Test getting stopped for unknown reasons
We are doing load testing for 6500 users using 8 Vmware's Win xp Pro 2 GB ram for each VM.
When we start the test in all 8 VM's after some time opensta test failed in one of the VM.
Shows task group 1 failed in audit log.
Also in error log getting the following message
Exception getting VU count from TG Executer Could be TG stopped or completed.
1)We dont know why the test stops.
2)Also we found a strange behaviour i.e If I start the test in one VM the test starts automatically in another VM.Same thing happens when we stop the test.
Had anyone come across this situation
Re: Test getting stopped for unknown reasons
Couple of ideas on the task group stopping;
1) If all VMs on one physical server, try adding more physical servers and using fewer VMs per physical server.
2) Try having the master node not run a task group.
3) Try disabling http data collection. Set HTTP data users to 0
4) Monitor memory on the injector process and see if it is getting close to 1.5 GB in size. It usually quits shortly after that. Fast growth of the injector process working set can indicate a memory leak. Calling subroutines or scripts can cause a memory leak. Passing parameters makes it worse.
> 2)Also we found a strange behaviour i.e If I start the test in one VM the test starts automatically in another VM.Same thing happens when we stop the test.
The way this is supposed to work is you define a master node that "owns" the repository and you create "slave" nodes by configuring the name server on the "slaves" to point to the repository that resides on the master (do this by right mousing on the name server icon in systray and configure the name server specifying the node name of the "master". All opensta instances are a "master" until you configure the name server to use another node's repository.)
Once you do this, you have multiple instances of opensta ready to cooperate as a "cluster". Next, you assign task groups to individual nodes in the cluster using the OpenSTA commander. You start the test using commander on the master node and it instructs slave nodes to start their task groups according task group parameters.
Now, with this description, what exactly is your problem?