| Author |
Message
|
| cd |
 Posted: Wed Nov 08, 2006 8:20 am Post subject: amqzlaa0 -f parameter |
 |
|
Novice
Joined: 16 May 2001
Posts: 20
|
Hi all,
I am running MQSeries 5.3 at CSD11 on AIX 5.3.
I am looking for information on the parameters to the amqzlaa0 process. Specifically, what is the -f parameter value used for? What does it tell me?
For example:
The filtered results of a ps -ef command gives me the following entry, among others:
amqzlaa0 -mPABLOP1 -fip27
What does the 'ip27' value mean?
Can I use this to associate this amqzlaa0 process to some other process(es) on this AIX server?
Thanks,
Curt |
|
| Back to top |
|
 |
| Vitor |
| Posted: Wed Nov 08, 2006 8:26 am Post subject: |
|
|
Grand High Poobah
Joined: 11 Nov 2005
Posts: 26093
Location: Texas, USA
|
amqzlaa0 is an internal MQ process. It's a very bad idea to be messing with these directly. IIRC this is the agent process for the queue manager named in the -m flag, but what advantage this knowledge gives you is questionable.
What are you trying to achieve?
_________________
Honesty is the best policy.
Insanity is the best defence. |
|
| Back to top |
|
|
| cd |
| Posted: Wed Nov 08, 2006 8:33 am Post subject: |
|
|
Novice
Joined: 16 May 2001
Posts: 20
|
Hi,
The process appears to be in a tight loop, as evidenced by the various ps command cpu statistics. This is confirmed by one of our Unix capacity planners. I am trying to help someone determine if the problem is with this particular amqzlaa0 instance itself, or with something that is associated with it. Past experience tells us that killing the process is a bad idea. Since it is production, I am limited as to when I can recycle the queue manager. So I am trying to investigate this specific amqzlaa0 instance. Working with MQSeries traces and AIX system level traces has not provided anything meaningful yet. Any other suggestions on what to look at would also be much appreciated.
Thanks,
Curt |
|
| Back to top |
|
|
| Vitor |
| Posted: Wed Nov 08, 2006 8:39 am Post subject: |
|
|
Grand High Poobah
Joined: 11 Nov 2005
Posts: 26093
Location: Texas, USA
|
| cd wrote: |
| Past experience tells us that killing the process is a bad idea. |
That I believe! 
| cd wrote: |
| Since it is production |
Again with the belief!
Frankly if the queue manager's not under exceptional load (or perhaps even if it is) I'd consider raising a PMR with IBM. Before doing that you might consider putting on CSD12 - they'll almost certainly suggest that so steal a march. 
Aside from that, I'd look into any use of XA with an external coordinator. Also long running transactions - make sure they're commiting units of work in a timely fashion.
_________________
Honesty is the best policy.
Insanity is the best defence. |
|
| Back to top |
|
|
| cd |
| Posted: Wed Nov 08, 2006 10:44 am Post subject: |
|
|
Novice
Joined: 16 May 2001
Posts: 20
|
Thanks for your comments. I appreciate the help.
The MQ manuals are silent as to parameter definitions for amqzlaa0. I would still like to know what is the -f parameter, and what is the significance of the value passed.
Thanks again,
Curt |
|
| Back to top |
|
|
| jefflowrey |
| Posted: Wed Nov 08, 2006 10:52 am Post subject: |
|
|
Grand Poobah
Joined: 16 Oct 2002
Posts: 19981
|
It probably does you absolutely no good to know these things.
_________________
I am *not* the model of the modern major general. |
|
| Back to top |
|
|
| mvic |
| Posted: Wed Nov 08, 2006 2:46 pm Post subject: |
|
|
Jedi
Joined: 09 Mar 2004
Posts: 2080
|
| cd wrote: |
| The process appears to be in a tight loop, as evidenced by the various ps command cpu statistics. |
Could you share a few more details of this evidence please? And what is the definition of "in a tight loop" being used here? |
|
| Back to top |
|
|
| Vitor |
| Posted: Thu Nov 09, 2006 12:31 am Post subject: |
|
|
Grand High Poobah
Joined: 11 Nov 2005
Posts: 26093
Location: Texas, USA
|
| cd wrote: |
| The MQ manuals are silent as to parameter definitions for amqzlaa0. |
This is a) because to write it in the manual will result in someone with less reason than yourself fiddling with it to "improve" the product and getting in a terrible mess and b) because it's an internal MQ process, is no more documented than the rest of the internal workings of the software and probably protected by copyright / intellectual property right / some other kind of right.
If you really want to know what it means, raise a PMR with IBM and ask them. I'm sure with proper legal protection (signed forms and so forth) they'll give you the information.
I still remain unconvinced how much help this will be to you. 
_________________
Honesty is the best policy.
Insanity is the best defence. |
|
| Back to top |
|
|
| cd |
| Posted: Thu Nov 09, 2006 8:25 am Post subject: |
|
|
Novice
Joined: 16 May 2001
Posts: 20
|
Hi,
Okay, I will be content to let internals be internals...
As to mvic's question about how I know a specific amqzlaa0 process is in a tight loop...
