| Author |
Message
|
| JAKEZ |
 Posted: Wed Nov 03, 2004 8:46 am Post subject: Calculate the use of an mqs server for each client cxion? |
 |
|
Apprentice
Joined: 03 Oct 2002
Posts: 32
|
Hi,
- having an mqs server as a hub in front of a number of different mqs client we wonder how to measure the service of the server for each client cxion?
- has anybody already had this pb?
- some idea to treat this question?
thanks in advance for help
Jakez |
|
| Back to top |
|
 |
| csmith28 |
| Posted: Wed Nov 03, 2004 5:44 pm Post subject: |
|
|
Grand Master
Joined: 15 Jul 2003
Posts: 1196
Location: Arizona
|
Could you be more verbose on what you mean by,
| Quote: |
| "we wonder how to measure the service of the server for each client" |
It would also help if you tell us what OS/version, MQ version.
I understand that you have one MQManager with multiple remote Clients but you guestion is kinda vague.
_________________
Yes, I am an agent of Satan but my duties are largely ceremonial. |
|
| Back to top |
|
|
| JAKEZ |
| Posted: Wed Nov 03, 2004 11:38 pm Post subject: |
|
|
Apprentice
Joined: 03 Oct 2002
Posts: 32
|
Hi guy,
thanks fo replying - here is more information:
- we are running a Qmgr on an MQS server 5.3 on HPUX 11 box which is used as a hub for many MQS client hosts (about 400) coming there to put and get messages and treatment on them by applications
- we want to make "pay" each MQS client for the service they receive from this MQS hub and we wonder which could be the better way to do that:
+ counting the number of messages treated for each client on the server?
+ calculating the time passed by each client cxion?
and the different means are there to be able to do that?
and get feeling of people who have already done that kind of exercice?
regards,
Jakez |
|
| Back to top |
|
|
| Nigelg |
| Posted: Thu Nov 04, 2004 1:24 am Post subject: |
|
|
Grand Master
Joined: 02 Aug 2004
Posts: 1046
|
| You could use either method of accounting for client time. There are no readily available statistics on client connections. The best way would be to write a receive exit to run on the SVRCONN channel. You can identify the client by the IP address in the MQCD passed to the exit, and the type of client API call by byte 10 of the data flow from the client - there is a table of exposed values of this byte in the Intercomms manual. |
|
| Back to top |
|
|
| JAKEZ |
| Posted: Thu Nov 04, 2004 1:39 am Post subject: |
|
|
Apprentice
Joined: 03 Oct 2002
Posts: 32
|
thanks Nigelg,
- i think that a receive exit on svrconn is a good idea
- but i thaught that it was not possible to pass the MQMD of the message to the receive exit?
cheers
Jakez |
|
| Back to top |
|
|
| Nigelg |
| Posted: Thu Nov 04, 2004 2:34 am Post subject: |
|
|
Grand Master
Joined: 02 Aug 2004
Posts: 1046
|
IBM do not expose the meaning of the data structures passed in client data flows other than byte 10 of the TSH, so strictly speaking you cannot parse the msg out of the data.
However, you can tell what type of data flow it is, e.g. x86 is an MQPUT request, x85 is an MQGET request, and so on. See table 53 in the Intercomms manual at
http://www.mqseries.net/manuals/intercommunication/csqzae0604.htm#FT_TBLDUP0007
It should not be too hard for you to find the size of the dataflow or the msg by inspecting the data in an MQPUT request, x86, or an MQGET reply, x95, and to find the completion code & reason code in an MQGET reply. |
|
| Back to top |
|
|
|
|
