[Bioclusters] Application not writing to disk with OpenMosix
Joe Landman
bioclusters@bioinformatics.org
Fri, 20 Aug 2004 12:46:06 -0400
Hi Thomas:
T. Grzybowski wrote:
> Question: We are running an application called NONMEM on a small
> OpenMosix cluster. We use the "system" call under perl (together
> with "&") to spawn a few hundred tasks at a time. Each tasks is
> spawned with the present working directory set to a unique ID. The
> problem is that although the tasks seem to run OK (reading their input
> data running for a while and all), the output data files (in the PWD)
> are empty. I am not using the oMFS, but relying on the home node to
> handle all IO. Anyone have relevant experience with NONMEM or R who
> might be able to give me a clue to the problem?
Strace is your friend. From what you are saying, and without looking
at the code, I would guess that you might be trying to write to
directories that either do not exist, or have insufficient privileges to
write into, or are writing somewhere you do not expect. If you do not
allow the jobs to migrate, do they still work (albeit slowly, take a
tiny subset of say 10 jobs, let them run, and do they generate correct
output ...)
Another issue is that since you are using system, you can inspect the
result code in $?
From the perlfunc man page:
You can check all the failure possibilities by
inspecting $? like this:
if ($? == -1) {
print "failed to execute: $!\n";
}
elsif ($? & 127) {
printf "child died with signal %d, %s coredump\n",
($? & 127), ($? & 128) ? 'with' : 'without';
}
else {
printf "child exited with value %d\n", $? >> 8;
}
Either way, you want to look at the $! value and the $? code.
Joe
>
> Thanks,
> Thomas Grzybowski
>
>
> _______________________________________________
> Bioclusters maillist - Bioclusters@bioinformatics.org
> https://bioinformatics.org/mailman/listinfo/bioclusters
--
Joseph Landman, Ph.D
Scalable Informatics LLC,
email: landman@scalableinformatics.com
web : http://www.scalableinformatics.com
phone: +1 734 612 4615