| Author |
Message
|
| yonny |
 Posted: Mon Oct 14, 2002 7:29 am Post subject: QMGR cannot join a cluster |
 |
|
Apprentice
Joined: 08 Jul 2001
Posts: 49
Location: Santo Domingo
|
Hello, I have a cluster where the repositories queue managers are MQSeries 5.1 for AIX with the last CSD. When tried to join a QM (MQSeries for Win2K 5.2.1 with last CSD) I found that the repositories queue managers appear as SYSTEM.TEMPQMGR[connaname]. I am unable to put messages on remote cluster queues from the new queue manager, mq returns 2085, I also got this reason code when trying to put messages from the Repo qmgrs to cluster queues on the new queue manager.
The cluster is working fine with other qmgrs but don't accept new ones.
Refresh cluster did not work, Reset the new queue manager from the repositories and trying to reconnect it again did not work either.
The following are error messages found on the new qmgr log. Thanks a lot for your help.
Yonny R. Serrano
10/13/2002 20:54:47
AMQ9419: No cluster-receiver channels for cluster 'ENG1'
EXPLANATION:
The repository manager has received information about a cluster for which no
cluster-receiver channels are known.
ACTION:
Define cluster-receiver channels for the cluster on the local queue manager.
-------------------------------------------------------------------------------
10/13/2002 20:54:47
AMQ9419: No cluster-receiver channels for cluster 'ENG1'
EXPLANATION:
The repository manager has received information about a cluster for which no
cluster-receiver channels are known.
ACTION:
Define cluster-receiver channels for the cluster on the local queue manager.
-------------------------------------------------------------------------------
10/13/2002 20:54:47
AMQ9442: Refresh Cluster command processed.
EXPLANATION:
The Refresh Cluster(ENG1) command caused 11 objects to be refreshed and 2
objects to be republished.
ACTION:
None.
.....
-------------------------------------------------------------------------------
10/13/2002 21:40:56
AMQ9519: Channel 'TO.QMPROENG1' not found.
EXPLANATION:
The requested operation failed because the program could not find a definition
of channel 'TO.QMPROENG1'.
ACTION:
Check that the name is specified correctly and the channel definition is
available.
-------------------------------------------------------------------------------
10/13/2002 21:40:56
AMQ9999: Channel program ended abnormally.
EXPLANATION:
Channel program 'TO.QMPROENG1' ended abnormally.
ACTION:
Look at previous error messages for channel program 'TO.QMPROENG1' in the error
files to determine the cause of the failure.
-------------------------------------------------------------------------------
10/13/2002 21:41:47
AMQ9420: No repositories for cluster ENG1.
EXPLANATION:
The queue manager has received information about a cluster for which no
repositories are known.
ACTION:
Alter the REPOS or REPOSNL attribute of the queue manager, that is to have a
full repository for the cluster, to specify the cluster name.
-------------------------------------------------------------------------------
 |
|
| Back to top |
|
 |
| bduncan |
| Posted: Mon Oct 14, 2002 10:01 pm Post subject: |
|
|
Padawan
Joined: 11 Apr 2001
Posts: 1554
Location: Silicon Valley
|
Can you by any chance post the exact commands you put into runmqsc to define the cluster sender and cluster receiver channel on the queue manager you were attempting to add to the cluster? You obviously know how to add queue managers to a cluster, since you already have a cluster existing (otherwise I'd point you to the MQSeries Using Clusters manual which describes how to add queue managers to a cluster; as you've probably noticed, the order in which you do things is essential).
I've seen this problem (and caused it too!) many times before and it invariably turned out to be a misconfiguration. You pasted the logs from (I'm presuming) the repository. Are there any log entries on the queue manager you are trying to add to the cluster? Also, have you tried using the PING channel command on the cluster sender channel on the new queue manager? Apparently that is working since the repository is learning about the new queue manager, but is unable to send information back to it. In this case I would look to see if the repository has generated a cluster sender channel pointing back to the new queue manager, and then check its status as well as try the ping channel command.
