Re: When kill fails...
On Wed, 1 Dec 1999, Roland McGrath wrote:
> > Uhmmm....
> > I might have missed something in all UNIX - related manuals I have seen,
> > but...
> >
> > if
> > # kill -9 <some pid>
> > fails for a program (that has hung) , what should I do then...
This can happen on Solaris if you have hung NFS processes (and other
network problems). I have had it happen on Linux when there are SCSI
errors. Sometimes the only solution is to power-cycle the machine.
I think the reason is that there are locked kernel processes --- I don't
know whether these cases are "bugs" or "design errors". Someone on this
list surely knows more about this than I do.
> >
> > Does there exist any kommand, or call, UNIX specific, HURD specific, or Mach
> > specific - that completely and unfrendly removes a program from the
> > tasklist, and whipes it out of the virtual memory?
>
> That's what kill -9 should be doing. Unlike other signals, SIGKILL (9) is
> a special case and in fact does use the low-level Mach task_terminate call
> to nuke the process with extreme prejudice. Can you give us a specific
> reproducible case of an unkillable process?
>
> What does "fails" mean? Please show us the full context and the actual
> output, rather than your interpretation of what happened.
>
>
> --
> To UNSUBSCRIBE, email to debian-hurd-request@lists.debian.org
> with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
>
>
----
Guy W. Hulbert At Work:
guy@interlog.com guy@bioinfo.sickkids.on.ca
Reply to: