| Author |
Message
|
| hkhan12 |
 Posted: Fri Dec 14, 2007 10:56 am Post subject: *Error cc=2 rc=2189 Cannot Open |
 |
|
Voyager
Joined: 08 Aug 2002
Posts: 98
|
Hi MQ Experts,
I have cluster environment. My application connecting to Solaris queue manager(partial repo) throwing rc=2189 when its trying to put a message on my another partial repo queue manager running on aix.
Initially, I have removed one of my windows queue manager from the same cluster. After a couple of attempts, I was able to suspend and removed this windows queue manager from cluster but my application was getting 2189. Hence, I have ran REFRESH CLUSTER(CLUSTER_NAME) REPOS(YES) at my partial repo queue managers. I came to know later that REFRESH CLUSTER is a risky one to perform.
Initially the problem was fixed but again its appearing. Please help me. It is very urgent.
Thanks a lot for all your help in advance. |
|
| Back to top |
|
 |
| KramJ |
| Posted: Fri Dec 14, 2007 11:17 am Post subject: |
|
|
Voyager
Joined: 09 Jan 2006
Posts: 80
Location: Atlanta
|
|
| Back to top |
|
|
| hkhan12 |
| Posted: Fri Dec 14, 2007 5:05 pm Post subject: |
|
|
Voyager
Joined: 08 Aug 2002
Posts: 98
|
Dear KramJ,
Thanks a lot for your response to my post. I have looked at the link you sent me and found that amqrrmfa process was not running on one of my partial repo queue manager.
I have stop and re-start my partial repo queue manager but does not help. Now I have found that any queue I create on receiving partial repo queue manager as a clustered, its not visible and I can not put messages in it from my one specific queeu manager.
Please help me here. Any body can contribute to this issue would be highly appreciated. I need this help badly.
My environment again as follows:
1. Two repos running on aix and solaris. (AIX with CSD13, Solaris with CSD 10 - both running WMQ5.3)
2. One sending partial repo running on solaris WMQ6 (plain mq6)
3. One receinign partial repo running on AIX with WMQ 5.3 CSD 13.
Thank you again for your help!!! |
|
| Back to top |
|
|
| fjb_saper |
| Posted: Fri Dec 14, 2007 6:55 pm Post subject: |
|
|
Grand High Poobah
Joined: 18 Nov 2003
Posts: 20697
Location: LI,NY
|
If you can't get amqrrmfa to run you will need to open a pmr with IBM.
Check for any FDC's and info in the error logs.
Make sure that all your cluster channels are in status running (check through channel status) if need be start them manually and then bounce the qmgr.
Check if you can see process amqrrmfa running.
Have fun.
_________________
MQ & Broker admin |
|
| Back to top |
|
|
| PeterPotkay |
| Posted: Fri Dec 14, 2007 6:56 pm Post subject: |
|
|
Poobah
Joined: 15 May 2001
Posts: 7716
|
This partial repository QM that's having a problem:
It has one and only 1 CLUSSNDR channel defined to 1 Full Repository? That channel is defined 100% correctly? (its clustered, its got the correct host name and port of the server hosting the FR?) Are you 100% sure?
It has one and only 1 CLUSRCVR channel defined? That channel is defined 100% correctly? (it has the correct host name and port # for the server this Partial QM is running on, its clustered, etc) Are you 100% sure?
Are any channels retrying?
Is the S.C.T.Q. empty?
Is the S.C.C.Q. empty? Does it have an IPROCS of one, which should be your Repository Manager.
Are there any messages in the MQ Error logs?
Are there any FDC files?
(this is all basic problem resolution for any cluster problem)
If you need help badly please open a PMR with IBM instead of relying on help that may or may not come from some stranger on the internet.
I notice you have your Full Repositories at MQ 5.3 and 1 or more partial repositories at MQ 6.0.
Its highly recommended that your Full Repositories are always at the latest version in a cluster and never behind any Partials. And MQ 5.3 is no longer supported.
You say this QM is at MQ 6.0.0.0? You really really should upgrade that to 6.0.2.2. I would rather be at MQ 5.3.0.13 than MQ 6.0.0.0, even though 5.3 is no longer supported. Its never a good idea to be at the initial release of a major version if you can help it.
