| Author |
Message
|
| cc |
 Posted: Mon Jul 18, 2011 8:43 am Post subject: CLUSQMGR not found |
 |
|
Novice
Joined: 28 May 2008
Posts: 19
|
I have a very simple configuration with a single machine sending messages to two backend machines. All 3 machines are in a cluster with 2 FRs. If I create a cluster queue on the each of the two backend machines, the FRs see the queues and the front-end sending machine sees the queues. So when I do a dis qc(QUEUE_NAME) on the front-end machine, both instances are displayed. However, if I do a dis clusqmgr(*) on the front-end machine only one of the backend queue managers is displayed. Therefore, when I send messages from the front-end to the backend queue all messages are routed to one of the instances.
How do I get the front-end machine to have the CLUSQMGR definition in its partial repository so it does load balancing?
I tried a REFRESH CLUSTER with both REPOS YES and NO on the front-end machine.
If I do a dis clusqmgr(*) on the backend servers, only the one I can send too, knows about the front-end machine. However, if I do a dis clusqmgr(*) on either of the FRs they know about the front-end queue manager.
Thanks in advance, |
|
| Back to top |
|
 |
| exerk |
| Posted: Mon Jul 18, 2011 8:46 am Post subject: |
|
|
Jedi Council
Joined: 02 Nov 2006
Posts: 6339
|
Establish that your FRs can 'see' each other, i.e. that dis clusqmgr(*) on either FR shows the other FR. If not, fault-diagnose the set-up between the FRs and correct.
_________________
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 |
|
|
| Vitor |
| Posted: Mon Jul 18, 2011 8:52 am Post subject: Re: CLUSQMGR not found |
|
|
Grand High Poobah
Joined: 11 Nov 2005
Posts: 26093
Location: Texas, USA
|
| cc wrote: |
| I tried a REFRESH CLUSTER with both REPOS YES and NO on the front-end machine. |
Don't do that.
| cc wrote: |
| If I do a dis clusqmgr(*) on the backend servers, only the one I can send too, knows about the front-end machine. However, if I do a dis clusqmgr(*) on either of the FRs they know about the front-end queue manager. |
It's right that only an FR knows about a queue until it's been used (and I'm a bit confused about which machines in your set up are the FRs).
Standard advice for clusters:
- make sure all the channels are running
- make sure there are no cluster queue managers showing on any machine as TEMP.something
Resolve & fix those and if the problem persists when all the queues are properly visible remember that the application controls if load balancing is used & might be preventing it (so there's nothing actually wrong with the cluster).
_________________
Honesty is the best policy.
Insanity is the best defence. |
|
| Back to top |
|
|
| cc |
| Posted: Mon Jul 18, 2011 9:25 am Post subject: |
|
|
Novice
Joined: 28 May 2008
Posts: 19
|
The Cluster setup on the front-end looks ok. If I do an amqsput of a message destined for the cluster queue on the backend servers. I see the channels for both FRs are running and the CLUSSDRA to the one backend server that the front-end knows about. The defbind on the cluster queue is set to fixed not on open. However all of the messages go to the single backend server.
If I create a local clustered queue on the front-end, I see the CLUSSDR start up to both FRs and the CLUSRCVR has connection back from the FRs. If I go and look at both of the FRs, they know about the new local cluster queue on the front end. Given that I believe my configuration to the FRs from the Front-end is correct. |
|
| Back to top |
|
|
| exerk |
| Posted: Mon Jul 18, 2011 9:29 am Post subject: |
|
|
Jedi Council
Joined: 02 Nov 2006
Posts: 6339
|
| cc wrote: |
| ...Given that I believe my configuration to the FRs from the Front-end is correct. |
So you are absolutely sure that you see running CLUSSDR/CLUSRCVR channel instances from each FR to each FR? Do all the messages go to the same 'back-end' server if you use the amqsput sample? Can you confirm you are NOT specifying a destination queue manager during the PUT?
_________________
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 |
|
|
| cc |
| Posted: Mon Jul 18, 2011 9:39 am Post subject: |
|
|
Novice
Joined: 28 May 2008
Posts: 19
|
Yes I am sure there are channels in both directions from the two FRs.
On each FR I do a dis chs(TO.CLUSRCVR.CHANNEL) CURRENT and I see connection from the other FR.
I do an amqsput <QUEUE NAME> <LOCAL QMG> from the front end. |
|
| Back to top |
|
|
| fjb_saper |
| Posted: Mon Jul 18, 2011 12:16 pm Post subject: |
|
|
Grand High Poobah
Joined: 18 Nov 2003
Posts: 20756
Location: LI,NY
|
| cc wrote: |
The Cluster setup on the front-end looks ok. If I do an amqsput of a message destined for the cluster queue on the backend servers. I see the channels for both FRs are running and the CLUSSDRA to the one backend server that the front-end knows about. The defbind on the cluster queue is set to fixed not on open. However all of the messages go to the single backend server.
If I create a local clustered queue on the front-end, I see the CLUSSDR start up to both FRs and the CLUSRCVR has connection back from the FRs. If I go and look at both of the FRs, they know about the new local cluster queue on the front end. Given that I believe my configuration to the FRs from the Front-end is correct. |
Be more specific. You have 2 possible values here
So which one did you choose? 
_________________
MQ & Broker admin |
|
| Back to top |
|
|
| cc |
| Posted: Mon Jul 18, 2011 12:19 pm Post subject: |
|
|
Novice
Joined: 28 May 2008
Posts: 19
|
|
| Back to top |
|
|
| fjb_saper |
| Posted: Mon Jul 18, 2011 12:22 pm Post subject: |
|
|
Grand High Poobah
Joined: 18 Nov 2003
Posts: 20756
Location: LI,NY
|
| cc wrote: |
| DEFBIND(NOTFIXED) |
You might want to check your cluster channels manually from the front end...
_________________
MQ & Broker admin |
|
| Back to top |
|
|
| Vitor |
| Posted: Mon Jul 18, 2011 12:32 pm Post subject: |
|
|
Grand High Poobah
Joined: 11 Nov 2005
Posts: 26093
Location: Texas, USA
|
| cc wrote: |
| DEFBIND(NOTFIXED) |
And for the umpeenth time that doesn't guarantee that BIND_ON_OPEN won't be used.
But in the specific case of amqsput (unless it's been modified) we can have some confidence it's honouring the queue setting.
_________________
Honesty is the best policy.
Insanity is the best defence. |
|
| Back to top |
|
|
| dutchman |
| Posted: Tue Jul 19, 2011 12:03 am Post subject: |
|
|
Acolyte
Joined: 15 May 2001
Posts: 71
Location: Netherlands
|
Might be worth displaying some of the cluster related attributes. I suggest you do the following on the 'sending' qmgr:
1. dis clusqmgr(*) all
2. dis qcluster(*) all
Regards.. Ruud |
|
| Back to top |
|
|
|
|
