If for some reason the process immediately exits with non-zero exit code (often due to some failure condition or misconfiguration), it will be immediately restarted in a tight loop. This can load to high load and filling log files when things go wrong. It might be nice to either have some moderate delay before retrying, or a limit on how many time it can be restarted before giving up. Either of these could be configurable. The backoff logic used in ready wait could perhaps be reused somehow?
If for some reason the process immediately exits with non-zero exit code (often due to some failure condition or misconfiguration), it will be immediately restarted in a tight loop. This can load to high load and filling log files when things go wrong. It might be nice to either have some moderate delay before retrying, or a limit on how many time it can be restarted before giving up. Either of these could be configurable. The backoff logic used in ready wait could perhaps be reused somehow?