Currently, if the engine run called here ends, the process hangs until it's killed externally.  A common pattern is to stop the software and let its caller (systemd, kubernetes, etc) handle a restart.  Providing an option to do so would be appreciated.