[
https://issues.apache.org/jira/browse/SOLR-7065?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14299052#comment-14299052
]
Mark Miller commented on SOLR-7065:
-----------------------------------
bq. You mentioned at one point that you wondered whether the whole "watch the
guy in front" and ZKs ephemeral-sequential node was the right way to approach
this.
Right. This entire approach is an elegant ZooKeeper recipe that is actually
quite difficult to program perfectly. It's point is to prevent a thundering
herd problem when you have tons of nodes involved in the election - with
simpler approaches, if a leader goes down, now you can have everyone in the
election checking the same nodes about what has changed and this can cause
problems. Except that you never have more than a handful of replicas. Even 20
replicas is kind of crazy, and it's still not even close to a herd.
This elegant solution is hard to nail, hard to test properly, and as can be
seen, not very good for dealing with priorities and altering the election line.
A very simple solution that involves the overseer or optimistic locking /
writing would be much, much simpler for re ordering the election.
> Let a replica become the leader regardless of it's last published state if
> all replicas participate in the election process.
> ----------------------------------------------------------------------------------------------------------------------------
>
> Key: SOLR-7065
> URL: https://issues.apache.org/jira/browse/SOLR-7065
> Project: Solr
> Issue Type: Improvement
> Reporter: Mark Miller
> Assignee: Mark Miller
> Attachments: SOLR-7065.patch, SOLR-7065.patch
>
>
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
