Author |
Message
|
bustacap |
Posted: Tue Mar 20, 2007 6:36 am Post subject: Queue manager ending automatically |
|
|
Novice
Joined: 20 Mar 2007 Posts: 10 Location: Toronto, Canada
|
Hi there, I'm new to MQ and was just wondering if in doubt transactions will causes a queue manager to shut down automatically? See an excerpt of my error log below. Since there are many people who have access to this server, I'm not sure if the shut down command was entered manually by another administrator or if the queue manager shut down by itself.
-------------------------------------------------------------------------------
03/05/07 04:40:23
AMQ7605: The XA resource manager MQSeries has returned an unexpected return
code -3, when called for xa_commit.
EXPLANATION:
WebSphere MQ received an unexpected return code when calling XA resource
manager MQSeries at its xa_commit entry point. This indicates an internal
error, either within MQ or the resource manager.
ACTION:
Try to determine the source of the error. A trace of the failure could be used
to look at the XA flows between MQ and the resource manager. MQ has allocated
an RMId of 0 to this resource manager. This will be useful when isolating the
flows associated with the resource manager concerned. If the error occurs on an
xa_commit or xa_rollback request, the queue manager will not attempt to
redeliver the commit or rollback instruction for this transaction, until after
the queue manager has been restarted. The transaction indoubt is identified by
the following XID of
X'abc123hex'. If you think
that the error lies within the queue manager, contact your IBM support center.
Do not discard any information describing the problem until after the problem
has been resolved.
----- amqtmcma.c : 855 --------------------------------------------------------
03/05/07 04:40:24
AMQ7606: A transaction has been committed but one or more resource managers
have backed out.
EXPLANATION:
WebSphere MQ was processing the commit operation for a transaction involving
external resource managers. One or more of these resource managers failed to
obey the commit request and instead rolled back their updates. The outcome of
the transaction is now mixed and the resources owned by these resource managers
may now be out of synchronization. MQ will issue further messages to indicate
which resource managers failed to commit their updates.
ACTION:
The transaction with the mixed outcome is identified by the following XID of
X'abc123hex'. The messages
which identify the failing resource managers will also contain this same XID.
If the transaction has completed it won't be displayed by the dspmqtrn command
and all other transaction participants will have committed their updates. If
the transaction is displayed by the dspmqtrn command then there are some
participants still in prepared state. In order to preserve data integrity you
will need to perform recovery steps local to the failing resource managers.
----- amqtmfga.c : 353 --------------------------------------------------------
03/05/07 08:30:24
AMQ9542: Queue manager is ending.
EXPLANATION:
The program will end because the queue manager is quiescing.
ACTION:
None. |
|
Back to top |
|
 |
Michael Dag |
Posted: Tue Mar 20, 2007 6:49 am Post subject: |
|
|
 Jedi Knight
Joined: 13 Jun 2002 Posts: 2607 Location: The Netherlands (Amsterdam)
|
look at the times, there is 3:50 hours in between...
quiescing is something that results from a normal shutdown...
so look at your syslogs and see what commands were entered, if you can find any. _________________ Michael
MQSystems Facebook page |
|
Back to top |
|
 |
bustacap |
Posted: Tue Mar 20, 2007 6:52 am Post subject: |
|
|
Novice
Joined: 20 Mar 2007 Posts: 10 Location: Toronto, Canada
|
actually... I'm pretty sure the queue manager shut down by itself, but I just want to get a second opinion on whether this is actually possible.
Thanks. |
|
Back to top |
|
 |
Vitor |
Posted: Tue Mar 20, 2007 6:55 am Post subject: |
|
|
 Grand High Poobah
Joined: 11 Nov 2005 Posts: 26093 Location: Texas, USA
|
bustacap wrote: |
actually... I'm pretty sure the queue manager shut down by itself, but I just want to get a second opinion on whether this is actually possible.
|
The log you posted looks like the response to an endmqm command IMHO. Here's a question - how soon after the shutdown was it restarted? If there's a short gap, it adds weight to the theory that someone bounced it when they got in that morning. If there's a longer gap, it implies that it was restarted after someone noticed it had closed itself down.
