| |
|
RSS Feed - WebSphere MQ Support
|
RSS Feed - Message Broker Support
|
  |
|
| How to know the broker's or an execution group's work load? |
« View previous topic :: View next topic » |
| Author |
Message
|
| karach |
 Posted: Thu Jan 05, 2006 2:24 am Post subject: How to know the broker's or an execution group's work load? |
 |
|
Newbie
Joined: 05 Jan 2006
Posts: 6
|
Hi all,
I'm a newby on this forum, and I'm just tasting WMQ products since a few months...
--------------------
I currently work on a system where the deployment can - possibly - take too much ressources.
Here is the problem : while I am re-deploying a flow, as this one is already in use (in a given EG), the program using it sometimes gets a time out.
This might be because of the fact that Websphere uses to stop a flow, then delete it and put the newly deployed one instead. If this same flow is under use (or is going to be so), then, the client program just waits for a response and if it doesn't get in time, sends a time out.
What I want to know then is if there are possibilities to measure the ressources taken by an execution group during deployment and use. And if so, with which tool or how inside the WBIMB Toolkit.
This will allow me to define a "step through deploy process" that will help avoiding an overuse of the Broker in the deploy time frame.
Yours,
Arash |
|
| Back to top |
|
 |
| tillywern |
| Posted: Thu Jan 05, 2006 12:40 pm Post subject: top |
|
|
Centurion
Joined: 28 Jan 2003
Posts: 109
Location: Colorado
|
Depends on the platform. Depends on OS.
Solaris: prstat
AIX: topas
linux: top
hpux: top
If you are on unix you can always run a "ps -ef |grep Data" to see how much cpu time has been burned by the execution group processes.
On Windows one might look at the task manager while doing a deploy. You might be able to use various unix lik emulators to actually get stats like you would from a unix box.
So your DAtaFlowEngines are you execution group processes. This is easiest to see when you start a broker. you can see the DFEs grab a bunch of CPU and while they start up. You could do the same with a deploy.
Take the system off load and run a script to get the information you want in a loop with time delay. Then you can get some measurements.
The variants of top are nice in that they show process size. But you can probably dump that with ps. Check the man page for ps. 'ps' woth the '-eo' option will allow you choose what aspects of the process informationyou what to display.
The short answer is.. "No easy way." O |
|
| Back to top |
|
|
| tillywern |
| Posted: Thu Jan 05, 2006 12:44 pm Post subject: One other thing. |
|
|
Centurion
Joined: 28 Jan 2003
Posts: 109
Location: Colorado
|
I have worked on systems where a deply took over 45 min.
Always consider your implementation architecture. If your broker database is not local to the broker (ie you traverse a network) deploy times can be affected negativly.
But start by checking CPU, memory, and swap usage. These will all give you a hit as to how to find the bottleneck in the process. |
|
| 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
|
|
|
|
