[
https://issues.apache.org/jira/browse/SOLR-11577?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18132076#comment-18132076
]
Nick Shanin commented on SOLR-11577:
------------------------------------
>From code reading against apache/main b73f03976fa, shard selection reads the
>cached cluster state
>(solr/core/src/java/org/apache/solr/handler/component/CloudReplicaSource.java:63).
> HttpShardHandler throws a 503 when no active server is listed
>(solr/core/src/java/org/apache/solr/handler/component/HttpShardHandler.java:538-539),
> with no retry in the handler. SolrJ's CloudSolrClient retries
>(solr/solrj/src/java/org/apache/solr/client/solrj/impl/CloudSolrClient.java:98-100),
> but direct HTTP callers do not. Is a query right after a node stop still
>failing with 503, with shards.tolerant unset?
🤖 AI-generated 🤖
> HttpShardHandler do not retry on stale clusterstate
> ---------------------------------------------------
>
> Key: SOLR-11577
> URL: https://issues.apache.org/jira/browse/SOLR-11577
> Project: Solr
> Issue Type: Bug
> Reporter: Cao Manh Dat
> Assignee: Cao Manh Dat
> Priority: Major
>
> Found this problem when doing SOLR-11576. The case here is
> 1. collection1 has 2 shards and 1 replicas each (on node1 and node2)
> 2. restart node2
> 3. do indexing then commit using CloudSolrClient
> 4. assert that docs are found
> Step 3 is always success, cause CloudSolrClient will do retry requesting on
> stale clusterstate (SOLR-11484). But HttpShardHandler does not. So there is a
> chance that we will meet an assertion error on step 4 because
> HttpShardHandler will throw a SolrException if it can not found an alive
> replica for a shard ( no servers hosting shard: ... )
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
