Critical: brings down perf testbeds. Properly detect if T-rex is running with max number of pkill retries (with timeout between each try) prior new instance is started. If Jenkins Job is forced to stop then T-rex remains in running state changing its parent process and once new instance is being initialized, T-rex gets self into D state with no option of recovery. After TBx down due to this problem - best case we can log in to tg-host and troubleshoot TRex process (we are not able to kill process) and the only solution is to reboot the tg-host, worst case kernel messed up and we are not able to login in and we need to reboot the tg-host.