ASG
IBM
Zystems
Cressida
Icon
Netflexity
 
  MQSeries.net
Search  Search       Tech Exchange      Education      Certifications      Library      Info Center      SupportPacs      LinkedIn  Search  Search                                                                   FAQ  FAQ   Usergroups  Usergroups
 
Register  ::  Log in Log in to check your private messages
 
RSS Feed - WebSphere MQ Support RSS Feed - Message Broker Support

MQSeries.net Forum Index » WebSphere Message Broker (ACE) Support » regarding workload balancing

Post new topic  Reply to topic
 regarding workload balancing « View previous topic :: View next topic » 
Author Message
mq_pm
PostPosted: Mon Oct 18, 2004 10:06 am    Post subject: regarding workload balancing Reply with quote

Centurion

Joined: 27 Aug 2003
Posts: 132

Hi,

I have a scenario like this:

2 QMs, QM1 and QM2 with full repositories on two servers. Each server has broker installed with same configuration and same message flow and set. These are in cluster. There is a one more QM3 which is a part of this cluster.
Now the situation is when QM3 puts a message to a cluster queue X which is defined on both server QMs from which the broker message flow picks it up and process it. But when one broker is down say server1 broker, when QM3 puts a message one after the other, to X queue, then it distributes the message to both queues X defined on servers doing work load balancing. But since one broker is down, the message stays on that queue X of QM1 and when message arrives to queue of QM2, it processes it since it is up. So how can i make the all the messages to be processed even though one broker is down. Otherwise having 2 brokers for failover or workload balancing is of no use....Any suggestions please.

Thanks.
Back to top
View user's profile Send private message Send e-mail
vennela
PostPosted: Mon Oct 18, 2004 10:56 am    Post subject: Reply with quote

Jedi Knight

Joined: 11 Aug 2002
Posts: 4055
Location: Hyderabad, India

MQ clustering does not help failover scenarios. It is used for work load balancing. You can achieve this using a hardware failover like HACMP Veritas or Service Gaurd etc.
Back to top
View user's profile Send private message Send e-mail Visit poster's website
kirani
PostPosted: Mon Oct 18, 2004 1:41 pm    Post subject: Reply with quote

Jedi Knight

Joined: 05 Sep 2001
Posts: 3779
Location: Torrance, CA, USA

Or write a Cluster workload exit to route messages to Active Broker's queue manager.
_________________
Kiran


IBM Cert. Solution Designer & System Administrator - WBIMB V5
IBM Cert. Solutions Expert - WMQI
IBM Cert. Specialist - WMQI, MQSeries
IBM Cert. Developer - MQSeries

Back to top
View user's profile Send private message Visit poster's website
jefflowrey
PostPosted: Mon Oct 18, 2004 2:21 pm    Post subject: Reply with quote

Grand Poobah

Joined: 16 Oct 2002
Posts: 19981

Or write a script that gets fired when the broker shuts down, that also shuts down the queue manager - or otherwise disables it from the cluster.
_________________
I am *not* the model of the modern major general.
Back to top
View user's profile Send private message
Nigelg
PostPosted: Thu Oct 21, 2004 12:34 am    Post subject: Reply with quote

Grand Master

Joined: 02 Aug 2004
Posts: 1046

When a CLUSSDR channel stops, if there are any msgs remaining on the cluster xmitq for the channel the msgs are all read and put through the workload balancing algorithm. However, all other things being equal, a channel status of RUNNING is the same, as far as workload balancing is concerned, as INACTIVE, and so the msgs will still be distributed to the inactive qmgr.
You can change this by:
Altering the channel status to STOPPED; this has a lower priority than RUNNING and so the msgs will be rerouted to the RUNNING qmgr
Suspending the inactive qmgr, so that the msgs will be routed to the qmgr joined to the cluster
You can do this after the qmgr has stopped; take whichever action you choose to lower the priority of the inactive qmgr, and then start the CLUSSDR channel. The channel will not start, but in ending and going to RETRYING it will reroute the msgs on the cluster xmitq. RERYING also has a lower priority than RUNNING. After the msgs have been rerouted you can stop the channel again.
Back to top
View user's profile Send private message
mverh
PostPosted: Thu Oct 21, 2004 7:00 am    Post subject: Reply with quote

Voyager

Joined: 06 Mar 2002
Posts: 97

Quote:
You can change this by:
Altering the channel status to STOPPED; this has a lower priority than RUNNING and so the msgs will be rerouted to the RUNNING qmgr
Suspending the inactive qmgr, so that the msgs will be routed to the qmgr joined to the cluster


Suspending the inactive qmgr will not necessarily result in messages no longer being routed to it. I too thought this would work but my testing proved otherwise. Never really figured out why you would want to issue a suspend, if not this exact reason. In any case stopping the clussdr will work.
Back to top
View user's profile Send private message
Nigelg
PostPosted: Thu Oct 21, 2004 7:51 am    Post subject: Reply with quote

Grand Master

Joined: 02 Aug 2004
Posts: 1046

There are several posts on this forum, including one by me, which gives the exact critieria for which qmgr is selected by the workload balancing algorithm.
Suspending a qmgr will not stop a msg going to it if no other destination with a higher priority is available.
Back to top
View user's profile Send private message
mq_pm
PostPosted: Mon Oct 25, 2004 6:29 am    Post subject: Reply with quote

Centurion

Joined: 27 Aug 2003
Posts: 132

Hi,

As I mentioned the scenario of the environment above, I do have a doubt. Since the queue is a cluster defined on both cluster qmgrs, if in the flow, in the output node if I mention only the queue and no Qmgr name, then the message is processed by the active broker. Is there any other way I can do this because we need to mention the Qmgr name and queue name in the flow for the message to go.
I hope my question is clear.

Thanks.
Back to top
View user's profile Send private message Send e-mail
vennela
PostPosted: Mon Oct 25, 2004 7:19 am    Post subject: Reply with quote

Jedi Knight

Joined: 11 Aug 2002
Posts: 4055
Location: Hyderabad, India

There is no way that MQ knows whether the broker is running or not.
Your question is not clear. Because you say:
Quote:
Since the queue is a cluster defined on both cluster qmgrs, if in the flow, in the output node if I mention only the queue and no Qmgr name, then the message is processed by the active broker.


A message that is put in the queue defined on the output node is not usually processed by broker. That's the node(queue) where the broker puts the message.
Back to top
View user's profile Send private message Send e-mail Visit poster's website
Display posts from previous:   
Post new topic  Reply to topic Page 1 of 1

MQSeries.net Forum Index » WebSphere Message Broker (ACE) Support » regarding workload balancing
Jump to:  



You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
Protected by Anti-Spam ACP
 
 


Theme by Dustin Baccetti
Powered by phpBB © 2001, 2002 phpBB Group

Copyright © MQSeries.net. All rights reserved.