Author |
Message
|
OzgurAydin |
Posted: Tue Oct 24, 2017 1:27 am Post subject: Cluster Queue not up to date |
|
|
 Apprentice
Joined: 08 Sep 2008 Posts: 25
|
Hi,
We have a MQ Cluster consisting of multiple queue managers residing on both the mainframe side and the open side. Last weekend we changed one queue property of a mainframde queue. We had a maintanence break and we did put inhibited one queue. All the Queue Manager got their cluster queues refreshed and up to date. But one of them did not change the status of the clustered queue.
The one WHO did not change was a partial repository. Did anyone of you had such an issue with MQ Cluster ?
The version of the open side is 7.5
The version of the mainframe MQ is 9
Thanks |
|
Back to top |
|
 |
fjb_saper |
Posted: Tue Oct 24, 2017 5:26 am Post subject: |
|
|
 Grand Poobah
Joined: 18 Nov 2003 Posts: 19619 Location: LI,NY
|
Avenues of research:
Check the communication between full reps. Was the full rep the PR is attaching to correctly updated?
Check the communication between the PR and the full reps. Does it flow as expected?
Check the queue bindings, on the putting app, are they bindings not fixed?  _________________ MQ & Broker admin |
|
Back to top |
|
 |
bruce2359 |
Posted: Tue Oct 24, 2017 6:40 am Post subject: Re: Cluster Queue not up to date |
|
|
 Poobah
Joined: 05 Jan 2008 Posts: 7979 Location: US: west coast, almost. Otherwise, enroute.
|
OzgurAydin wrote: |
But one of them did not change the status of the clustered queue. |
On the affected qmgr, are the CLUSSDR and CLUSRCVR channels to/from the FR in RUNNING state?
Are the CLUSSDR and CLUSRCVR channels between the two FRs in RUNNING state? _________________ I would tell you a UDP joke, but you might not get it. |
|
Back to top |
|
 |
tczielke |
Posted: Tue Oct 24, 2017 8:58 am Post subject: |
|
|
Yatiri
Joined: 08 Jul 2010 Posts: 691 Location: Illinois, USA
|
Yes, I have seen quirky behavior like this before with clustering, and the underlying issue is not always a connectivity issue between the partial and full repositories. MQ clustering seems to have quirks, based on my own personal experience.
Please note that IBM recommends that you check all the partial repositories after doing any kind of clustering change like this to confirm that the command did fully take.
https://www.ibm.com/support/knowledgecenter/SSFKSJ_9.0.0/com.ibm.mq.adm.doc/q021225_.htm#q021225_ _________________ MQ administrator since 2010. |
|
Back to top |
|
 |
fjb_saper |
Posted: Wed Oct 25, 2017 5:04 am Post subject: |
|
|
 Grand Poobah
Joined: 18 Nov 2003 Posts: 19619 Location: LI,NY
|
tczielke wrote: |
Yes, I have seen quirky behavior like this before with clustering, and the underlying issue is not always a connectivity issue between the partial and full repositories. MQ clustering seems to have quirks, based on my own personal experience.
Please note that IBM recommends that you check all the partial repositories after doing any kind of clustering change like this to confirm that the command did fully take.
https://www.ibm.com/support/knowledgecenter/SSFKSJ_9.0.0/com.ibm.mq.adm.doc/q021225_.htm#q021225_ |
For a check if the operation was successfully communicated to the offending PR, and as it is a queue inhibit, you would need to check that the operation is reflected in each of the FRs, that the PR is subscribed for the queue at at least one of the FRs and that the PR's SCCQ (system.cluster.command.queue) is being processed by the queue manager and is empty...
Even in a small cluster it pays to have the FRs only carry cluster command traffic and no data traffic as it helps isolate any communications problems between command traffic between FR and PR.
If the FRs did not get the queue inhibit information, then clearly you have done something wrong in the handling of communications between the originating PR and the FRs.
Have fun  _________________ MQ & Broker admin |
|
Back to top |
|
 |
|