You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@cassandra.apache.org by "Jignesh Dhruv (JIRA)" <ji...@apache.org> on 2010/07/20 23:40:52 UTC

[jira] Commented: (CASSANDRA-1291) nodetool ring prints incorrect IP address

    [ https://issues.apache.org/jira/browse/CASSANDRA-1291?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12890466#action_12890466 ] 

Jignesh Dhruv commented on CASSANDRA-1291:
------------------------------------------

So bug https://issues.apache.org/jira/browse/CASSANDRA-1304 has been marked duplicate of this bug.

But there is a major issue that I am seeing. The load is being divided among the nodes that are seen through nodetool ring command while some of the nodes in the cluster hardly have 2% of data. These are the nodes that are not visible through ring command.

When I run the loadbalance command, the data gets transferred again to a node that is visible in the ring command. The nodes that are not visible in the ring command has minimal of less than 10% of data. load-balance also doesn't help in this case.

Thus in my cluster of 7 nodes, 3 nodes has 90$ of data and other 4 nodes has 10% of data. And when I run loadbalance data keeps getting transferred within the top 3 nodes.

I am changing the priority to major. Again I am seeing this issue only with DatacenterShardStrategy.

Jignesh

> nodetool ring prints incorrect IP address
> -----------------------------------------
>
>                 Key: CASSANDRA-1291
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-1291
>             Project: Cassandra
>          Issue Type: Bug
>    Affects Versions: 0.7
>            Reporter: Brandon Williams
>            Priority: Minor
>             Fix For: 0.7
>
>
> Nodetool's ring output on trunk seems to duplicate an ipaddress instead of printing them all.
> To reproduce: spin up a 3 node cluster and create a keyspace, examine nodetool ring.  One ip address will show up twice, though the tokens are correctly displayed for the three machines.
> This is a cosmetic error, if you call getLiveNodes via JMX you can see all three IPs.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.