[jira] [Issue Comment Edited] (CASSANDRA-2393) Ring changes should stream from the replicas that are losing responsibility

["Chris Goffinet (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/CASSANDRA-2393?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13011912#comment-13011912 ] 

Chris Goffinet edited comment on CASSANDRA-2393 at 3/28/11 4:29 AM:
--------------------------------------------------------------------

>From what I can tell, this should only effect counters w/ Repair on Write turned off. We just need to stream from the replica that is leaving.

      was (Author: lenn0x):
    From what I can tell, this should only effect counters w/ Repair on Write turned off. We should really stream from all replicas if you have ROW turned off..
  
> Ring changes should stream from the replicas that are losing responsibility
> ---------------------------------------------------------------------------
>
>                 Key: CASSANDRA-2393
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-2393
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Core
>            Reporter: Stu Hood
>            Priority: Minor
>
> During a bootstrap or move operation, the replica that is leaving the replica set should be the preferred node to stream data to the node that is joining the replica set. Currently, an effectively random node will be chosen.
> Since the data held by the leaving node may not agree with the other replicas, it is important for consistency that it be the one that streams.

--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira