[
https://issues.apache.org/jira/browse/SOLR-12033?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16378377#comment-16378377
]
Shalin Shekhar Mangar commented on SOLR-12033:
----------------------------------------------
Attached test that demonstrates the issue. Pinging [~noble.paul].
> Autoscaling policy does not resolve all violations in one attempt
> -----------------------------------------------------------------
>
> Key: SOLR-12033
> URL: https://issues.apache.org/jira/browse/SOLR-12033
> Project: Solr
> Issue Type: Bug
> Security Level: Public(Default Security Level. Issues are Public)
> Components: AutoScaling, SolrCloud
> Affects Versions: 7.2
> Reporter: Shalin Shekhar Mangar
> Priority: Major
> Fix For: master (8.0), 7.3
>
> Attachments: SOLR-12033.patch
>
>
> I found a case where the computed plan did not resolve all violations even
> though it was possible.
> The use-case is as follows:
> # Cluster has 2 nodes
> # Collection has 2 shards, 2 replicas, created with maxShardsPerNode=2 so
> each node has exactly two cores
> # Add a cluster policy {{'cores': '<2', 'node' : '#ANY'}}. At this point
> there are exactly two violations (as returned by diagnostics API)
> # Use a trigger to call compute plan action (I used the metric trigger)
> # The plan computed has only 1 move replica operation, there by leaving 1
> violation behind
> With two move replica operations, all violations could have been corrected
> but for some reason that was not suggested.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
