|
RSS Feed - WebSphere MQ Support
|
RSS Feed - Message Broker Support
|
 |
|
Interpreting MQ Trace files |
« View previous topic :: View next topic » |
Author |
Message
|
g4Prakash |
Posted: Tue Jun 29, 2004 5:02 am Post subject: Interpreting MQ Trace files |
|
|
 Apprentice
Joined: 17 Dec 2002 Posts: 25 Location: INDIA
|
Hi,
I have a synchronous application (ASP & COM Interop) sending & receiving (non-persistent) requests & replies (from MDB, EJB's)) messages using message-id & correl-id.
I want to find out the time consumed by MQ at each step ...MQCONN, MQOPEN, MQPUT, MQGET(WAIT), MQCLOSE(RQST & RPLY), and MQDISC.
Also I am interested in the time consumed by the inner layers of MQ ...the MCA, the channels, network transit time, etc ...
I enabled trace on my windows system as follows:
strmqtrc -t api -t comms -t parms and generated the trace files.
I viewed the files corresponding to the following processes:
AMQRMPPA,exe, inetinfo.exe, mmc.exe, and AMQPCSEA.EXE ....but I could not figure out how to extract the information that I want.
Could you please help me in this ... Is there a tool available that can be used to extract such information.
If required I could send across the trace files.
All help & guidance appreciated. Thanks. |
|
Back to top |
|
 |
JasonE |
Posted: Wed Jun 30, 2004 1:19 am Post subject: |
|
|
Grand Master
Joined: 03 Nov 2003 Posts: 1220 Location: Hursley
|
All that information is in the trace files - its just a case of understanding them
I'd start with working out api call times - You should have a trace for your app.exe. Note that if PID reuse occurs, then the traces concatenate, so not the pid of your test program, and the trace you are after is AMQ<pid>.trc.
If this is a server app, all the API's have trace points like "MQGET >>" and "MQGET <<", eg. for a windows 5.3 trace:
Quote: |
Counter Timestamp PID.TID Tracepoint
--------------------------------------------------------------------------
00010561 09:32:00.200255 852.1 !! - __________
00010562 09:32:00.200261 852.1 !! - MQPUT >>
00010563 09:32:00.200266 852.1 !! - Hconn:
|
For client apps its harder and you'll need to tie up sends/recvs.
As to the other areas you mention (consumed by the inner layers of MQ ...the MCA, the channels, network transit time, etc ...) they are just a case of finding the appropriate trace points. |
|
Back to top |
|
 |
g4Prakash |
Posted: Wed Jul 07, 2004 12:50 am Post subject: |
|
|
 Apprentice
Joined: 17 Dec 2002 Posts: 25 Location: INDIA
|
I am actually looking forward to some guidance in understanding the trace information. Is there any article/manual/redbook that can help me with this?
I am confused when I see MQGET >> and MQGET << and such others ....I am not sure which to refer to !!!
If it is not asking for too much I could send out the trace generated for 5 request messages sent across the network. The average round trip response time is around 400 ms
I used the command: strmqtrc -t api -t comms -t parms
to generate the trace.
thanks. |
|
Back to top |
|
 |
JasonE |
Posted: Wed Jul 07, 2004 1:29 am Post subject: |
|
|
Grand Master
Joined: 03 Nov 2003 Posts: 1220 Location: Hursley
|
Well, the trace has always been considered an internal tool, and hence there is no external documentation. I can talk you through some bits, depending on what you want to know.
MQGET >> is entry into the MQ libraries in the application workspace for the process/thread that tracepoint is for. MQGET << is the nearly the last tracepoint from the MQ libraries before we go back to the application.
PM a location (url / ftp site) I can get the traces from, so at least we are looking at the same set, and I'll try to give you some examples of what is going on. (I do have a tool which does some of what you are asking, but its internal only, as it does things we cant publish, unfortunately). |
|
Back to top |
|
 |
|
|
 |
|
Page 1 of 1 |
|
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum
|
|
|
|