[
https://issues.apache.org/jira/browse/CASSANDRA-2991?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13079303#comment-13079303
]
Sylvain Lebresne commented on CASSANDRA-2991:
---------------------------------------------
bq. you can't use sstableloader from an existing cluster member
Well, you can using the JMX bulkload call
bq. plus it has to rebuild secondaries
True. Though I do admit that I think sstableloader will be simpler in most case
of "moving sstables between clusters", especially if the destination cluster
doesn't have the same number of node/replication factor (as you don't have to
think which sstable should be copied where and avoid both cleanup and repair),
and thus wonder if a 'load new sstables' command would be so useful. Another
way to put it is that I think a 'load new sstables' command will be more hacky,
in that you will have to run it on every nodes, know which sstable goes where
and not forget the cleanup and repair in most case, so I would be in favor of
not adding such hacky command unless it really makes a difference.
> Add a 'load new sstables' JMX/nodetool command
> ----------------------------------------------
>
> Key: CASSANDRA-2991
> URL: https://issues.apache.org/jira/browse/CASSANDRA-2991
> Project: Cassandra
> Issue Type: New Feature
> Reporter: Brandon Williams
> Priority: Minor
> Fix For: 0.8.4
>
>
> Sometimes people have to create a new cluster to get around a problem and
> need to copy sstables around. It would be convenient to be able to trigger
> this from nodetool or JMX instead of doing a restart of the node.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
