hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Elliott Clark (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-9312) Lower StochasticLoadBalancer's default max run time
Date Fri, 23 Aug 2013 20:09:51 GMT

    [ https://issues.apache.org/jira/browse/HBASE-9312?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13748948#comment-13748948

Elliott Clark commented on HBASE-9312:

Since it's runtime based I'll wait for hadoop qa run to pass.  If it does I'll commit to trunk
and 95.
> Lower StochasticLoadBalancer's default max run time 
> ----------------------------------------------------
>                 Key: HBASE-9312
>                 URL: https://issues.apache.org/jira/browse/HBASE-9312
>             Project: HBase
>          Issue Type: Bug
>    Affects Versions: 0.95.2
>            Reporter: Jean-Daniel Cryans
>            Assignee: Elliott Clark
>             Fix For: 0.98.0, 0.96.0
>         Attachments: HBASE-9312-0.patch
> Three things about the 1 minute it takes to decide if we want to balance or not:
>  - 1 minute is also the default socket timeout, so if you call "balancer" in the shell
you often get a SocketTimeoutException back. Bad user experience.
>  - Elliott was telling me that we might not even need 1 minute to compute a good balance
anyways because of other improvements. I tested 30 seconds on a badly balanced cluster and
it worked well.
>  - I personally think that 1 minute is too much time (Elliott disagrees).

This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

View raw message