I can use the ps command and see that the CPU time consumed by the process is keeping up second for second with clock/elapsed time, and has been doing so for the past 6 days. We also have a monitoring tool here named Glance Plus that is telling me that this specific amqzlaa0 process is constantly using between 97% and 98% of a CPU on a 4-CPU server, and never decreases.
No applications people are complaining about their applications that use MQSeries on this same server. I am wondering if this specific amqzlaa0 process is actually being used by any applications. I am trying to make the association between one or more applications and this specific amqzlaa0 process, but I do not know how to do that. Any suggestions?
We currently have a PMR open with IBM on this.
Thanks,
Curt |
|
| Back to top |
|
|
| Vitor |
| Posted: Thu Nov 09, 2006 8:36 am Post subject: |
|
|
Grand High Poobah
Joined: 11 Nov 2005
Posts: 26093
Location: Texas, USA
|
| cd wrote: |
Okay, I will be content to let internals be internals...
|
Or at least let their parents at Hursley look after them!
In so far as I'm aware (and I'm the first to admit that it's not far) this process is connected to the XA-compliant part of the queue manager; hence my initial suggestion for problem investigation. I would doubt that there's any kind of direct link between the queue manager processes and the applications in use and even if you find one, what good does it do you? If there's no application issues being reported then the applications are performing normally even in the face of the "looping" process, so whatever it does is not likely caused by application action & changing the application will not therefore solve your problem. You've already proved that tampering with the process can have dire consequences so if you do establish cause and effect your courses of action are somewhat limited.
I agree that establishing a link would provide additional information for the PMR, but you're still reliant on IBM for a resolution.
Bottom line - it's broke, you need IBM to fix it. Where "fix it" may involve telling you what you've done to confuse the little mite. IMHO it should be impossible for user action to cause system level software to fail like this, but we live in an imperfect world. 
Do post any resolution or information you get - I'm interested to see how this pans out.
_________________
Honesty is the best policy.
Insanity is the best defence. |
|
| Back to top |
|
|
| mvic |
| Posted: Thu Nov 09, 2006 8:46 am Post subject: |
|
|
Jedi
Joined: 09 Mar 2004
Posts: 2080
|
| cd wrote: |
| We also have a monitoring tool here named Glance Plus that is telling me that this specific amqzlaa0 process is constantly using between 97% and 98% of a CPU on a 4-CPU server, and never decreases. |
Thanks for the extra info. High CPU usage could just be sign that the queue manager is doing a lot of work for apps. The fact that the app people don't complain is definitely encouraging in this regard. In case it helps: when I run a CPU-intensive task like a big code build, I expect the machine to have high CPU usage until the task is finished. I guess IBM Support will work with you to identify whether the high CPU usage you see is bad or good. |
|
| Back to top |
|
|
| mvic |
| Posted: Thu Nov 09, 2006 8:52 am Post subject: |
|
|
Jedi
Joined: 09 Mar 2004
Posts: 2080
|
| Vitor wrote: |
| In so far as I'm aware ... this process is connected to the XA-compliant part of the queue manager |
amqzlaa0 is a part of the queue manager that serves your application connections. Whether the apps use XA or not does not alter this.
| Quote: |
| I would doubt that there's any kind of direct link between the queue manager processes and the applications in use |
One amqzlaa0 generally serves many apps. If an amqzlaa0 is doing work, it's doing it on behalf of an app - including user apps, channel programs, trigger monitor etc. |
|
| Back to top |
|
|
| malammik |
| Posted: Thu Nov 09, 2006 11:52 am Post subject: |
|
|
Partisan
Joined: 27 Jan 2005
Posts: 397
Location: Philadelphia, PA
|
| Vitor wrote: |
amqzlaa0 is an internal MQ process. It's a very bad idea to be messing with these directly. IIRC this is the agent process for the queue manager named in the -m flag, but what advantage this knowledge gives you is questionable.
|
The sky will break open an unleash the hell on all of us. If you dont know the answer and dont care than just be quiet. I dont think knowing more about the product inside or outside has ever hurt anyone.
_________________
Mikhail Malamud
http://www.netflexity.com
http://groups.google.com/group/qflex |
|
| Back to top |
|
|
| jefflowrey |
| Posted: Thu Nov 09, 2006 11:54 am Post subject: |
|
|
Grand Poobah
Joined: 16 Oct 2002
Posts: 19981
|
Nobody here has said that having the knowledge will hurt, just that it won't help.
_________________
I am *not* the model of the modern major general. |
|
| Back to top |
|
|
| malammik |
| Posted: Thu Nov 09, 2006 12:00 pm Post subject: |
|
|
Partisan
Joined: 27 Jan 2005
Posts: 397
Location: Philadelphia, PA
|
I just get this feeling that some ppl are over protective. "Boo that internal, thou shall not possess this knowledge." Guy wants to know what it is even it is completely useless, its his right!. I just dont like this overprotective white-box mentality e.g. you cannot decompile java code or you dont need to know what -f flag is. Really bothers me.
_________________
Mikhail Malamud
http://www.netflexity.com
http://groups.google.com/group/qflex |
|
| Back to top |
|
|
|
|
