[
https://issues.apache.org/jira/browse/SOLR-17616?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17912149#comment-17912149
]
David Smiley commented on SOLR-17616:
-------------------------------------
A simple alternative that's probably good enough could be for the
ClusterStateProvider to return a list of URLs for admin commands. The
HttpClusterStateProvider would simply return the configured nodes. The ZK one
could use live nodes.
an aside: I filed this purely as a wish/wouldn't-it-be-nice. I don't think
it's important or warrants sophistication. Simple is better. Nothing is fine
too :-)
> CloudSolrClient admin routing should consider node roles and randomization
> --------------------------------------------------------------------------
>
> Key: SOLR-17616
> URL: https://issues.apache.org/jira/browse/SOLR-17616
> Project: Solr
> Issue Type: Improvement
> Components: SolrCloud, SolrJ
> Reporter: David Smiley
> Priority: Minor
>
> CloudSolrClient delivers an admin request to the same node given a list of
> liveNodes (no ordering). If it's a V2 request, it'll be randomly chosen,
> which is a little better. If the non-Overseer so-called "distributed"
> cluster processing booleans are set, it's preferable to more thoughtfully
> distribute this work instead of consistently picking the same node. Node
> roles are an existing way of designating nodes for certain purposes.
> Instead, CloudSolrClient should pick randomly amongst the OVERSEER node
> roles, then amongst the COORDINATOR node roles, then amongst remaining.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@solr.apache.org
For additional commands, e-mail: issues-h...@solr.apache.org
