lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Cao Manh Dat (JIRA)" <>
Subject [jira] [Commented] (SOLR-12110) Replica which failed to register in Zk can become leader
Date Fri, 16 Mar 2018 01:55:00 GMT


Cao Manh Dat commented on SOLR-12110:

The idea of the patch is calling ZkController.unregister when an exception is thrown by ZkController.register

> Replica which failed to register in Zk can become leader
> --------------------------------------------------------
>                 Key: SOLR-12110
>                 URL:
>             Project: Solr
>          Issue Type: Bug
>      Security Level: Public(Default Security Level. Issues are Public) 
>    Affects Versions: 7.3
>            Reporter: Cao Manh Dat
>            Assignee: Cao Manh Dat
>            Priority: Blocker
>         Attachments: SOLR-12110.patch
> In case of when an exception is thrown in ZkController.register() a replica can still
be able to joinElection and become leader after that. This will cause many problems, one
of the problems can happen (the patch including a test which lead to this failure) is
> A replica with DOWN state can become a leader and the shard will be stuck in this state
forever until the replica is removed or the node contains the replica is restarted.
> This won't be a problem in Solr 7.2.1 because a replica with last published state = DOWN
can't become a leader, only since SOLR-7034 get resolved (by SOLR-12011)

This message was sent by Atlassian JIRA

To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message