cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Marcus Eriksson (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (CASSANDRA-11451) Don't mark sstables as repairing when doing sub range repair
Date Tue, 29 Mar 2016 12:37:25 GMT

     [ https://issues.apache.org/jira/browse/CASSANDRA-11451?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Marcus Eriksson updated CASSANDRA-11451:
----------------------------------------
    Description: 
Since CASSANDRA-10422 we don't do anticompaction when a user issues a sub range repair (-st
X -et Y), but we still mark sstables as repairing.

We should avoid marking them as users might want to run many sub range repair sessions in
parallel. 

The reason we mark sstables is that we don't want another repair session to steal the sstables
before we do anticompaction, and since we do no anticompaction with sub range repair we have
no benefit from the marking.

  was:
Since CASSANDRA-10422 we don't do anticompaction when a user issues a sub range repair (-st
X -et Y), but we still mark sstables as repairing.

We should avoid marking them as users might want to run many sub range repair sessions in
parallel. 

The reason we mark sstables is that we need don't want another repair session to steal the
sstables before we do anticompaction, and since we do no anticompaction with sub range repair
we have no benefit from the marking.


> Don't mark sstables as repairing when doing sub range repair
> ------------------------------------------------------------
>
>                 Key: CASSANDRA-11451
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-11451
>             Project: Cassandra
>          Issue Type: Bug
>            Reporter: Marcus Eriksson
>             Fix For: 2.2.x, 3.0.x, 3.x
>
>
> Since CASSANDRA-10422 we don't do anticompaction when a user issues a sub range repair
(-st X -et Y), but we still mark sstables as repairing.
> We should avoid marking them as users might want to run many sub range repair sessions
in parallel. 
> The reason we mark sstables is that we don't want another repair session to steal the
sstables before we do anticompaction, and since we do no anticompaction with sub range repair
we have no benefit from the marking.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message