| Author |
Message
|
| olan022 |
 Posted: Thu Feb 06, 2003 11:57 am Post subject: rcdmqimg and cleanmqlogs |
 |
|
Acolyte
Joined: 17 Apr 2002
Posts: 50
|
I am using the recommended settings for a daily rcdmqimg and cleanmqlogs process which normally works in cleaning out the linear logs we have setup in MQSeries for AIX. However, it's nearly been two months now and I've noticed that the cleanmqlogs has not compressed any of more linear logs for the last two months! I have about 120 4MB logs files now and am growing concerned that we could be running out of space soon.
Here are the script commands I am running:
/usr/mqm/bin/rcdmqimg -m QM1 -t all \*
/usr/local/scripts/system/cleanmqlogs -z QM1
Like I said this has worked in the past and the QMgr has been stopped and started a lot in the last two months so I can't figure out what is needed to produce a checkpoint or closeout any old transactions or whatever else could be holding this cleanup process up.
Any ideas/suggestions would be greatly appreciated. |
|
| Back to top |
|
 |
| jhalstead |
| Posted: Fri Feb 07, 2003 1:45 am Post subject: |
|
|
Master
Joined: 16 Aug 2001
Posts: 258
Location: London
|
Use the -l option in the rcdmqimg command. (This ensures that the AMQ7467 & AMQ7468 messages that are written in AMQERR01.LOG after you record your media image. Similarly this is echoed to screen (stderr)
Note down the log file numbers specified.
Now shut the qmgr down
Start the qmgr up and again check the new AMQ7467 & AMQ7468 messages in the AMQERR01.LOG.
Do these numbers correspond to those recorded earlier?
If these are not the same then there could be a problem with rcdmqimg.
Let us know how you get along
Jamie |
|
| Back to top |
|
|
| yaakovd |
| Posted: Fri Feb 07, 2003 1:47 am Post subject: |
|
|
Partisan
Joined: 20 Jan 2003
Posts: 319
Location: Israel
|
Try to find following message in MQSeries log (looking for newest):
"The oldest log file required to perform media recovery of queue manager ..."
It will help you to understand where the problem in QMgr or in using the script command.
_________________
Best regards.
Yaakov
SWG, IBM Commerce, Israel |
|
| Back to top |
|
|
| olan022 |
| Posted: Fri Feb 07, 2003 10:21 am Post subject: |
|
|
Acolyte
Joined: 17 Apr 2002
Posts: 50
|
Here is the output I get from rcdmqimg
QM1: Needed for Media Recovery: S0000054.LOG
QM1: Needed for Qmgr Restart: S0000181.LOG
Looking at the logs, I see:
AMQ7468: The oldest log file required to perform media recovery of queue
manager QM1 is S0000054.LOG.
AMQ7467: The oldest log file required to start queue manager QM1 is
S0000181.LOG.
So the logs match up with what the command is outputting. Moreover, the rcdmqimg completes every time without any errors. This is odd. |
|
| Back to top |
|
|
| yaakovd |
| Posted: Fri Feb 07, 2003 12:04 pm Post subject: |
|
|
Partisan
Joined: 20 Jan 2003
Posts: 319
Location: Israel
|
This is good news  I think that your MQ is OK.
Is this log (S0000181.LOG) the oldest in your directory?
Do you have some big non-processed message in Q? Check also system Q-s. 480 MB logs it looks too mutch for "empty" QMgr.
_________________
Best regards.
Yaakov
SWG, IBM Commerce, Israel |
|
| Back to top |
|
|
| olan022 |
| Posted: Tue Feb 11, 2003 9:02 am Post subject: |
|
|
Acolyte
Joined: 17 Apr 2002
Posts: 50
|
S0000181.LOG is the newest.
Here is what IBM states:
------------
Abstract:
AMQ7468 message not incrementing when rcdmqimg run
Customer Symptoms:
AMQ7467 and AMQ7468 are output to the MQSeries error log whenever an action is taken which means that older
log files are no longer required for Queue manager restart and Media recovery respectively.
Using the record media image command (RCDMQIMG) to record the state of all objects on the queue manager would
be expected to move the media recovery log forward - when using MQSeries v5.2 this does not appear to occur.
Customer Solution:
The problem would appear to be caused by unused 'ghost' queues.
These unused ghost queues can be removed in one of 2 ways, so by way of a workaround customers can:
Delete and Redefine their model queues.
Stop the queue manager, set the ghostpool sizes to 0 (in the QMSTATUS.ini file) then restart the queue
manager.
Clearly the former of these has the least impact.
Technical Description:
For each model queue there is a pool of re-usable ghost queues. When a dynamic queue is created it requires some significant infrastructure to be put in place in terms of directories and files - at 5.2 instead of throwing away all of this data when a dynamic queue is created it is kept, and stored as a member of the ghostpool associated with that model queue.
It appears that the problem with rcdmqimg is related to the existence of ghost queues and the fact that whilst they are potentially MQSeries objects, whilst they are in the 'ghost' state, they are not real objects.
This is more than likely to do with the calculation of media restart LSN including these ghost queues when it doesn't need to - There is a development defect opened, 61889, to address this, in the short term, there are workarounds - see 'Solution' section.
---------
Option #1 did not work. Not happy about option #2, but during this time I have noticed that it WILL fire off and start compressing LOG files, it just could take over 2 months to do it! For now I am allocating more space and watching. I might try option #2 but would rather avoid. |
|
| Back to top |
|
|
|
|
