Patrick Gleichmann wrote: > At the moment I have a script, which sends a kill signal to leave the > endless loop = end vdr. > The problem is that in runvdr I simply can't decide whether a normal > shutdown through my script happened or some other signal, e.g. watchdog, > kicked in, since all signals return "0". I'm going a different way: I've implemented signal handling in my runvdr itself. That way I can kill runvdr, and runvdr will kill vdr, and of course knows that vdr doesn't need to be restarted. Simplified, it works like this: SIG= trap "SIG=TERM" SIGTERM # run vdr eval "$VDRCMD" & VDRPID=$! # wait for VDR terminating or signal caught wait $VDRPID VDRRETURN=$? if [ "$SIG" = "TERM" ] ; then # runvdr was killed kill $VDRPID # dont restart else # normal termination, decide on errorlevel $VDRRETURN fi Cheers, Udo