[ 
https://issues.apache.org/jira/browse/SOLR-10935?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16061047#comment-16061047
 ] 

Erick Erickson commented on SOLR-10935:
---------------------------------------

Part of the new autoscaling work will superseded this.

Oh, and BALANCESHARDUNIQUE does _not_ change the actual leadership role, it 
simply distributes the "preferredLeader" _property_. You must issue the 
REBALANCELEADERS API command to actually change leadership to all the nodes 
with the "preferredLeader" property set to true.

If you look at the "preferredLeader" property in the state.json file associated 
with that collection, is it evenly distributed?

> BALANCESHARDUNIQUE does not distribute properties correctly
> -----------------------------------------------------------
>
>                 Key: SOLR-10935
>                 URL: https://issues.apache.org/jira/browse/SOLR-10935
>             Project: Solr
>          Issue Type: Bug
>      Security Level: Public(Default Security Level. Issues are Public) 
>          Components: SolrCloud
>    Affects Versions: 6.2
>            Reporter: Daisy.Yuan
>
> Create a collection of 8 slices on 4 nodes, 2 replicas of each slice.
> Node IP is:
> 192.168.182.246:21101
> 192.168.182.247:21104
> 192.168.182.248:21101
> 192.168.182.149:21104
> After executing the BALANCESHARDUNIQUE command, the leader node is balanced 
> as follows:
> Cloud Graph (* Leader)
> shard1 |- * 192.168.182.248:21101
>             |-   192.168.182.247.21104
> shard2 |- * 192.168.182.249:21104
>             |-   192.168.182.246:21101
> shard3 |-   192.168.182.247:21104
>             |- * 192.168.182.246:21101
> shard4 |-   192.168.182.248:21101
>             |- * 192.168.182.249:21104
> shard5 |-   192.168.182.249:21104
>             |- * 192.168.182.246:21101
> shard6 |- * 192.168.182.248:21101
>             |-   192.168.182.247:21104
> shard7 |-   192.168.182.248:21101
>             |- * 192.168.182.249:21104
> shard8 |- * 192.168.182.247:21104
>             |-   192.168.182.246:21101
> The correct expected result should be that there are two leader on each node.
> But the actual result is..
>   there are 3 leaders on 192.168.182.249:21104,
>   and only one Leader on 192.168.182.247:21104
>   the others are distributed correctly.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

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

Reply via email to