You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@activemq.apache.org by "Dejan Bosanac (Resolved) (JIRA)" <ji...@apache.org> on 2012/02/10 12:57:59 UTC

[jira] [Resolved] (AMQ-3706) Dynamic Failover Randomize is Not Evenly Balanced

     [ https://issues.apache.org/jira/browse/AMQ-3706?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Dejan Bosanac resolved AMQ-3706.
--------------------------------

    Resolution: Fixed
      Assignee: Dejan Bosanac

Fixed with svn revision 1242748

The problem was that every client was responsible for randomizing his list of urls, and as every client uses it's own rnd generator we haven't had a uniform distribution.
if the cluster rebalance is used, now we're doing a randomization on the broker side (always using same rnd generator) and clients now respect the url order broker supplied. It give much better distribution in this case.
                
> Dynamic Failover Randomize is Not Evenly Balanced
> -------------------------------------------------
>
>                 Key: AMQ-3706
>                 URL: https://issues.apache.org/jira/browse/AMQ-3706
>             Project: ActiveMQ
>          Issue Type: Sub-task
>          Components: Broker
>    Affects Versions: 5.5.0
>            Reporter: Scott England-Sullivan
>            Assignee: Dejan Bosanac
>             Fix For: 5.6.0
>
>         Attachments: AMQ-3706.patch
>
>
> When using dynamic failover with randomize enabled, clients are not consistently rebalanced in an even manner.  A test case is added that allows a minimum of 25% per broker on a three broker test and 40% distribution on a 2 broker test.

--
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