| Author |
Message
|
| yoyo |
 Posted: Tue Feb 21, 2017 9:36 pm Post subject: |
 |
|
Novice
Joined: 19 Feb 2017
Posts: 10
|
|
| Back to top |
|
 |
| fjb_saper |
| Posted: Wed Feb 22, 2017 6:09 am Post subject: |
|
|
Grand High Poobah
Joined: 18 Nov 2003
Posts: 20756
Location: LI,NY
|
If you are using JMS you should not be dealing with buffer lengths at all. Just make sure your max message length on the queue is greater than MQMD + RFH2 + payload length...
Have fun 
_________________
MQ & Broker admin |
|
| Back to top |
|
|
| yoyo |
| Posted: Wed Feb 22, 2017 9:46 am Post subject: |
|
|
Novice
Joined: 19 Feb 2017
Posts: 10
|
Hi fjb_saper,
That does not solve the aim.
My aim was to verify if message property's length is counted as part of message length or not.
The docs state otherwise, and a simple test gives me a different result. |
|
| Back to top |
|
|
| fjb_saper |
| Posted: Wed Feb 22, 2017 2:15 pm Post subject: |
|
|
Grand High Poobah
Joined: 18 Nov 2003
Posts: 20756
Location: LI,NY
|
| yoyo wrote: |
Hi fjb_saper,
That does not solve the aim.
My aim was to verify if message property's length is counted as part of message length or not.
The docs state otherwise, and a simple test gives me a different result. |
The docs are correct. It is not counted as part of the message length per se. However when persisting the message you still need to account for it.
So when you use gmo/pmo option MESSAGE_PROPERTIES_IN_HANDLE or equivalent, the message length is really the payload length but you also have a properties length to account for....
Like I said it may be splitting hairs but the reality is that the total length needs to be accounted for when persisting.... (MQMD + properties (RFH2?) + payload )
_________________
MQ & Broker admin |
|
| Back to top |
|
|
| gbaddeley |
| Posted: Wed Feb 22, 2017 3:45 pm Post subject: |
|
|
Jedi Knight
Joined: 25 Mar 2003
Posts: 2538
Location: Melbourne, Australia
|
At times its frustrating that IBM included RFH / properties properties in the Message Data. A new part of a MQ message should have been invented. 1) MQMD, 2) RFH properties & possibly other common MQ structures, 3) Message Data = app data payload usually in string format.
Too late by a couple of decades....
_________________
Glenn |
|
| Back to top |
|
|
| yoyo |
| Posted: Fri Feb 24, 2017 10:35 am Post subject: |
|
|
Novice
Joined: 19 Feb 2017
Posts: 10
|
Hi gbaddeley,
Yup. That should've been the case in order to not count them in message's length.
Hi fjb_saper,
Yes it is splitting hairs! But I still don't see how the docs clearly convey that. I still feel they are incorrect.
So to conclude I think we can agree that yes they are accounted. |
|
| Back to top |
|
|
| fjb_saper |
| Posted: Fri Feb 24, 2017 2:10 pm Post subject: |
|
|
Grand High Poobah
Joined: 18 Nov 2003
Posts: 20756
Location: LI,NY
|
| yoyo wrote: |
Hi gbaddeley,
Yup. That should've been the case in order to not count them in message's length.
Hi fjb_saper,
Yes it is splitting hairs! But I still don't see how the docs clearly convey that. I still feel they are incorrect.
So to conclude I think we can agree that yes they are accounted. |
The docs convey the API. In the API the properties are not part of the message length. When serialized and stored on disk you have to count them
_________________
MQ & Broker admin |
|
| Back to top |
|
|
| rekarm01 |
| Posted: Fri Feb 24, 2017 7:15 pm Post subject: Re: Message properties wrongly included in message length |
|
|
Grand Master
Joined: 25 Jun 2008
Posts: 1415
|
What MQ version and platform is the application using? Whatever the issue may be, unless it's using MQ v7.0.0 on z/OS, it's probably not related to this APAR.
That's not quite what the doc says:
| Quote: |
It is possible for an application to successfully put a message with a buffer that is larger than the value of MaxMsgLength, if the buffer includes properties. ...
The MQRFH2 header fields add to the properties length only if ... every folder in the header contains properties.
If ... any folder does not contain properties, the MQRFH2 header fields count toward the message length instead. |
Does every folder in the MQRFH2 header contain properties, or are there folder(s) that don't contain properties?
| yoyo wrote: |
| But a simple hello world example gives a different outcome inconsistent with what the doc said. |
If so, then it's time for a PMR, to fix one or the other. |
|
| Back to top |
|
|
| yoyo |
| Posted: Tue Feb 28, 2017 6:50 am Post subject: |
|
|
Novice
Joined: 19 Feb 2017
Posts: 10
|
Hi rekarm01,
MQ version is 7.1
And platform is unix
Yes you are right, the APAR contains a different platform and version, but it is exactly the same issue.
Only the usr folder contains properties I believe.
How do I view these folders? |
|
| Back to top |
|
|
| rekarm01 |
| Posted: Tue Feb 28, 2017 4:16 pm Post subject: Re: Message properties wrongly included in message length |
|
|
Grand Master
Joined: 25 Jun 2008
Posts: 1415
|
| yoyo wrote: |
| Yes you are right, the APAR contains a different platform and version, but it is exactly the same issue. |
The Unix platform does not have a CSQMPOM module with a CSQMDMCM macro to calculate the message length, for which an available zSeries fix can be applied. It is not the same issue.
| yoyo wrote: |
Only the usr folder contains properties I believe.
How do I view these folders? |
There are plenty of tools available for browsing messages on a queue, such as IBM's sample program amqsbcg0, IH03 (rfhutil) supportpac, and various third party applications to choose from. |
|
| Back to top |
|
|
|
|
