[
https://issues.apache.org/jira/browse/SOLR-6411?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14107610#comment-14107610
]
Mark Miller commented on SOLR-6411:
-----------------------------------
Rare hiccups that occur around the same time for all replicas can leave a shard
in a state where no thinks they can become the leader and the only safe
recovery is to restart the whole shard.
We should be able to do better than this - we should be able to simulate
something similar to restarting the shard as long as all replicas look like
they are up and will participate in the leader election. For instance, if it
looks to a candidate that all replicas for a shard are down but on live nodes,
it could try and become leader even though it currently would think it cannot
and if it's peer sync can talk to all the replicas, we should allow it to
become the leader even though it wasn't previously active.
> A shard should be able to recover even if all nodes don't think they can if
> all replicas in a shard appear to be involved.
> --------------------------------------------------------------------------------------------------------------------------
>
> Key: SOLR-6411
> URL: https://issues.apache.org/jira/browse/SOLR-6411
> Project: Solr
> Issue Type: Improvement
> Components: SolrCloud
> Reporter: Mark Miller
> Assignee: Mark Miller
>
--
This message was sent by Atlassian JIRA
(v6.2#6252)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