If the queue manager had decided to shut down, it should have shut down not quiesced. There should also be a raft of messages describing why it was taking the action & probably some FDC files as such a shutdown should be the last resort after an unrecoverable error. _________________ Honesty is the best policy.
Insanity is the best defence.
Last edited by Vitor on Tue Mar 20, 2007 6:58 am; edited 1 time in total |
|
Back to top |
|
 |
Michael Dag |
Posted: Tue Mar 20, 2007 6:57 am Post subject: |
|
|
 Jedi Knight
Joined: 13 Jun 2002 Posts: 2607 Location: The Netherlands (Amsterdam)
|
as far as I can remember I have never seen a queuemanager shutdown (not crash! ) by itself...
what platform are you on? on unix you should be able to retrieve the keyboard input from user mqm , although 5th of march seems a long time ago... _________________ Michael
MQSystems Facebook page |
|
Back to top |
|
 |
bustacap |
Posted: Tue Mar 20, 2007 7:00 am Post subject: |
|
|
Novice
Joined: 20 Mar 2007 Posts: 10 Location: Toronto, Canada
|
Very true indeed. Thanks a lot. I'll try and track down the culprit!!
 |
|
Back to top |
|
 |
Vitor |
Posted: Tue Mar 20, 2007 7:10 am Post subject: |
|
|
 Grand High Poobah
Joined: 11 Nov 2005 Posts: 26093 Location: Texas, USA
|
Michael Dag wrote: |
as far as I can remember I have never seen a queuemanager shutdown (not crash! ) by itself...
|
There's a thin line between a crash and a very urgent, immediate, unstoppable shutdown.  _________________ Honesty is the best policy.
Insanity is the best defence. |
|
Back to top |
|
 |
jefflowrey |
Posted: Tue Mar 20, 2007 7:13 am Post subject: |
|
|
Grand Poobah
Joined: 16 Oct 2002 Posts: 19981
|
I've seen queue managers shut down by themselves, when for example the filesystem is completely full.
Nothing quite like seeing a pile full of FDC files, all 0 bytes long, to make your day *exciting*. _________________ I am *not* the model of the modern major general. |
|
Back to top |
|
 |
Michael Dag |
Posted: Tue Mar 20, 2007 7:18 am Post subject: |
|
|
 Jedi Knight
Joined: 13 Jun 2002 Posts: 2607 Location: The Netherlands (Amsterdam)
|
jefflowrey wrote: |
I've seen queue managers shut down by themselves, when for example the filesystem is completely full.
Nothing quite like seeing a pile full of FDC files, all 0 bytes long, to make your day *exciting*. |
but do they quiesce then or end immediately? _________________ Michael
MQSystems Facebook page |
|
Back to top |
|
 |
Vitor |
Posted: Tue Mar 20, 2007 7:18 am Post subject: |
|
|
 Grand High Poobah
Joined: 11 Nov 2005 Posts: 26093 Location: Texas, USA
|
jefflowrey wrote: |
Nothing quite like seeing a pile full of FDC files, all 0 bytes long, to make your day *exciting*. |
I said they'd be FDC files. Never claimed there would be anything in them.....  _________________ Honesty is the best policy.
Insanity is the best defence. |
|
Back to top |
|
 |
jefflowrey |
Posted: Tue Mar 20, 2007 8:33 am Post subject: |
|
|
Grand Poobah
Joined: 16 Oct 2002 Posts: 19981
|
I'm not sure there's a significant difference in this case between quiesce and end immediately. If I recall correctly, it probably tried to quiesce.
And, of course, there would be data in the FDCs if there was any space to write the data!
Regardless, though, this wasn't the case for bustacap, because there was a notice in the AMQERR logs that the qmgr was quiescing. If the file system was full, there wouldn't be room to write that notice either... _________________ I am *not* the model of the modern major general. |
|
Back to top |
|
 |
|