On Wed 2003-02-12 at 14:22:41 -0700, [EMAIL PROTECTED] wrote: [...] > what i'm doing: > ps ax | grep <process> > su > (passwd) > kill -s <signal> <pid> > > using all kinds of signals.... starting with sigterm, sigkill,
Those two is all you need. "kill -s TERM" will ask the process to
terminate (the process may refuse), "kill -s KILL" forcefully tries to
kill the process (.
If it doesn't terminate after a KILL signal, there is no way to do
it. Usually it means that your kernel got a hickup which shouldn't
normally happen. There are cases where the kernel cannot remove a
process due to its internal state[1], but I only encountered these with
either broken kernels (i.e. an update fixed it) or with broken
hardware.
> signals like sigrtmin+2 even were [and man kill does not say])
man 7 signal
(that's mentioned in "man kill")
> not wanting to get mucking around too much with something i didn't
> fully understand i didn't use them ALL but it seems like sigkill
> should kill just about anything :-/
Correct. If it does not, there is nothing a mere user (or admin) can
do about it.
HTH,
Benjamin.
[1] AFAIK, a killed process which currently happens to be somewhere in
the "kernel space", will not terminate until it jumped back to
user space. So one explanation for such an unkillable process is
that it got stuck in the kernel space somehow.
msg119277/pgp00000.pgp
Description: PGP signature
