavior was noisy and less efficient; 
a bug perhaps albeit without real consequence.

Co-authored-by: Pierre Salagnac <[email protected]>

> When rejoining shard leader election, we try to leave the election twice
> ------------------------------------------------------------------------
>
>                 Key: SOLR-17077
>                 URL: https://issues.apache.org/jira/browse/SOLR-17077
>             Project: Solr
>          Issue Type: Bug
>      Security Level: Public(Default Security Level. Issues are Public) 
>          Components: SolrCloud
>    Affects Versions: 8.11.2, 9.4
>            Reporter: Pierre Salagnac
>            Priority: Minor
>          Time Spent: 1h 10m
>  Remaining Estimate: 0h
>
> When rejoining leader election, we tries to delete the leader election twice 
> in Zookeeper.
> This has no impact on the outcome of the leader election and this does not 
> cause other failure. This is visible in logs (see bellow).
>  
> *Scénario details*
> Method {{ZkController.rejoinShardLeaderElection()}} leaves the election by 
> removing the sequential election node and the leader registration node. This 
> is done directly from there by invoking {{{}prevContext.cancelElection(){}}}.
> Then {{LeaderElector.retryElection()}} does that again 
> (this.context.cancelElection()) but the second time around always fails as 
> the Zookeeper nodes were already removed by the first attempt.
> Following log line is logged each time we invoke {{REJOINELECTION}} of a 
> shard.
> _"No version found for ephemeral leader parent node, won't remove previous 
> leader registration."_



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

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

Reply via email to