| Author |
Message
|
| kd |
 Posted: Mon Apr 16, 2007 6:53 am Post subject: Channel reset not persistent! |
 |
|
Novice
Joined: 10 Aug 2006
Posts: 19
|
Hi,
I've recently upgrade my 5.3.12 Queue Manager to 6.0.2.0. I'm having a strange problem with on of my RCVR channels. We got into a situation were the sequence numbers were out of sync. Here's what happens:
1. Host sends me seqnum 123, I am expecting 120. I reset to 123 and all works fine, messages flowing through
2. After no messages in a while, channel sleeps. When it comes back after new messages, host sends me seqnum 456, I am expecting 123!
3. So I reset again (to 456), and a few hours later (after sleeping and triggering again) the host sends 789, but I'm back expecting 456 again.. and so on and so on.
Sending hosts is Win2003 with MQ 5.3.
At first I thought it was because I wasn't stopping the RCVR channel before resetting, but I tried this the next time it happened and the same thing again.
I have now asked the sender to reset from their end, I'm currently waiting for an error, but don't expect it...
I have never seen this behaviour in 5.3, I was always able to reset both ends of a channel. Is it a bug introduced in version 6? Has anyone else experienced this?
Thanks in advance.
Kevin |
|
| Back to top |
|
 |
| tleichen |
| Posted: Mon Apr 16, 2007 8:20 am Post subject: |
|
|
Yatiri
Joined: 11 Apr 2005
Posts: 663
Location: Center of the USA
|
Hmmm... are both platforms Windows? Just wondering. 
_________________
IBM Certified MQSeries Specialist
IBM Certified MQSeries Developer |
|
| Back to top |
|
|
| kd |
| Posted: Mon Apr 16, 2007 8:22 am Post subject: |
|
|
Novice
Joined: 10 Aug 2006
Posts: 19
|
No,
My QM is running on Solaris 10
Thanks |
|
| Back to top |
|
|
| tleichen |
| Posted: Mon Apr 16, 2007 8:29 am Post subject: |
|
|
Yatiri
Joined: 11 Apr 2005
Posts: 663
Location: Center of the USA
|
When these would get out of sync between mainframe and unix systems, we would always reset them back to 1, I believe. I never had the synching problem that you have mentioned, but have never tried this with version 6, as yet.
_________________
IBM Certified MQSeries Specialist
IBM Certified MQSeries Developer |
|
| Back to top |
|
|
| Ward |
| Posted: Wed Dec 05, 2007 4:22 am Post subject: |
|
|
Voyager
Joined: 27 Jun 2001
Posts: 98
Location: Europe
|
Hi,
we are experiencing the exact same problem here. Receiving queue manager is a 6.0.2.1 on Solaris; sending a 5.3 queue manager...
every time the channel has gone down the receiving end expects the next sequence number to be 1.
Did you find a solution to this ? Have you logged the problem with IBM ?
Cheers, Ward. |
|
| Back to top |
|
|
| kd |
| Posted: Wed Dec 05, 2007 4:31 am Post subject: |
|
|
Novice
Joined: 10 Aug 2006
Posts: 19
|
We didn't raise with IBM, though we found a fix was to recreate channels on both sides. It seems that the sequence number object is stored in one of the SYSTEM.* queues, and in some cases the "message" can be corrupted somehow. Recreating the channels ensures the QM looks at a new message in that queue...
Hope this helps. |
|
| Back to top |
|
|
| Ward |
| Posted: Wed Dec 05, 2007 5:04 am Post subject: |
|
|
Voyager
Joined: 27 Jun 2001
Posts: 98
Location: Europe
|
| Thanks, this should do the trick here as well then... |
|
| Back to top |
|
|
| JosephGramig |
| Posted: Wed Dec 05, 2007 5:13 am Post subject: |
|
|
Grand Master
Joined: 09 Feb 2006
Posts: 1244
Location: Gold Coast of Florida, USA
|
Hmmm,
In any case, the current maintenance level is 6.0.2.2.