_________________
Brandon Duncan
IBM Certified MQSeries Specialist
MQSeries.net forum moderator |
|
| Back to top |
|
|
| yonny |
| Posted: Tue Oct 15, 2002 8:41 am Post subject: |
|
|
Apprentice
Joined: 08 Jul 2001
Posts: 49
Location: Santo Domingo
|
Brandon,
The logs that I posted are from the qmgr thar Im trying to add. I'm able to start the sender channel from the new qmgr to the repo qmgr, however, the repo qmgr do not generate a sender channel pointing back to the new qmgr (I have to do it manually).
So, I defined manually the cluster reciever channel at the new qmgr and both cluster sender channels, then I start the sender channels and they become running.
Another clue: I cannot join any other qmgr to the cluster, even with a differente name over a different machine. I see always the repository qmgrs as SYSTEM.TEMPQMGR[conname].
Thank you
Yonny Serrano |
|
| Back to top |
|
|
| 2189 |
| Posted: Tue Oct 22, 2002 2:58 am Post subject: |
|
|
Apprentice
Joined: 22 Oct 2002
Posts: 31
|
I've had this problem before ages ago. Although we fixed it I think we were getting the same problem every time we added a new queue manager so I suspect it a probelm with the repositories. You could risk setting up another qmgr and seeing if you get the same results, this wud identify if the problem is with the repos or the new qmgr.
I reported my fault to IBM and they provided this response:
The SYSTEM.TEMPQMGR entries are created until the QMgr
has contacted the remote qmgr. When it contacts it the
attributes of the remote QMgrs CLUSRCVR are used to "fill in" the
details of the local QMgrs temp. Therefore the
temp.QMgr entries imply that the QMgr has had difficulty
contacting the destination QMgr e.g QMPSB2TZ cannot
contact QMPSB2T0
There are, of course, many reasons why this may be the case. Here is a
brief list of some of the most common ones:
1) If there is no CLUSRCVR defined for that channel at the remote QMgr
you will see this problem.
2) If there are any misspellings in a channel definition this will cause
similar problems.
3) Networking problems which prevent the QMgr contacting the remote
machine
4) No QMgr or listener running on the remote machine at the time of the
CLUSSDR channel definition.
The majority of these are configuration issues, which Brandon mentioned, but I assume you've already checked all that stuff out. To confirm there isn't a connectivity problem you could always setup some temporary non-clustered channels and see if they start ok.
I think I've still got a test qmgr in this state so I'll have a play and see if I come up with anything.
Gareth
---------
2189 errors make me  |
|
| Back to top |
|
|
| 2189 |
| Posted: Tue Oct 22, 2002 3:02 am Post subject: |
|
|
Apprentice
Joined: 22 Oct 2002
Posts: 31
|
Oops, having reread your last post seems you've already tried creating a new qmgr with the same error. I personally think it's a repository fault which isn't gonna be easy to fix. If I do still have the same problem on a qmgr I'll investigate it and get back to you.
Gareth
----------
2189 errors make me |
|
| Back to top |
|
|
| leongor |
| Posted: Wed Oct 23, 2002 2:12 am Post subject: |
|
|
Master
Joined: 13 May 2002
Posts: 264
Location: Israel
|
Allow me ask you this basic question :
have you defined cluster-receiver channel on the new Qmgr before the cluster-sender or after ?
( Should be before )
_________________
Regards.
Leonid.
IBM Certified MQSeries Specialist. |
|
| Back to top |
|
|
| 2189 |
| Posted: Wed Oct 23, 2002 8:02 am Post subject: |
|
|
Apprentice
Joined: 22 Oct 2002
Posts: 31
|
I did have a qmgr with the exact same problem on one of my test qmgrs in a dev environment and I'm convinced in my situation that it's a problem with one of the repositories.
Whilst the channel to the repos started ok if I tried starting the channel from the repos back to the qmgr manually it said it didn't exist, so clearly despite the qmgr having contacted the repos, the repos did not know about the new qmgr.
I therefore tried altering the CLUSRCVR channel, to hopefully remind the repos that it existed. This started the channel to the repos again but no receiver started.
I then created a second CLUSSDR channel to the other repos. As soon as I created this it started and the receiver channels started and also the CLUSQMGR returned the correct results.
My next step is to fix the repos by the rather radical method of clearing out it's repository to start it again from scratch.
Hope this is of use.
_________________
2189 errors make me |
|
| Back to top |
|
|
|
|
