You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@cassandra.apache.org by "Brandon Williams (JIRA)" <ji...@apache.org> on 2014/06/17 23:45:21 UTC

[jira] [Commented] (CASSANDRA-7411) Node enables vnodes when bounced

    [ https://issues.apache.org/jira/browse/CASSANDRA-7411?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14034427#comment-14034427 ] 

Brandon Williams commented on CASSANDRA-7411:
---------------------------------------------

This is because once the initial_token is set in the system table, it's unused from the yaml afterward.  Upon rebooted the num_tokens takes precedence, as that would be the procedure to begin upgrading to vnodes.  You really shouldn't ever set both of these, especially with a single initial_token.

> Node enables vnodes when bounced
> --------------------------------
>
>                 Key: CASSANDRA-7411
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-7411
>             Project: Cassandra
>          Issue Type: Bug
>         Environment: OSX 9
>            Reporter: Philip Thompson
>         Attachments: system.log
>
>
> According to cassandra.yaml, in the information for the num_tokens setting, "Specifying initial_token will override this setting." So if exactly one initial token is set, then vnodes are disabled, regardless of if or what num_tokens are set to. This behavior is inconsistent when a node is started, versus if it has been bounced.
> From a fresh checkout of C*, if I build, then edit cassandra.yaml so that:
> num_tokens: 256
> initial_token: -9223372036854775808
> then run bin/cassandra, C* will start correctly. I can run bin/nodetool ring and see that the node has exactly one token and it is what I set in initial_token. If I gracefully shutdown C*, then restart the node, running bin/nodetool ring shows that the node now has vnodes enabled and has 256 tokens.
> I have been able to reproduce this locally on OSX using 2.0.8, 2.1 rc1, and trunk. I have not yet tested in Linux or Windows to see if it occurs there.



--
This message was sent by Atlassian JIRA
(v6.2#6252)