| Author |
Message
|
| J.D |
 Posted: Mon Sep 13, 2010 4:09 pm Post subject: Load Balancing |
 |
|
Voyager
Joined: 18 Dec 2009
Posts: 92
Location: United States
|
I have added new Qmgr to existing cluster as PR and created a local queue (TEST) with DEFBIND set to NOTFIXED. Already there was a same instance of Queue on different PR qmgr in same cluster with same properties. When application tried to send messages, all of them are coming to new qmgr and 0 msgs are going to old qmgr which has same instance. Cluster channels are working fine on FR and PR Qmgrs.
Env:
QMGR A (FR) (Front-End)
QMGR B (FR) (Front-End)
QMGR C (PR) (Back-End)
QMGR D (PR) (Back-End)
QMGR C was added to cluster before by creating CLUSSDR chl to QMGR A. But the newly created QMGR D is added to cluster by creating CLUSSDR chl to QMGR B which is also a full repository. And the other differences between QMGR C & D are primary log files and secondary log files are different.
_________________
IBM WebSphere MQ & WAS Administrator |
|
| Back to top |
|
 |
| Vitor |
| Posted: Mon Sep 13, 2010 4:17 pm Post subject: Re: Load Balancing |
|
|
Grand High Poobah
Joined: 11 Nov 2005
Posts: 26093
Location: Texas, USA
|
| J.D wrote: |
| I have added new Qmgr to existing cluster as PR and created a local queue (TEST) with DEFBIND set to NOTFIXED. |
It's been said before and will be said again - that's only a default setting and there's no reason the application will honour it. Check that the application is either using that setting, or using NOTFIXED for it's own reasons.
After that, do all the usual cluster things. Check both FRs can see both instances of this TEST queue. Check that all the cluster channels are running (I know you said they're fine, check again that they are running). Check all the channels & queues are of equal priority & weight. Etc. Etc.
You fix cluster problems by grinding through the configuration, dotting i's and crossing t's.
I don't see log sizes being a factor personally.
_________________
Honesty is the best policy.
Insanity is the best defence. |
|
| Back to top |
|
|
| PeterPotkay |
| Posted: Mon Sep 13, 2010 5:32 pm Post subject: |
|
|
Poobah
Joined: 15 May 2001
Posts: 7722
|
|
| Back to top |
|
|
| J.D |
| Posted: Mon Sep 13, 2010 11:20 pm Post subject: |
|
|
Voyager
Joined: 18 Dec 2009
Posts: 92
Location: United States
|
I did "DIS CLUSQMGR(*)" and "DIS QCLUSTER(*)" on both the front end Queue managers A & B and evrything looks fine. Our applications connect to front end qmgrs A & B to publish messages to Cluster Queue which is defined on back end Qmgrs C & D.
I even tried sending msg using amqsput from two front end which are FRs but msgs are going to newly created Cluster member D only.
_________________
IBM WebSphere MQ & WAS Administrator |
|
| Back to top |
|
|
| Vitor |
| Posted: Tue Sep 14, 2010 4:04 am Post subject: |
|
|
Grand High Poobah
Joined: 11 Nov 2005
Posts: 26093
Location: Texas, USA
|
| J.D wrote: |
| I did "DIS CLUSQMGR(*)" and "DIS QCLUSTER(*)" on both the front end Queue managers A & B and evrything looks fine. |
So both A & B display C & D as members of the cluster? Both A & B show 2 instances of the queue TEST, one on each of C & D?
Both instances of TEST are put enabled? All messages put are going to the one instance of TEST, none are sitting on a SCTQ?
And all the channels are running?
_________________
Honesty is the best policy.
Insanity is the best defence. |
|
| Back to top |
|
|
| J.D |
| Posted: Tue Sep 14, 2010 4:17 am Post subject: |
|
|
Voyager
Joined: 18 Dec 2009
Posts: 92
Location: United States
|
| Vitor wrote: |
| So both A & B display C & D as members of the cluster? Both A & B show 2 instances of the queue TEST, one on each of C & D? |
Yes
| Vitor wrote: |
| Both instances of TEST are put enabled? All messages put are going to the one instance of TEST, none are sitting on a SCTQ? |
Yes. I am seeing all msgs published in only one instance. None on SCTQ.
| Vitor wrote: |
| And all the channels are running? |
Clussdr from FRs to one of the PR was in inactive state. I manually started from both the FRs before publishing msgs.
_________________
IBM WebSphere MQ & WAS Administrator |
|
| Back to top |
|
|
| exerk |
| Posted: Tue Sep 14, 2010 4:27 am Post subject: |
|
|
Jedi Council
Joined: 02 Nov 2006
Posts: 6339
|
| J.D wrote: |
| Clussdr from FRs to one of the PR was in inactive state. I manually started from both the FRs before publishing msgs. |
I do hope you mean these were dynamically-defined CLUSSDR's, not explicit ones 
_________________
It's puzzling, I don't think I've ever seen anything quite like this before...and it's hard to soar like an eagle when you're surrounded by turkeys. |
|
| Back to top |
|
|
| J.D |
| Posted: Tue Sep 14, 2010 4:47 am Post subject: |
|
|
Voyager
Joined: 18 Dec 2009
Posts: 92
Location: United States
|
| exerk wrote: |
Clussdr from FRs to one of the PR was in inactive state. I manually started from both the FRs before publishing msgs.
I do hope you mean these were dynamically-defined CLUSSDR's, not explicit ones |
Yes they are dynamically defined channels. I manually started to make sure clussdr channel to both PRs are working.
_________________
IBM WebSphere MQ & WAS Administrator |
|
| Back to top |
|
|
| PeterPotkay |
| Posted: Tue Sep 14, 2010 2:50 pm Post subject: |
|
|
Poobah
Joined: 15 May 2001
Posts: 7722
|
MQ doesn't load balance at the queue level. Refer to the link I posted above for an explanation on why you might be seeing the behavior you are after adding a new QM to the cluster.
_________________
Peter Potkay
Keep Calm and MQ On |
|
| Back to top |
|
|
| mvic |
| Posted: Tue Sep 14, 2010 3:14 pm Post subject: |
|
|
Jedi
Joined: 09 Mar 2004
Posts: 2080
|
|
| Back to top |
|
|
| J.D |
| Posted: Tue Sep 14, 2010 5:48 pm Post subject: |
|
|
Voyager
Joined: 18 Dec 2009
Posts: 92
Location: United States
|
Thank You. Looks like 12th point also resolves my problem. But i followed a different method provided by IBM in the below link.
| PeterPotkay wrote: |
| http://www-01.ibm.com/support/docview.wss?rs=171&uid=swg1IZ51783 |
Thanks Peter. Load balancing started working after altering one of the cluster workload attributes of Queue instance in both old and new qmgr as mentioned in the above link.
_________________
IBM WebSphere MQ & WAS Administrator |
|
| Back to top |
|
|
|
|
