| |
|
RSS Feed - WebSphere MQ Support
|
RSS Feed - Message Broker Support
|
| Workload balancing not working as expected |
« View previous topic :: View next topic » |
| Author |
Message
|
| JasonE |
 Posted: Mon Oct 04, 2004 2:38 am Post subject: |
 |
|
Grand Master
Joined: 03 Nov 2003
Posts: 1220
Location: Hursley
|
| I think its in a chinit dump, formatted with clus=1 or 2 (Dont quote me, just some rough notes I have). Its the same info though, laid out in pretty much the same format |
|
| Back to top |
|
 |
| cicsprog |
| Posted: Tue Nov 09, 2004 1:02 pm Post subject: |
|
|
Partisan
Joined: 27 Jan 2002
Posts: 314
|
Litle more on what JasonE said:
You may want to take an SVC dump of MQxxCHIN (mover). Then thru IPCS do a VERBEXIT CSQXDPRD,CLUS=1. IPCS and SVC dumps are not too easy for first timers. Th CLUS=1 will give you Cluster report including the cluster repository known on the queue manager. See z/OS Problem Determination Guide.
Also, as stated before, thru the ISPF Dialogs or via modify do a DISPLAY CLUSQMGR(*) ALL. That should report states of MQM's in the CLUSTER(s).
Lastly, ask IBM if this utility could help:ts/fromibm/mqseries files CSQ4CCHE.EXPORT and CSQ4CCHE.INFO. I used this utility in the CLUSTER corruption problem.
Best
Jason |
|
| Back to top |
|
|
| mverh |
| Posted: Fri Jan 14, 2005 6:40 am Post subject: |
|
|
Voyager
Joined: 06 Mar 2002
Posts: 97
|
I've read this thread and was also surprised that the balancing algorithm worked as Nigel stated. I do not disagree with he states but I have an issue with it.
I have 3 qmgrs, one partial repos (PR), and two full (F1, F2). The two full repos house a clustered appl queue Q1.
I am sending messages from the partial repos to Q1 on the full repos.
My channel seq num are: PR->F1: 1577756, PR->F2: 1141
When I run my app with bind option of not fixed all messages go to Q1 on F2. I assume because the seq no on the channel PR-F2 is lower.
I now want to evenly load balance so I look at doing a "reset channel seqnum(1)" on the PR cluster sender channels but this has no effect.
How do I get the seq numbers reset on the cluster sender channels so I can truly load balance? |
|
| Back to top |
|
|
| PeterPotkay |
| Posted: Fri Jan 14, 2005 2:42 pm Post subject: |
|
|
Poobah
Joined: 15 May 2001
Posts: 7716
|
You don't need to do this.
Reopen the queue, start putting the messages, and as longs as BOTH instances of the queue on FR1 and FR2 are enabled, and you opened the queue with BIND_NOT_ FIXED, and both channels can start, the message will round robin. Trust me. They will. If they don't something else is wrong, or you have another app running that is increasing the seq # of one of the channels very quickly. Make sure you put enough messages to let the algorithim do its things. If both chabbels are Inactive and you put just 2 messages, the first channel that starts up will take all the messages before the second channel fires up. It will look like round robining is not working. Either put a lot of messages (like a 100) to be sure both channels have a shot at the messages, or make sure both channels are RUNNING if you are only going to put 2 messages.
Cluster Administrative messages are often shipped between the FRs and the PRs. Those messages up the sequence numbers. As do other applications. I don't think IBM would put out a product that needed its channels manually reset every time you needed round robining to occur.
What version are you at? I just tested this again for the umpteenth time, replicating your scenrio, and had no problems at all. I am at 5.3 CSD8.
_________________
Peter Potkay
Keep Calm and MQ On |
|
| Back to top |
|
|
|
|
|
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
|
|
|
