RPT with Agent Controller?
I got a problem with the RPT. I read about the functionallity, to work with 2 or more cross-linked RPT maschines. I think so you can start a test on one of the maschines, and the maschine ballanced througt an agent controller the vusers to both (or more) maschines(RPT installations).
For example: We have got 2 Windwos-systems with 2 full installed RPTs. We want to test a system with 600 vusers. So we want to start one of the RPT installations, and it should start the second RPT maschine itself and ballanced the vusers to 300 vusers per RPT maschine.
But I found no understandable Documentaion about:
- How to start and configure the agent Controller.
- How to config the RPT to work with the agent Controller.
- How to config the RPT/AgentC to work together and make them knowing on which maschines are the other RPT installations.
- How to start the main RPT Installation with use of agent controller
- How to start a test on the main maschine with the functionallity, that the main maschine split the vusers per agent controller to the other RPT maschines (RPT installations on different server).
Thanks in advanced!
Re: RPT with Agent Controller?
i'm not used to work with the agent contoller, but as far as i understand the readme, you are right. it's used to control and synchronize a test, which is executed by different testing machines. for example the result graphs should be merged over all testing machines.
unfortunately i didn't find a proper installation/configuration guide. perhaps someone else can give you more detailed information.
Re: RPT with Agent Controller?
It is a best practice to dedicate one machine to the user interface (workbench) for controlling the test. So, you may need three machines for best results: one for the workbench, and two to share 300 users each.
On Windows the agent controller is started automatically. Directory structures vary by RPT version number. On Windows, you should see an ACWinService.exe with a recent version of RPT (eg 7+). If you are not sure, run netstat -a. There should be something listening on port 10002. On Linux, you must run SetConfig.sh and start the controller manually (./RAStart.sh).
The installation of RPT generally happens using the Installation Manager program. On Windows, the agent controller is installed during this process. After that, the agent controller is only used for assisting in running tests.
To use RPT with the playback computers and the agent controllers running on them you must create a Performance Schedule (eg File > New > Performance Schedule).
There are tutorials available from the Welcome page (Help > Welcome) depending on your version.
Adding User Groups to the Performance Schedule is how you spread the users over machines. In the case you describe, look for the Locations area for the User Group. If the schedule has 600 users, add two locations for the two computers for playback. RPT will automatically split the 600 users over those two machines. It is always better to start with a few users and work up (eg 10, 100, etc.).
When you add a new location you need to specify four attributes:
name - any name you choose
hostname - the computer host name
deployment directory - where RPT will store assets (eg c:\deploy)
o/s - the computer o/s (Windows, Linux)