You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@cassandra.apache.org by "Terje Marthinussen (JIRA)" <ji...@apache.org> on 2011/06/28 03:17:17 UTC
[jira] [Created] (CASSANDRA-2834) Avoid repair getting started
twice at the same time for the same CF
Avoid repair getting started twice at the same time for the same CF
-------------------------------------------------------------------
Key: CASSANDRA-2834
URL: https://issues.apache.org/jira/browse/CASSANDRA-2834
Project: Cassandra
Issue Type: Improvement
Reporter: Terje Marthinussen
It may seem like it is possible to start repair twice at the same time on the same CF?
Not 100% verified, but if this is indeed the case, we may want to consider avoiding that including making nodetool repair abort and return and error if repair is attempted on the same CF as one which already have repair running.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (CASSANDRA-2834) Avoid repair getting started
twice at the same time for the same CF
Posted by "Sylvain Lebresne (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/CASSANDRA-2834?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13056399#comment-13056399 ]
Sylvain Lebresne commented on CASSANDRA-2834:
---------------------------------------------
bq. It may seem like it is possible to start repair twice at the same time on the same CF?
It is possible. Right now the only cases where we abort a repair quickly is if some neighbors are dead.
Repairing twice on the same CF is indeed useless and we can try to avoid it. This is however not totally trivial because the two repairs can be started on different nodes so we'll have to synchronize somehow. Rather, it's not hard per se, but this will require some addition to the network protocol and is thus a little longer term that one could hope.
Note that this may be made simpler by CASSANDRA-1740 in that it would propose to have a way to abort a repair (which don't have so far).
> Avoid repair getting started twice at the same time for the same CF
> -------------------------------------------------------------------
>
> Key: CASSANDRA-2834
> URL: https://issues.apache.org/jira/browse/CASSANDRA-2834
> Project: Cassandra
> Issue Type: Improvement
> Reporter: Terje Marthinussen
>
> It may seem like it is possible to start repair twice at the same time on the same CF?
> Not 100% verified, but if this is indeed the case, we may want to consider avoiding that including making nodetool repair abort and return and error if repair is attempted on the same CF as one which already have repair running.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (CASSANDRA-2834) Avoid repair getting started
twice at the same time for the same CF
Posted by "Jonathan Ellis (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/CASSANDRA-2834?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13277159#comment-13277159 ]
Jonathan Ellis commented on CASSANDRA-2834:
-------------------------------------------
Do we really need to coordinate cross-node? Or can we just say, "I'm already running a validation compaction for repair session X, so I'm going to skip session Y."
> Avoid repair getting started twice at the same time for the same CF
> -------------------------------------------------------------------
>
> Key: CASSANDRA-2834
> URL: https://issues.apache.org/jira/browse/CASSANDRA-2834
> Project: Cassandra
> Issue Type: Improvement
> Reporter: Terje Marthinussen
> Assignee: Sylvain Lebresne
> Priority: Minor
>
> It may seem like it is possible to start repair twice at the same time on the same CF?
> Not 100% verified, but if this is indeed the case, we may want to consider avoiding that including making nodetool repair abort and return and error if repair is attempted on the same CF as one which already have repair running.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (CASSANDRA-2834) Avoid repair getting started
twice at the same time for the same CF
Posted by "Sylvain Lebresne (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/CASSANDRA-2834?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13277694#comment-13277694 ]
Sylvain Lebresne commented on CASSANDRA-2834:
---------------------------------------------
You mean having a node skip a validation compaction if it is already running one for the same range? If we do that, what do we answer to the node that asked for a a merkle tree but we've decided to not do it? And even if we answer to the node a 'won't do it' (which probably already require a small protocol change), then what does that node do? Either it continue a repair that won't be complete, or he tries to abort it, but then we need a way to abort a repair. Besides, if 2 repair X and Y are started on 2 different nodes, there is no guarantee that one node won't skip a repair for session X, and another for session Y, so we're not even sure to make correct progress on any of those session.
> Avoid repair getting started twice at the same time for the same CF
> -------------------------------------------------------------------
>
> Key: CASSANDRA-2834
> URL: https://issues.apache.org/jira/browse/CASSANDRA-2834
> Project: Cassandra
> Issue Type: Improvement
> Reporter: Terje Marthinussen
> Assignee: Sylvain Lebresne
> Priority: Minor
>
> It may seem like it is possible to start repair twice at the same time on the same CF?
> Not 100% verified, but if this is indeed the case, we may want to consider avoiding that including making nodetool repair abort and return and error if repair is attempted on the same CF as one which already have repair running.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Updated] (CASSANDRA-2834) Avoid repair getting started
twice at the same time for the same CF
Posted by "Jonathan Ellis (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/CASSANDRA-2834?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Jonathan Ellis updated CASSANDRA-2834:
--------------------------------------
Priority: Minor (was: Major)
Assignee: Sylvain Lebresne
> Avoid repair getting started twice at the same time for the same CF
> -------------------------------------------------------------------
>
> Key: CASSANDRA-2834
> URL: https://issues.apache.org/jira/browse/CASSANDRA-2834
> Project: Cassandra
> Issue Type: Improvement
> Reporter: Terje Marthinussen
> Assignee: Sylvain Lebresne
> Priority: Minor
>
> It may seem like it is possible to start repair twice at the same time on the same CF?
> Not 100% verified, but if this is indeed the case, we may want to consider avoiding that including making nodetool repair abort and return and error if repair is attempted on the same CF as one which already have repair running.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira