[
https://issues.apache.org/jira/browse/SOLR-18131?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18119547#comment-18119547
]
Jan Høydahl commented on SOLR-18131:
------------------------------------
You could of course filter the envent-stream on a list of collections too, but
it is still an active connection and transferred data that may not be needed.
Of course a ClusterStateProvider could be smarter and disconnect from the
event-stream as long as it does not need the state for routing decisions, and
then re-connect on demand, requesting only the set of events that have changed
since that timestamp. That would be more light weight. But you are right, it is
best suited for a system needing real time info, such as a dashboard.
> SolrCloud should return HTTP header(s) of certain ZK node versions
> ------------------------------------------------------------------
>
> Key: SOLR-18131
> URL: https://issues.apache.org/jira/browse/SOLR-18131
> Project: Solr
> Issue Type: Sub-task
> Reporter: David Smiley
> Priority: Major
>
> (see parent issue) Solr should return an HTTP header of some ZK Stat version
> numbers of pertinent ZK nodes. Relates to ClusterState tracking. This
> enables a client to invalidate a cache of related information (perhaps
> CloudSolrClient / HttpClusterStateProvider) or similar for custom clients not
> using SolrJ.
> It could be JSON to hold multiple, or we use multiple headers. Need to
> decide on suitable name(s).
> All SolrCloud interaction can use this:
> live_nodes: ZK cversion (initial 'c' is for children)
> Interactions involving one (typical) or more collections can incorporate:
> collectionName: ZK version
> Interactions involving alias resolution can incorporate:
> aliases: ZK version
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