_________________
Joseph
Administrator - IBM WebSphere MQ (WMQ) V6.0, IBM WebSphere Message Broker (WMB) V6.1 & V6.0
Solution Designer - WMQ V6.0
Solution Developer - WMB V6.1 & V6.0, WMQ V5.3 |
|
| Back to top |
|
|
| PeterPotkay |
| Posted: Wed Dec 05, 2007 4:16 pm Post subject: |
|
|
Poobah
Joined: 15 May 2001
Posts: 7722
|
kd,
What ID are the RCVR channels running under? Do you have something coded in the MCAUSER? And/or do you have your listener running under an ID other than the mqm group?
I ran into this same problem when I upgraded from 5.3 to 6.0. Because channels are now objects in 6.0, they need the rights to be able to do certain things. If they run as an ID in the mqm group you never realize this problem. Otherwise you have to use setmqaut to grant them +ctrlx, plus one other aut that escapes me right now (I have it documented at work and will post it tmrw)
I had the same exact symptoms until I did this.
_________________
Peter Potkay
Keep Calm and MQ On |
|
| Back to top |
|
|
| fjb_saper |
| Posted: Wed Dec 05, 2007 5:04 pm Post subject: |
|
|
Grand High Poobah
Joined: 18 Nov 2003
Posts: 20756
Location: LI,NY
|
Also note that you should probably open a PMR with IBM. For the mainframe they will send you a utility to fix the problem.(multiple entries for the channel in the SYSTEM.CHANNEL.SYNCQ) 
_________________
MQ & Broker admin |
|
| Back to top |
|
|
| Nigelg |
| Posted: Wed Dec 05, 2007 11:01 pm Post subject: |
|
|
Grand Master
Joined: 02 Aug 2004
Posts: 1046
|
It is not supported to run channel programs under any user other than mqm, or its equivalent in Windows. In UNIX all the apps are setuid, so the situation cannot arise that the effective user is not mqm, unless the app permissions are changed which is definitely not supported.
The MCAUSER does not matter; this is only used to authorise access to WMQ objects, not to internal channel and qmgr data.
_________________
MQSeries.net helps those who help themselves.. |
|
| Back to top |
|
|
| PeterPotkay |
| Posted: Thu Dec 06, 2007 9:51 am Post subject: |
|
|
Poobah
Joined: 15 May 2001
Posts: 7722
|
Nigel, if you code an MCAUSER on a RCVR channel, in MQ 6.0 that ID needs +dsp and +ctrlx against the channel object if sequence # resets are to stick.
_________________
Peter Potkay
Keep Calm and MQ On |
|
| Back to top |
|
|
| kd |
| Posted: Thu Dec 06, 2007 10:06 am Post subject: |
|
|
Novice
Joined: 10 Aug 2006
Posts: 19
|
Thanks Peter,
Will give this a try next time I see the error (haven't seen it for a while though...)
Kevin |
|
| Back to top |
|
|
| PeterPotkay |
| Posted: Thu Dec 06, 2007 1:34 pm Post subject: |
|
|
Poobah
Joined: 15 May 2001
Posts: 7722
|
You got something waiting to break that you can fix easily ahead of time and you're not gonna fix it? Your jinxing yourself for this channel to have a sequence # problem at 3 AM! 
Create a new dummy SNDR/RCVR pair on this QM to test the problem and the fix, and then just implement it across the board for all your other channels.
_________________
Peter Potkay
Keep Calm and MQ On |
|
| Back to top |
|
|
| Nigelg |
| Posted: Thu Dec 06, 2007 9:39 pm Post subject: |
|
|
Grand Master
Joined: 02 Aug 2004
Posts: 1046
|
| Quote: |
Nigel, if you code an MCAUSER on a RCVR channel, in MQ 6.0 that ID needs +dsp and +ctrlx against the channel object if sequence # resets are to stick.
|
This is a bug, fixed by APAR IY97736 (qv) included in 6.0.2.3.
_________________
MQSeries.net helps those who help themselves.. |
|
| Back to top |
|
|
|
|
