I went back to this bug report where Ross noticed that very large smbd processes (for instance due to a memory leak such as ones we had in some versions) mais not have enough time to terminate before the init script ends.
As a consequence, on immediate "start", smbd will not start because a running smbd is detected: the one that was stopped earlier but hasn't stopped....yet. That smbd eventually terminates but, then, the machine is left without any running smbd. Of course, memory leaks are bad and, actually, we don't have any of these left, AFAIK. Still, could we improve the init script to better handle this? Should we use the "--retry" option of start-stop-daemon? If so, what argument should be give to it? --
signature.asc
Description: Digital signature
