[ 
https://issues.apache.org/jira/browse/SOLR-8696?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15153226#comment-15153226
 ] 

Shalin Shekhar Mangar commented on SOLR-8696:
---------------------------------------------

Thanks Scott.

The ZkStateReader refresh is a safe guard to ensure that before processing any 
operation, the effects of previous operations are reflected in the local 
cluster state. This is less important now because most operations do wait for 
the cluster state to reflect their results (e.g. collection creation/deletion) 
etc.

I audited all overseer operations and the following operations do not have such 
a wait loop:
# ADDREPLICA does not wait for the replica to show up in the local cluster 
state (which may cause the same core name to be assigned again for a subsequent 
addReplica operation)
# ADDREPLICAPROP
# DELETEREPLICAPROP
# BALANCESHARDUNIQUE
# MODIFYCOLLECTION

Of all these only add replica should be fixed. I don't foresee any damage for 
the others.

> Optimize overseer + startup
> ---------------------------
>
>                 Key: SOLR-8696
>                 URL: https://issues.apache.org/jira/browse/SOLR-8696
>             Project: Solr
>          Issue Type: Improvement
>          Components: SolrCloud
>    Affects Versions: 5.4.1
>            Reporter: Scott Blum
>              Labels: patch, performance, solrcloud, startup
>         Attachments: SOLR-8696.patch
>
>
> ZkController.publishAndWaitForDownStates() occurs before overseer election.  
> That means if there is currently no overseer, there is ironically no one to 
> actually service the down state changes it's waiting on.  This particularly 
> affects a single-node cluster such as you might run locally for development.
> Additionally, we're doing an unnecessary ZkStateReader forced refresh on all 
> Overseer operations.  This isn't necessary because ZkStateReader keeps itself 
> up to date.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org

Reply via email to