hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "nkeywal (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-6960) Support fast and graceful master failover
Date Tue, 09 Oct 2012 05:48:03 GMT

    [ https://issues.apache.org/jira/browse/HBASE-6960?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13472154#comment-13472154
] 

nkeywal commented on HBASE-6960:
--------------------------------

fwiw, it's partially done in HBASE-5926. Partially because HBASE-5926 would not work with
private launch scripts.
                
> Support fast and graceful master failover
> -----------------------------------------
>
>                 Key: HBASE-6960
>                 URL: https://issues.apache.org/jira/browse/HBASE-6960
>             Project: HBase
>          Issue Type: Bug
>          Components: master
>    Affects Versions: 0.94.2, 0.96.0
>            Reporter: Jonathan Hsieh
>
> If you have multiple masters, you should be able to kill the primary and have the backups
assume control within seconds instead of having to wait the default 3mins before an ephemeral
znode expires.  
> The expected behavior is similar to how region servers when killed (not -9) trigger the
servers shutdown handler and quickly pass of its regions for reassignment.  In this case,
a master that is killed is removes the znode and allow another to become the primary.  A related
follow on feature could be the ability to failover to another primary without taking the orignal
primary down.

--
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

Mime
View raw message