| Author |
Message
|
| Bear |
 Posted: Tue Jan 17, 2006 10:09 am Post subject: amqiclen and chkmqipc |
 |
|
Voyager
Joined: 17 Jan 2006
Posts: 28
Location: Montreal Canada
|
| We have MQ 5.3 on Sun Solaris and having QMs hang because of the number of MQ agents spawned and excess use of resourses such as memory and semaphores. Often we have had to force a shutdown of the QM. It is impossible to do it graciously. I have found on IBM support and independent forums many references to this problem and in some cases it is suggested that we use amqicln and chkmqipc but these seem to be something you use after the shutdown not while the QM is running. Has anyone used these facilities in an attempt to keep the QM up and restore it to good health. OR, is there another facility that we can attempt to save our QM from a forced shutdown? Our QMs run 24/7 and should rarely shut down except for maintenance. Thanx for any assistance here. BB. |
|
| Back to top |
|
 |
| wschutz |
| Posted: Tue Jan 17, 2006 11:59 am Post subject: |
|
|
Jedi Knight
Joined: 02 Jun 2005
Posts: 3316
Location: IBM (retired)
|
I don't think you want to use those commands on running qmgrs, it probably wouldn't be pretty.
Have you tried opening a PMR to see what service has to suggest?
Are you up-to-date on CSDs on these machines?
also, please don't DOUBLE post...
_________________
-wayne |
|
| Back to top |
|
|
| fjb_saper |
| Posted: Tue Jan 17, 2006 3:56 pm Post subject: |
|
|
Grand High Poobah
Joined: 18 Nov 2003
Posts: 20756
Location: LI,NY
|
Wayne
Please remember that you can run amqiclen in 2 different modes:
a) display -c
b) delete -x
The display mode does just what it says. It displays the shared memory usage.
The delete mode can be used once a qmgr has been shut down and avoids numerous ipcrm commands... after a crash ...
You can use it either for the full system -s
or for a specific qmgr -m <qmgr>
I don't know about using -x on a running qmgr (I would not try that) but I certainly use -xF (Force) on a queue manager that ended unexpectedly...(eyecatcher memory problem ... moving it to CSD 11) after having killed the remaining processes.
Enjoy 
_________________
MQ & Broker admin |
|
| Back to top |
|
|
| csmith28 |
| Posted: Tue Jan 17, 2006 4:43 pm Post subject: |
|
|
Grand Master
Joined: 15 Jul 2003
Posts: 1196
Location: Arizona
|
Please do not post the same topic twice.
_________________
Yes, I am an agent of Satan but my duties are largely ceremonial. |
|
| Back to top |
|
|
| SAFraser |
| Posted: Wed Jan 18, 2006 10:07 am Post subject: |
|
|
Shaman
Joined: 22 Oct 2003
Posts: 742
Location: Austin, Texas, USA
|
Am I the only curious person.... Why are these resources being orphaned?
When we have had this happen, we worked pretty hard to find the root cause and then correct it. Because (in my experience) you either shut down your queue manager and clean up the orphaned resources, or the queue manager will eventually stop responding and you must shut down anyway.
Do you know the cause, Bear?
Shirley |
|
| Back to top |
|
|
| Bear |
| Posted: Wed Jan 18, 2006 12:27 pm Post subject: Various |
|
|
Voyager
Joined: 17 Jan 2006
Posts: 28
Location: Montreal Canada
|
First, sorry for the doule post.
Don't know the cause of the orphaned processes, still checking. I want to find a way to improve the health of the QM and hopefully not shut it down while clients still attached. And then I will dig deeper for the root cause. I have been looking at the TuningParameters stanza in the qm.ini file to see if any of these might alleviate the excessive use of resources such as the AgentClassLimit and the DefaultCellGroup. If any can suggest if I'm barking up the wrong tree I would appreciate muchly. Cheers. |
|
| Back to top |
|
|
| PeterPotkay |
| Posted: Wed Jan 18, 2006 5:05 pm Post subject: |
|
|
Poobah
Joined: 15 May 2001
Posts: 7722
|
Are you at base 5.3, or do you have any CSDs applied?
You did go ahead and tweak all the Kernel settings as described in the Quick Beginings manual for your platform, right?
_________________
Peter Potkay
Keep Calm and MQ On |
|
| Back to top |
|
|
| Bear |
| Posted: Thu Jan 19, 2006 6:52 am Post subject: |
|
|
Voyager
Joined: 17 Jan 2006
Posts: 28
Location: Montreal Canada
|
| We are at Fix Pack (CSD) 11 on V5.3. We tend to keep up to date as much as possible without trying to be bleeding edge. Yes, we have tweaked the kernel parms as suggested in thd QB guide. I am new to the project and have experience on AIX but the other MQ folks here assure me every thing is up to date and well tested and tried. We have several production Solaris platforms and lots of QMs that run 24/7 and we do everything to keep them alive and well. Thanx for you interest and help. BB. |
|
| Back to top |
|
|
| Tibor |
| Posted: Thu Jan 19, 2006 10:12 am Post subject: |
|
|
Grand Master
Joined: 20 May 2001
Posts: 1033
Location: Hungary
|
Bear,
Those values in the guide 'Quick Beginnings' are only for recommendation. If other applications are running on your server, your settings of shared memory segments or semaphores may be too slight... 
Tibor |
|
| Back to top |
|
|
| fjb_saper |
| Posted: Thu Jan 19, 2006 7:39 pm Post subject: |
|
|
Grand High Poobah
Joined: 18 Nov 2003
Posts: 20756
Location: LI,NY
|
And remember that with CSD 11 there is an APAR about producers having a memory leak....
Enjoy 
_________________
MQ & Broker admin |
|
| Back to top |
|
|
|
|
