[
https://issues.apache.org/jira/browse/SOLR-14341?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17061346#comment-17061346
]
David Smiley commented on SOLR-14341:
-------------------------------------
There is obviously a back-compat concern here. I propose new collections have
the name added to state.json in 8x redundantly with the older location. On
lookup, we only need to read state.json if it's there, else fall back to the
legacy location. The fallback can trigger an immediate update to state.json.
In master, only this new location will be accessed. This will require users to
upgrade to a recent 8x version before proceeding to 9x.
> Move a collection's configSet name to state.json
> ------------------------------------------------
>
> Key: SOLR-14341
> URL: https://issues.apache.org/jira/browse/SOLR-14341
> Project: Solr
> Issue Type: Improvement
> Security Level: Public(Default Security Level. Issues are Public)
> Components: SolrCloud
> Reporter: David Smiley
> Priority: Major
>
> It's a bit odd that a collection's state.json knows everything about a
> collection except for perhaps the most important pointer -- the configSet
> name. Presently the configSet name is retrieved via
> {{ZkStateReader.getConfigName(collectionName)}} which looks at the zk path
> {{/collections/collectionName}} (an intermediate node) interpreted as a
> trivial JSON object. Combining the configSet name into state.json is simpler
> and also more efficient since many calls to grab the configset name _already_
> need the state.json (via a DocCollection object).
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
