[
https://issues.apache.org/jira/browse/SOLR-9765?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18132078#comment-18132078
]
Nick Shanin commented on SOLR-9765:
-----------------------------------
>From code reading against apache/main b73f03976fa, if one replica fails to
>reload after a schema change, the error is returned but the new field stays in
>ZooKeeper and on the coordinator
>(solr/core/src/java/org/apache/solr/schema/SchemaManager.java:149-173, 186;
>solr/core/src/java/org/apache/solr/schema/ManagedIndexSchema.java:293-297).
>There is no rollback. Is this still the case, or was it covered by SOLR-9751,
>which is resolved?
🤖 AI-generated 🤖
> In SolrCloud mode, schema modification should fail the operation and rollback
> to previous schema on all replicas if any replica fails to load the new schema
> ------------------------------------------------------------------------------------------------------------------------------------------------------------
>
> Key: SOLR-9765
> URL: https://issues.apache.org/jira/browse/SOLR-9765
> Project: Solr
> Issue Type: Bug
> Reporter: Steven Rowe
> Priority: Major
>
> When a schema modification request succeeds on the coordinating replica but
> fails on one or more other replicas, the request should fail and the schema
> should be rolled back to the pre-request schema on all replicas.
> An example of where this should have happened but didn't: SOLR-9751.
> Another scenario where this might occur: if nodes don't all have the same
> classpath, schema modifications might succeed on one node but not on another.
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
