[torqueusers] How to run mpirun of intel on torque

Brock Palen brockp at umich.edu
Thu Dec 20 07:49:11 MST 2012


Stock mpiexec from OSC is broken with torque 4,  

/home/software/rhel6/mpiexec/04292011/bin/mpiexec uptime
<uptimes from all the nodes printed>
*** Hang ***

This broke our matlab PCT install, 

If you install your own build of hydra from mpich though, you can build it with tm support:

Then tell this new mpiexec.hydra to use pbs:

export HYDRA_LAUNCHER=pbs
export HYDRA_RMK=pbs
./mpiexec.hydra uptime
<print all nodes uptime>
*** no hang ***

Though it does print errors, these errors are related to torque 4 bug that causes OSC to hang.  

[mpiexec at nyx5354.engin.umich.edu] HYDT_bscd_pbs_wait_for_completion (./tools/bootstrap/external/pbs_wait.c:68): tm_poll(obit_event) failed with TM error 17002
[mpiexec at nyx5354.engin.umich.edu] HYDT_bsci_wait_for_completion (./tools/bootstrap/src/bsci_wait.c:23): launcher returned error waiting for completion
[mpiexec at nyx5354.engin.umich.edu] HYD_pmci_wait_for_completion (./pm/pmiserv/pmiserv_pmci.c:216): launcher returned error waiting for completion
[mpiexec at nyx5354.engin.umich.edu] main (./ui/mpich/mpiexec.c:325): process manager error waiting for completion

But it should work.  Just an alternative.  

Brock Palen
www.umich.edu/~brockp
CAEN Advanced Computing
brockp at umich.edu
(734)936-1985



On Dec 20, 2012, at 6:54 AM, Chris Samuel wrote:

> On Thu, 20 Dec 2012 10:11:58 AM David Roman wrote:
> 
>> Yes, i did this after my reply
>> 
>> Did this test
>> 
>> echo 'hpc-node15: hostname' | mpiexec --comm=none -nostdin -config=-
>> 
>> But I have a segmentation fault
> 
> Umm, yes, quite probably. :-)
> 
>> I read the documentation to find my mistake
> 
> You should just need to do:
> 
> mpiexec program arguments
> 
> replacing program and arguments with the executable and any arguments you need 
> to pass to it.
> 
> Hope that helps!
> Chris
> -- 
>   Christopher Samuel - Senior Systems Administrator
> VLSCI - Victorian Life Sciences Computation Initiative
> Email: samuel at unimelb.edu.au Phone: +61 (0)3 903 55545
>         http://www.vlsci.unimelb.edu.au/
> 
> _______________________________________________
> torqueusers mailing list
> torqueusers at supercluster.org
> http://www.supercluster.org/mailman/listinfo/torqueusers



More information about the torqueusers mailing list