In my HPC SOA jobs based project I have to perform negative test cases where in if a Task fails, I have to confirm if the Job scheduler would retry the tasks. I believe the tasks are rerun as many times as set in the Job Scheduler Configuration
-> Error Handling section. I've tried taking the compute nodes offline while the job is being run, tried thowing exception from the WCF Service but I'm unable to simulate the scenario.
Kindly suggest the ways to force Service Tasks retries.
I think the easiest way is to kill task process which is HpcServiceHost.exe on the compute node, or in the service code to exit the process, eg: calling Envrionment.Exit. Scheduler would recreate the task in this case.