Interestingly the original version on of the remote agent controller was socket based and then upgraded to the named pipe approach ;-)
The most common way to deal with this scenario is like what has been done for WSDM and JMX based agents. Perhaps Mark or Balan can point to the source and plugin...
This basically has you implement your own agent infrastructure that is as sophisticated or simple as you need.
It become a question of how much process isolation you want/need from the agent infrastructure. Sri's proxy give the most isolation.
Thanks for your time.
--------------------------------------------------------------------------
Harm Sluiman, STSM,
phone:905-413-4032 fax: 4920
cell: 1-647-300-4758
mailto:[EMAIL PROTECTED]
Admin : Arlene Treanor [EMAIL PROTECTED] Tie: 969-2323 1-905-413-2323
| Navid Mehregani/Toronto/[EMAIL PROTECTED]
Sent by: [EMAIL PROTECTED] 07/11/2006 11:15 AM
|
|
Sorry Seungll for misunderstanding your original question. All agent and Agent Controller communication is done through named pipes. In order to have remote agents communicating with a remote Agent Controller, you'll need to replace the named pipe communication mechanism with sockets. I'm personally not aware of any public documentation that would help you to do this. Alternatively you can use the agent proxy approach that Sri has mentioned below.
Thanks,
Navid Mehregani
| "Doddapaneni, Srinivas
P" <[EMAIL PROTECTED]> Sent by: [EMAIL PROTECTED] 11/07/2006 10:54 AM
|
|
SeungIl Lee,
Navid seem to have missed the finer detail that you are trying to run agent controller and agent on different machines. TPTP agent controller manages and communicates with agents that are running on the same machine. For embedded targets for which running agent controller on the embedded target system is not possible, one needs to create a proxy agent to run on the same system as the agent controller. The proxy agent will handle the communication and control of the actual agent running in an embedded system.
This is not the best possible way to support the embedded systems use-cases, but can be implemented easily. For longer term, we are thinking on ways to enhance the data collection framework to allow for direction connection to agents both local and remote to Eclipse Workbench. My suggestion is to use the proxy agent approach.
Thanks,
-Sri Doddapaneni
TPTP Platform Project Lead
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Navid Mehregani
Sent: Tuesday, July 11, 2006 7:38 AM
To: TPTP Tracing and Profiling Tools Project developer discussions
Subject: Re: [tptp-tracing-profiling-tools-dev] Agent Controller and Agenton other machines..
Yes, of course this is possible. The Agent Controller was made for remote testing/profiling. Please visit our documentation page here: http://www.eclipse.org/tptp/home/documents/index.html and read the following documents:
- "Getting started with Agent Controller". This document explains how to set up and run the Agent Controller. Choose the appropriate version you're using
- "An Introduction in Profiling Java Applications". This document is a bit old, but it still does a good job of explaining some of the main concepts.
- "Demo: Profiling an application". You might also see some UI differences in this video and the latest TPTP driver.
- "Tips on speeding up TPTP's Java profiler". This document explains what you need to do to have an effective profiling session.
The most up-to-date documents are available in the help content of Eclipse so I also recommend skimming over those.
Remote profiling isn't very different from local profiling. All you have to do is set up and run the Agent Controller on the remote machine. Add the new host to your local machine, select it, and specify the location of the Java application you want to profile.
Navid Mehregani
| "SeungIl Lee"
<[EMAIL PROTECTED]> Sent by: [EMAIL PROTECTED] 11/07/2006 05:59 AM
|
|
I’m trying to modify Agent Controller to run on another machine from an agent.
Is it possible? I understand that normally Agent Controller and Agent should be run on the same machine.
But it is too heavy to run two java programs on my machine, (embedded system)
So I’d like to separate those two programs and make them run on separate machines.
(I mean Agent Controller and Agent communicate via socket or so.)
Please at least let me know where to start.
--
SeungIl Lee
_______________________________________________
tptp-tracing-profiling-tools-dev mailing list
tptp-tracing-profiling-tools-dev@eclipse.org
https://dev.eclipse.org/mailman/listinfo/tptp-tracing-profiling-tools-dev_______________________________________________
tptp-tracing-profiling-tools-dev mailing list
tptp-tracing-profiling-tools-dev@eclipse.org
https://dev.eclipse.org/mailman/listinfo/tptp-tracing-profiling-tools-dev
_______________________________________________
tptp-tracing-profiling-tools-dev mailing list
tptp-tracing-profiling-tools-dev@eclipse.org
https://dev.eclipse.org/mailman/listinfo/tptp-tracing-profiling-tools-dev
_______________________________________________ tptp-tracing-profiling-tools-dev mailing list tptp-tracing-profiling-tools-dev@eclipse.org https://dev.eclipse.org/mailman/listinfo/tptp-tracing-profiling-tools-dev
