Restart sometimes fails with "a server is already running"

I've just done a restart of a DM (with the init.d script, which calls nuxeoctl restart).
During the "stop", nuxeoctl got to the point where it force kills the server.
During the subsequent "start", it said there was already an instance of nuxeo running.

A manual start following this worked fine.

-> it seems in some cases, the "stop" can hand over control to the "start" before the server process is actually gone.