| Author |
Message
|
| abd.wsu |
 Posted: Tue Mar 03, 2015 8:48 am Post subject: Stale cluster information in queue manager |
 |
|
Acolyte
Joined: 12 Sep 2012
Posts: 65
|
Hello,
We have run into a minor issue with one our gateway queue managers. The queue manager was shared between 2 different clusters A & B.Recently one of my teammates performed a task to remove our queue manager from aCLUSB.
But after the task, we still see the CLUSB information on the queue manager. The queue manager is not advertised in the FR of CLUSB. All other queue managers in CLUSB don't have any info of the queue manager.
So I would say the task worked fine.
But, we have a autodefined clussdr to the FR on CLUSB in stopped state on the queue manager. We had deleted the manually defined CLUSSDR to FR on CLUSB as part of the task. And all the info about CLUSB is still on the queue manager.
We have message buildup on the SCTQ to the FR of CLUSB.
Now, my colleague swears he did all the steps correctly and ran a refresh cluster command after the queue manager was removed from the cluster, although I am not sure if he ran a refresh cluster repos(YES).
I am just confused if everything was done correctly as per IBM documentation why the strange behaviour.
Now, what are my options to clean it up. Would running a refresh cluster(*) repos(YES) fix the issue?
I know that the retained info would be erased after 90 days but i am not sure if that will happen here?
Please suggest. |
|
| Back to top |
|
 |
| JosephGramig |
| Posted: Tue Mar 03, 2015 12:47 pm Post subject: |
|
|
Grand Master
Joined: 09 Feb 2006
Posts: 1244
Location: Gold Coast of Florida, USA
|
Is the auto CLUSSDR still in stopped state?
Do any objects have the cluster info specified?
If no, try refresh cluster(bla) repos(yes) again. |
|
| Back to top |
|
|
| abd.wsu |
| Posted: Tue Mar 03, 2015 1:10 pm Post subject: |
|
|
Acolyte
Joined: 12 Sep 2012
Posts: 65
|
The auto CLUSSDR is in stopped state.
No objects refering to CLUSB. I am worried if running the refresh command will dump a bunch of messages on the SCTQ where there are all ready messages building up. |
|
| Back to top |
|
|
| fjb_saper |
| Posted: Wed Mar 04, 2015 5:23 am Post subject: |
|
|
Grand High Poobah
Joined: 18 Nov 2003
Posts: 20756
Location: LI,NY
|
What's the destination for the messages building up on your SCTQ?
_________________
MQ & Broker admin |
|
| Back to top |
|
|
| abd.wsu |
| Posted: Wed Mar 04, 2015 6:18 am Post subject: |
|
|
Acolyte
Joined: 12 Sep 2012
Posts: 65
|
| Looks like the first one is for SYSTEM.INTER.QMGR.CONTROL and rest of them are for SYSTEM.CLUSTER.COMMAND.QEUUE. |
|
| Back to top |
|
|
| JosephGramig |
| Posted: Wed Mar 04, 2015 7:45 am Post subject: |
|
|
Grand Master
Joined: 09 Feb 2006
Posts: 1244
Location: Gold Coast of Florida, USA
|
| abd.wsu wrote: |
| The auto CLUSSDR is in stopped state. |
This is why the cluster info does not go away. Try to change the channel state to stopped status(inactive).
When cleaning up cluster stuff, remove references to the cluster on all objects (channels last). The receiver should stop fairly quickly as the other Qmgrs get notified it is not part of the cluster any more. Then stop the sender with the status(inactive) or wait. Then give the 'refresh cluster(bla) repos(yes)' and poof it will be clean.
When you create the cluster receiver, that prepares the Qmgr to be part of the cluster. When you create the cluster sender, that injects the Qmgr into the cluster. Removal is the opposite actions. |
|
| Back to top |
|
|
| abd.wsu |
| Posted: Wed Mar 04, 2015 8:42 am Post subject: |
|
|
Acolyte
Joined: 12 Sep 2012
Posts: 65
|
@JosephGramig, I'll try that option. Hopefully it works.
Another issue(not sure if it's even an issue) was after @fjb_saber asked me, I browsed the SCTQ a bit closer than usual and saw some messages for SYSTEM.INTER.QMGR.CONTROL queue. A quick glance through the IBM documentation indicated that queue is used for Pub/Sub in the clusters.
Now, i went through my queue managers Pub/Sub setup and saw a bunch of SYSTEM.PROXY subscriptions to the queue managers in the now excluded cluster. No reference to that cluster in the topics again. No topics using the same topic strings as these Proxy subs.
Any idea how i can get rid of those? or would that be the reason for my cluster blues? |
|
| Back to top |
|
|
| rammer |
| Posted: Thu Mar 05, 2015 3:35 am Post subject: |
|
|
Partisan
Joined: 02 May 2002
Posts: 359
Location: England
|
| Hi what version of MQ do you have? There is a bug in 7.5.0.1 that is fixed in later release for not clearing Cluster QM's out of the REPOS cleanly all the time, even if you followed the process in the Knowledge Center correctly. |
|
| Back to top |
|
|
| abd.wsu |
| Posted: Thu Mar 05, 2015 6:42 am Post subject: |
|
|
Acolyte
Joined: 12 Sep 2012
Posts: 65
|
| We were on 7502 and moved to 7504 and then made the cluster changes on the same night but in that order. |
|
| Back to top |
|
|
|
|
