| Author |
Message
|
| Knut |
 Posted: Wed May 27, 2009 4:38 am Post subject: AMQ9219 on MQ6 HP/UX |
 |
|
Newbie
Joined: 27 May 2009
Posts: 6
|
HELLO!
First of all please excuse some rough spelling here since this isn't my native language.
One of our system are running MQ 6 on a HPUX system. We suddenly( with no other messages in the log the previous 6 hours) got the error :
05/27/09 01:09:56 - Process(13646.1) User(rtgs) Program(runmqlsr_nd)
AMQ9219: The TCP/IP listener program could not create a new connection for the
incoming conversation.
EXPLANATION:
An attempt was made to create a new socket because an attach request was received, but an error occurred.
We have been running MQ for many,many years but mainly on mainframe. But we have never experienced this error before. Can't find any good explanation for this but maybe some of you have run into this situation ??
Best regards Knut |
|
| Back to top |
|
 |
| Vitor |
| Posted: Wed May 27, 2009 4:48 am Post subject: |
|
|
Grand High Poobah
Joined: 11 Nov 2005
Posts: 26093
Location: Texas, USA
|
Why is the listener not running as mqm? How is the listener configured? What did / does netstat show? What level of WMQv6 are you using?
_________________
Honesty is the best policy.
Insanity is the best defence. |
|
| Back to top |
|
|
| Knut |
| Posted: Wed May 27, 2009 5:32 am Post subject: |
|
|
Newbie
Joined: 27 May 2009
Posts: 6
|
Lots of good questions.
the user rtgs is a member of the mqm group and have the same authoroties as mqm.
The version of WMQ is 6.02
The listener is defined on the basis of (SYSTEM.DEFAULT.LISTENER.TCP)
(LISTENER(SYSTEM.LISTENER.TCP.1) CONTROL(MANUAL)
TRPTYPE(TCP) PORT(1414)
IPADDR( ) BACKLOG(0)
DESCR( ) ALTDATE(2007-05-11)
ALTTIME(21.44.37)
Since this happend long before the chickens were up I am affraid we don't have any netstat. But this particular MQenvironment has been running since May2007. |
|
| Back to top |
|
|
| Vitor |
| Posted: Wed May 27, 2009 5:37 am Post subject: |
|
|
Grand High Poobah
Joined: 11 Nov 2005
Posts: 26093
Location: Texas, USA
|
| Knut wrote: |
| the user rtgs is a member of the mqm group and have the same authoroties as mqm. |
I repeat my question. The mqm user holds a special place in the WMQ hierarchy on distributed platforms.
| Knut wrote: |
| Since this happend long before the chickens were up I am affraid we don't have any netstat. But this particular MQenvironment has been running since May2007. |
I would suspect something bad happened on the box but there's not much evidence. I'd recommend changing the listener to run under the control of the queue manager rather than manually as it may help recovery from this sort of thing, but it may not help that much (!).
Other than that, capture some diagnostics when / if it happens again. Unless someone else in here has further, more useful insights.
_________________
Honesty is the best policy.
Insanity is the best defence. |
|
| Back to top |
|
|
| mqjeff |
| Posted: Wed May 27, 2009 5:50 am Post subject: |
|
|
Grand Master
Joined: 25 Jun 2008
Posts: 17447
|
| Revisit your platform QuickBeginnings guide for system setup for shmat and etc... and revisit your MaxActiveChannels and etc. etc. etc. |
|
| Back to top |
|
|
| Knut |
| Posted: Wed May 27, 2009 9:50 pm Post subject: |
|
|
Newbie
Joined: 27 May 2009
Posts: 6
|
Thank you for the input.
Forgot to mentioned that the probably worst thing with this was that the tcp listener ENDED!!!
----- amqclita.c : 928 --------------------------------------------------------
05/27/09 01:09:56 - Process(13619.1) User(rtgs) Program(amqzmgr0)
AMQ5027: The Listener 'SYSTEM.LISTENER.TCP.1' has ended. ProcessId(13646).
EXPLANATION:
The Listener process has ended.
ACTION:
None. |
|
| Back to top |
|
|
| Vitor |
| Posted: Thu May 28, 2009 12:13 am Post subject: |
|
|
Grand High Poobah
Joined: 11 Nov 2005
Posts: 26093
Location: Texas, USA
|
| Knut wrote: |
Thank you for the input.
Forgot to mentioned that the probably worst thing with this was that the tcp listener ENDED!!!
|
This additional information would have given up a clue as to why new TCP/IP connections were problematic... 
So something bad did happen on the box and it brought the listener down. I repeat my advice to put it under qmgr control.
And run it as mqm.
_________________
Honesty is the best policy.
Insanity is the best defence. |
|
| Back to top |
|
|
| Andyh |
| Posted: Sat Dec 10, 2022 9:12 am Post subject: |
|
|
Master
Joined: 29 Jul 2010
Posts: 239
|
Even if the listener failed to start a new connection it shouldn't have ended.
MQ V6 is pretty ancient, do you have an extended support contract? in which case you should take this up with IBM to see if it's a known issue.
The fact the listener did end might indicate a simple error handling issue on a code path not commonly taken, or might indicate a more fundamental issue affecting the whole listener process, for example if the listener had been running for a very long time might it have run out of some resource.
Accepting a new connection would involve creating a new file handle, so perhaps the process has run out of file handles (just speculation).
You might consider monitoring the number of file handles open in the listener process. If the listener were to be leaking file handles then a periodic restart might be in order (or upgrading to a more current service level). |
|
| Back to top |
|
|
|
|
