[jboss-jira] [JBoss JIRA] Commented: (JBMESSAGING-1230) Client failover support if one node of a cluster is cleanly shut down.

[Phillip Thurmond (JIRA)]

    [ http://jira.jboss.com/jira/browse/JBMESSAGING-1230?page=comments#action_12400607 ] 
            
Phillip Thurmond commented on JBMESSAGING-1230:
-----------------------------------------------

Just to address a couple of points from the forum posts at http://www.jboss.com/index.html?module=bb&op=viewtopic&t=110632

----snip
"Shutting down a node in the cluster is quite different from failing over. With a shutdown, it is likely the sys admin is just taking the node down temporarily and will restart it again shortly. In this case we *do not* want another node to take over and merge the queues of the node that is being taken down."

"Imagine the pathological case where the sysadmin takes every node in the cluster down one by one. By the time the sysadmin takes down the last node in the cluster, that node will have merged in *all* the queues from the other nodes! Not only will this be slow, but it means that when the cluster is brought back up, only one of the nodes (the last one to be taken down) will have all the messages."
----snip

I would argue that this behavior would be desirable in some common situations.  Specifically, if the sysadmin wants to do maintenance on several nodes without interrupting service to any of the clients.  Merging all the queues to one server might be desirable here since this type of maintenance is usually done during off-peak hours.

Also, without being too contrived, I think there are failure scenarios (ie, all nodes losing network connectivity except one) which could result in the above situation occurring even with the current failover implementation.

I can't speak for what's happening in the implementation when these different scenarios occur, but from a sysadmin's perspective they seem like very similar events which should be handled consistently.

Maybe it would be possible to add a flag which specified that a shutdown should be treated as a failover event?  That way it would be an option for those who wanted this functionality.

> Client failover support if one node of a cluster is cleanly shut down.
> ----------------------------------------------------------------------
>
>                 Key: JBMESSAGING-1230
>                 URL: http://jira.jboss.com/jira/browse/JBMESSAGING-1230
>             Project: JBoss Messaging
>          Issue Type: Feature Request
>            Reporter: Phillip Thurmond
>         Assigned To: Tim Fox
>             Fix For: Stable branch
>
>
> If one node of a cluster fails in someway, the client will attempt to failover transparently.  However, if one node is shutdown cleanly, the client will not automatically failover to another node in the cluster.  This functionality would be very useful for "rolling upgrades" where a portion of the cluster is left up while the other nodes are being updated.

-- 
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira