Open MPI logo

Open MPI User's Mailing List Archives

  |   Home   |   Support   |   FAQ   |   all Open MPI User's mailing list

Subject: Re: [OMPI users] OpenMPI exits when subsequent tail -f in script is interrupted
From: Ralph Castain (rhc_at_[hidden])
Date: 2011-04-23 11:15:16


On Apr 23, 2011, at 9:07 AM, Pablo Lopez Rios wrote:

>> what about:
>> ( trap "" sigint; exec mpiexec ...)&
>
> Yup, that's included in the workarounds I tried. Tried again with your specific suggestion; no luck.
>
>> Well, maybe mpiexec is adjusting it on its own
>> again. This can be checked in /proc/<pid>/status
>
> The signal masks in /proc/$!/status are:
>
> nompi (bash):
> SigBlk: 0000000000010000 -> 16 blocked
> SigIgn: 0000000000000006 -> 1,2 ignored
> SigCgt: 0000000000010000 -> 16 caught
>
> mpi (mpirun):
> SigBlk: 0000000000000000 -> none blocked
> SigIgn: 0000000000000004 -> 2 ignored
> SigCgt: 0000000180015ee2 -> 1,5,6,7,9,10,11,12,14,16,31,32 caught
>
> I think I'm off by one in interpreting the above masks (for instance I would expect signals 30 and 31 to be caught, not 31 and 32), but I'm already assuming that the least significant bit is "signal 0"; assuming it is "signal 1" would just worsen the values.
>
> Anyway, why does mpirun bypass the traps I try to set and how do I stop it doing so?

You can't - this is a design requirement for clean termination of MPI jobs when the user interrupts execution.

>
> Thanks,
> Pablo
>
> On 23/04/11 13:20, Reuti wrote:
>> Hi,
>>
>> Am 23.04.2011 um 04:31 schrieb Pablo Lopez Rios:
>>
>>> I'm having a bit of a problem with wrapping mpirun in a script. The script needs to run an MPI job in the background and tail -f the output. Pressing Ctrl+C should stop tail -f, and the MPI job should continue. However mpirun seems to detect the SIGINT that was meant for tail, and kills the job immediately. I've tried workarounds involving nohup, disown, trap, subshells (including calling the script from within itself), etc, to no avail.
>>>
>>> The problem is that this doesn't happen if I run the command directly instead, without mpirun. Attached is a script that reproduces the problem. It runs a simple counting script in the background which takes 10 seconds to run, and tails the output. If called with "nompi" as first argument, it will simply run bash -c "$SCRIPT">& "$out"&, and with "mpi" it will do the same with 'mpirun -np 1' prepended. The output I get is:
>> what about:
>>
>> ( trap "" sigint; exec mpiexec ...)&
>>
>> i.e. replace the subshell with changed interrupt handling with the mpiexec. Well, maybe mpiexec is adjusting it on its own again. This can be checked in /proc/<pid>/status
>>
>> -- Reuti
>>
>>> $ ./ompi_bug.sh mpi
>>> mpi:
>>> 1
>>> 2
>>> 3
>>> 4
>>> ^C
>>> $ ./ompi_bug.sh nompi
>>> nompi:
>>> 1
>>> 2
>>> 3
>>> 4
>>> ^C
>>> $ cat output.*
>>> mpi:
>>> 1
>>> 2
>>> 3
>>> 4
>>> mpirun: killing job...
>>>
>>> --------------------------------------------------------------------------
>>> mpirun noticed that process rank 0 with PID 1222 on node pablomme exited on signal 0 (Unknown signal 0).
>>> --------------------------------------------------------------------------
>>> mpirun: clean termination accomplished
>>>
>>> nompi:
>>> 1
>>> 2
>>> 3
>>> 4
>>> 5
>>> 6
>>> 7
>>> 8
>>> 9
>>> 10
>>> Done
>>>
>>>
>>> This convinces me that there is something strange with OpenMPI, since I expect no difference in signal handling when running a simple command with or without mpirun in the middle.
>>>
>>> I've tried looking for options to change this behaviour, but I don't seem to find any. Is there one, preferably in the form of an environment variable? Or is this a bug?
>>>
>>> I'm using OpenMPI v1.4.3 as distributed with Ubuntu 11.04, and also v1.2.8 as distributed with OpenSUSE 11.3.
>>>
>>> Thanks,
>>> Pablo
>>> <ompi_bug.sh.gz>_______________________________________________
>>> users mailing list
>>> users_at_[hidden]
>>> http://www.open-mpi.org/mailman/listinfo.cgi/users
>>
>> _______________________________________________
>> users mailing list
>> users_at_[hidden]
>> http://www.open-mpi.org/mailman/listinfo.cgi/users
>
> _______________________________________________
> users mailing list
> users_at_[hidden]
> http://www.open-mpi.org/mailman/listinfo.cgi/users