Open MPI logo

Open MPI User's Mailing List Archives

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

Subject: [OMPI users] Proper use of sigaction in Open MPI?
From: Keller, Jesse (jesse.keller_at_[hidden])
Date: 2008-04-24 09:27:36


Hello, all -

 

I have an OpenMPI application that generates a file while it runs. No
big deal. However, I'd like to delete the partial file if the job is
aborted via a user signal. In a non-MPI application, I'd use sigaction
to intercept the SIGTERM and delete the open files there. I'd then call
the "old" signal handler. When I tried this with my OpenMPI program,
the signal was caught, the files deleted, the processes exited, but the
MPI exec command as a whole did not exit. This is the technique, by
the way, that was described in this IBM MPI document:

 

http://publib.boulder.ibm.com/infocenter/clresctr/vxrx/index.jsp?topic=/
com.ibm.cluster.pe.doc/pe_linux42/am106l0037.html

 

My question is, what is the "right" way to do this under OpenMPI? The
only way I got the thing to work was by resetting the sigaction to the
old handler and re-raising the signal. It seems to work, but I want to
know if I am going to get "bit" by this. Specifically, am I "closing"
MPI correctly by doing this?

 

I am running OpenMPI 1.2.5 under Fedora 8 on Linux in a x86_64
environment. My compiler is gcc 4.1.2. This behavior happens when all
processes are running on the same node using shared memory and between
nodes when using TCP transport. I don't have access to any other
transport.

 

Thanks for your help.

 

Jesse Keller

454 Life Sciences

 

Here's a code snippet to demonstrate what I'm talking about.

 

------------------------------------------------------------------------
----------------------------

 

struct sigaction sa_old_term; /* Global. */

 

void

SIGTERM_handler(int signal , siginfo_t * siginfo , void * a)

{

    UnlinkOpenedFiles(); /* Global function to delete partial files. */

    /* The commented code doesn't work. */

    //if (sa_old_term.sa_sigaction)

    //{

    // sa_old_term.sa_flags =SA_SIGINFO;

    // (*sa_old_term.sa_sigaction)(signal,siginfo,a);

    //}

    sigaction(SIGTERM, &sa_old_term,NULL);

    raise(signal);

}

 

int main( int argc, char * argv)

{

    MPI::Init(argc, argv);

    

    struct sigaction sa_term;

    sigemptyset(&sa_term.sa_mask);

    sa_term.sa_flags = SA_SIGINFO;

    sa_term.sa_sigaction = SIGTERM_handler;

    sigaction(SIGTERM, &sa_term, &sa_old_term);

 

   doSomeMPIComputation();

   MPI::Finalize();

   return 0;

}