| Author |
Message
|
| jgwheeler |
 Posted: Thu Sep 13, 2007 1:44 pm Post subject: MQ V6 Queue Statistics |
 |
|
Newbie
Joined: 21 Sep 2006
Posts: 5
|
In taking a look at the Queue Statistics messages generated with MQ V6, I have noticed a discrepency with the GetCount and GetBytes parameters. It appears that these parameters are not being populated properly in the event that a particular queue is open for input the entire duration of the Statistics Interval that is set. Looking at the corresponding PutCount and PutBytes parameters, I can see that messages are being sent to a queue, however the GetCount and GetBytes parameters do not reflect the fact that the messages have actually been successfully read. In cases in which a queue is not continuously open for input, the Get parameters are updated as expected.
Has anyone else noticed similar behavior? If not, are there any suggestions on what I might be missing?
Thanks |
|
| Back to top |
|
 |
| jefflowrey |
| Posted: Thu Sep 13, 2007 1:48 pm Post subject: |
|
|
Grand Poobah
Joined: 16 Oct 2002
Posts: 19981
|
Do you see the current depth decrease during the interval?
Maybe, even though the queue is open for input, and messages have been PUT, they haven't been committed or GOT.
_________________
I am *not* the model of the modern major general. |
|
| Back to top |
|
|
| jgwheeler |
| Posted: Thu Sep 13, 2007 2:00 pm Post subject: |
|
|
Newbie
Joined: 21 Sep 2006
Posts: 5
|
| The messages are being consumed as quickly as they are being written, so the current depth remains low, 0 or 1. Incidentally, I have also noticed that the corresponding QMaxDepth parameter is always one, so if messages were being read but not committed, I would expect that parameter value to be higher. Also, the ExpiredMsgCount and PurgeCount are set to zero and the LastGetTime matches the LastPutTime on the queue. |
|
| Back to top |
|
|
| jefflowrey |
| Posted: Thu Sep 13, 2007 2:15 pm Post subject: |
|
|
Grand Poobah
Joined: 16 Oct 2002
Posts: 19981
|
Are the messages non-persistent?
_________________
I am *not* the model of the modern major general. |
|
| Back to top |
|
|
| PeterPotkay |
| Posted: Thu Sep 13, 2007 5:34 pm Post subject: |
|
|
Poobah
Joined: 15 May 2001
Posts: 7716
|
I think Jeff is going down the same path I am.
Its a new "feature" in MQ 6.0. If a non persistent message is put on a queue that has an outstanding MQGET with wait with no syncpoint, the message is passed in memory to the getter, bypassing the logic that ups the Get Count.
A feature for performance, yes. A contact admin bug for accounting purposes. I opened a PMR. They said it will not be changed anytime soon, if ever.
We were scratching our heads the day after we upgraded our z/OS QM to 6.0 and saw several queues' GET counts drop to zero, when we knew there was 1000s of messages zooming thru every hour.
_________________
Peter Potkay
Keep Calm and MQ On |
|
| Back to top |
|
|
| jefflowrey |
| Posted: Thu Sep 13, 2007 7:50 pm Post subject: |
|
|
Grand Poobah
Joined: 16 Oct 2002
Posts: 19981
|
| PeterPotkay wrote: |
| I think Jeff is going down the same path I am. |
I'd say "following in your footsteps".
I was just guessing that this performance enhancement would prevent deq from being updated.
You've apparently confirmed it, and been told "working as designed".
_________________
I am *not* the model of the modern major general. |
|
| Back to top |
|
|
| markt |
| Posted: Thu Sep 13, 2007 7:52 pm Post subject: |
|
|
Knight
Joined: 14 May 2002
Posts: 504
|
| I think that Peter's issue was only for z/OS, while the stats messages are only on Distributed. (MQ on Distributed has had "put-to-waiting-getter" for more than one release, and it does it for more than just NP messages.) |
|
| Back to top |
|
|
| jgwheeler |
| Posted: Fri Sep 14, 2007 6:41 am Post subject: |
|
|
Newbie
Joined: 21 Sep 2006
Posts: 5
|
| PeterPotkay wrote: |
| Its a new "feature" in MQ 6.0. If a non persistent message is put on a queue that has an outstanding MQGET with wait with no syncpoint, the message is passed in memory to the getter, bypassing the logic that ups the Get Count. |
I don't believe that this is the case in my situation. There is another parameter with Queue Statistics called NonQueuedMsgCount. According to the manual, NonQueuedMsgCount represents, "The number of messages that bypassed the queue and were transferred directly to a waiting application." This value is also showing up as zero.
Looks like I will be opening a PMR. Certainly let me know if you have any other thoughts or ideas. I will keep you posted. |
|
| Back to top |
|
|
| KeeferG |
| Posted: Tue Sep 18, 2007 4:04 am Post subject: |
|
|
Master
Joined: 15 Oct 2004
Posts: 215
Location: Basingstoke, UK
|
I just wanted to make clear that Queue avoidance is not new and has been in the product for a very long time. It is referenced in several 5.3 performance reports.
_________________
Keith Guttridge
-----------------
Using MQ since 1995 |
|
| Back to top |
|
|
|
|