_________________
Peter Potkay
Keep Calm and MQ On |
|
| Back to top |
|
|
| hkhan12 |
| Posted: Sat Dec 15, 2007 3:18 am Post subject: |
|
|
Voyager
Joined: 08 Aug 2002
Posts: 98
|
| PeterPotkay wrote: |
This partial repository QM that's having a problem:
It has one and only 1 CLUSSNDR channel defined to 1 Full Repository? That channel is defined 100% correctly? (its clustered, its got the correct host name and port of the server hosting the FR?) Are you 100% sure?
It has one and only 1 CLUSRCVR channel defined? That channel is defined 100% correctly? (it has the correct host name and port # for the server this Partial QM is running on, its clustered, etc) Are you 100% sure?
Are any channels retrying?
Is the S.C.T.Q. empty?
Is the S.C.C.Q. empty? Does it have an IPROCS of one, which should be your Repository Manager.
Are there any messages in the MQ Error logs?
Are there any FDC files?
(this is all basic problem resolution for any cluster problem)
If you need help badly please open a PMR with IBM instead of relying on help that may or may not come from some stranger on the internet.
I notice you have your Full Repositories at MQ 5.3 and 1 or more partial repositories at MQ 6.0.
Its highly recommended that your Full Repositories are always at the latest version in a cluster and never behind any Partials. And MQ 5.3 is no longer supported.
You say this QM is at MQ 6.0.0.0? You really really should upgrade that to 6.0.2.2. I would rather be at MQ 5.3.0.13 than MQ 6.0.0.0, even though 5.3 is no longer supported. Its never a good idea to be at the initial release of a major version if you can help it. |
Hi Peter,
yes, my S.C.C.Q is empty and it Does not have an IPROCS from my Repository Manager.
Anyway, I wanted to suspend this Solaris partial repository QM. When I suspend my PR using SUSPEND QMGR CLUSTER(CLUST_NAME) MODE(FORCE), my full repo is still showing SUSPEND(NO).
Idea was to replace one of our Windows PR QM with this Solaris PR QM. Now I tried it to suspend new PR on Solaris and bring back the Windows one.
Anyway, I brought back WIndows PR in cluster but even windows PR also throwing resolution error 2189. It was working fine earlier.
my QM is EESO081I, when I did "ps -ef | grep -i amq" it is showing this
0:04 /opt/mqm/bin/amqrrmfa -t2332800 -s2592000 -p2592000 -g5184000 -c3600 -m EESO081.
without last charater I. Is this correct or something wrong with it.
I'm really in big trouble now. I gues I will raise a PMR |
|
| Back to top |
|
|
| PeterPotkay |
| Posted: Sat Dec 15, 2007 7:17 am Post subject: |
|
|
Poobah
Joined: 15 May 2001
Posts: 7716
|
| hkhan12 wrote: |
yes, my S.C.C.Q is empty and it Does not have an IPROCS from my Repository Manager.
|
If the queue is empty but the Repository Manager is not running.......do you have something else consuming the cluster administration messages from the S.C.C.Q.?!
(you only answered one of my many questions)
_________________
Peter Potkay
Keep Calm and MQ On |
|
| Back to top |
|
|
| hkhan12 |
| Posted: Tue Dec 25, 2007 12:11 am Post subject: Partially Resolved |
|
|
Voyager
Joined: 08 Aug 2002
Posts: 98
|
Hi Peter,
2189 error has partially resolved when I broght down all the servers involved in the cluster along with queue managers.
One of the solution was to make one of the partial repository as full repository where I was getting 2189 and change it back to partial repository. I did it but I think the key thing is stop all manually defined cluster-sender & cluster-receiver channels before you bring down the servers.
