[
https://issues.apache.org/jira/browse/CASSANDRA-7710?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14095343#comment-14095343
]
Jimmy Mårdell commented on CASSANDRA-7710:
------------------------------------------
Our particular problem is that we have created our own SeedProvider. When you
run "nodetool ring" on a production machine, our custom SeedProvider gets
invoked by the nodetool process (as a side effect of DatabaseDescriptor being
invoked) which causes some minor issues. This seems like a bug to me; nodetool
ought to be a pure CLI fetching data through JMX.
Compare the 2.0 implementation (using DatabaseDescriptor.getNumTokens() to
determine if the cluster uses vnodes):
https://github.com/apache/cassandra/blob/cassandra-2.0/src/java/org/apache/cassandra/tools/NodeCmd.java#L296
and the 2.1 implementations (which determines if the cluster uses vnodes by
counting number of tokens through JMX)
https://github.com/apache/cassandra/blob/cassandra-2.1/src/java/org/apache/cassandra/tools/NodeTool.java#L437
The 2.1 implementation is clearly superior and doesn't have side effects. I
think this should be backported 2.0. I might do this myself, since it doesn't
seem to so hard.
> nodetool ring throws exception if run on machine without Cassandra
> ------------------------------------------------------------------
>
> Key: CASSANDRA-7710
> URL: https://issues.apache.org/jira/browse/CASSANDRA-7710
> Project: Cassandra
> Issue Type: Bug
> Components: Tools
> Reporter: Jimmy Mårdell
> Assignee: Michael Shuler
> Priority: Minor
>
> DatabaseDescriptor.getNumTokens() is invoked in the nodetool ring command
> which doesn't work so well when running on a machine where Cassandra doesnt'
> exist. And it has all kind of side effects as well.
> This seems fixed in 2.1 but would be nice if it was fixed in 2.0 as well.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
