I've had a bit more time to analyse this one. I made the wrong assumptions about the returned value of start-stop-daemon, and consequently thought that the script is faulty. But in fact, if start-stop-daemon works as expected, the script looks alright to me.
What is not working properly is this start-stop-daemon inside the docker container. I found the corresponding issue already reported in docker (https://github.com/docker/docker/issues/6800), so this bug can be safely be closed. Regards Alberto
