[slurm-users] Execute scripts on suspend and cancel

[Michael Jennings]

On Thursday, 17 October 2019, at 16:50:29 (+0000),
Goetz, Patrick G wrote:

> Are applications even aware when they've been hit by a SIGSTP?  This
> idea of a license being released under these circumstances just
> seems very unlikely.

No, which is why SIGSTOP cannot be caught.  The action is carried out
by the kernel, not the application, and no further time slices will be
provided to the application until it receives the SIGCONT signal (at
which point its SIGCONT handler, if any, is invoked).

The parent, however, *can* take action when one of its children
receives SIGSTOP by setting up a SIGCHLD handler; see sigaction(2) for
the gory details, along with wait4(2)/waitpid(2).

Michael

-- 
Michael E. Jennings <mej at lanl.gov>
HPC Systems Team, Los Alamos National Laboratory
Bldg. 03-2327, Rm. 2341     W: +1 (505) 606-0605