Here is what I did:
1. Change PR to FR where I was getting 2189
2. Stop all Cluster-Sender, Cluster-Receiver channels (all PRs & 3 FRs)
3. Bring up all Servers
4. Change new FR to PR where I got 2189
5. Perform REFRESH CLUSTER(CLUSTER_NAME) command on new FR
where I got 2189
6. Start all Cluster-Sender & Cluster-Receiver channels (All PRs & 3 FRs)
I mentioned partially resolved because I have suspend/removed my new PR (Solaris) and added old Windows PR. I have performed above tasks on my old Windows PR.
Now, I have to remove old Windows PR and put new Solaris PR back into Cluster. I know before I put Solaris PR into cluster, I have to apply at least WMQv6.0.2.1 FixPack.
I'm not sure how reliable WMQ6.0.2.2 FixPack. Please share your experience with WMQ v6.0.2.2 any issues you run into. I remember, last time I had WMQ6.0.2.2 and I ran into some issues which I don't remember now.
Any way, Thanks a lot for all of you who helped me. I really appreciate it.
Regards, |
|
| Back to top |
|
|
| hkhan12 |
| Posted: Tue Dec 25, 2007 4:12 am Post subject: Partially Resolved |
|
|
Voyager
Joined: 08 Aug 2002
Posts: 98
|
Hi There,
I just came to know that my management wants to uninstall WMQv6 from one of my new PR on Solaris and install WMQ5.3 in order to make whole MQ Cluster consistent since all other PRs and two FRs are running WMQ5.3 CSD13.
I know WMQ5.3 is unsupported by IBM but they are not thinking about it.
I have already informed and asked them to move forward to WMQv6, lookslike its not gonna happen.
Also they want to make new MQ Server as FR whereas we already have 2 FRs. It means we gonna have 3 FRs in our cluster all running WMQ5.3 some with CSD 10 and some with CSD 13.
My question is will there be any issues having 3 FRs in a cluster of around 25 - 30 queue managers?
Thanks a lot for your input and feedback.
Regards, |
|
| Back to top |
|
|
| fjb_saper |
| Posted: Tue Dec 25, 2007 4:56 pm Post subject: |
|
|
Grand High Poobah
Joined: 18 Nov 2003
Posts: 20697
Location: LI,NY
|
The first thing you upgrade in a cluster should be the FRs.
So what were you doing running a PR @ 6.0 against a 5.3 FR?? 
_________________
MQ & Broker admin |
|
| Back to top |
|
|
| PeterPotkay |
| Posted: Wed Dec 26, 2007 1:19 pm Post subject: Re: Partially Resolved |
|
|
Poobah
Joined: 15 May 2001
Posts: 7716
|
| hkhan12 wrote: |
| Also they want to make new MQ Server as FR whereas we already have 2 FRs. It means we gonna have 3 FRs in our cluster... |
Why?
| hkhan12 wrote: |
My question is will there be any issues having 3 FRs in a cluster of around 25 - 30 queue managers?
|
If done properly, no problem. But usually not necessary. It goes against best practices for MQ Clustering......and going against best practices usually leads to problems 
_________________
Peter Potkay
Keep Calm and MQ On |
|
| Back to top |
|
|
| Nigelg |
| Posted: Thu Dec 27, 2007 12:04 am Post subject: |
|
|
Grand Master
Joined: 02 Aug 2004
Posts: 1046
|
| Quote: |
My question is will there be any issues having 3 FRs in a cluster of around 25 - 30 queue managers?
|
You do not get any extra if you add more FRs to a cluster. Each PR will send details of its objects to only 2 FRs - the one it is 'connected' to, i.e. the one it has a manual CLUSSDR defined to, and one other. The result of having more than 2 FRs will be that each will have only a partial view of the cluster objects derived from the PRs, and what is worst, a randomly distributed partial view. Of course, each FR will also receive updates about all cluster objects from the other FRs in the cluster if you have also defined manual CLUSSDRs from each FR to each other FR.
However, if you have not done that or if the interconnecting channels go down for any reason, the FRs will be left wih a partial view of the cluster as objects are added & amended, since the FRs are relying on updates from each other and not from the PRs.
Use 2 and only 2 FRs in a cluster, and you eliminate these potentilal problems, e.g. causing 2085 & 2189.
_________________
MQSeries.net helps those who help themselves.. |
|
| Back to top |
|
|
|
|
