[maker-devel] running maker on TACC Stampede.

[Carson Holt]

Stampede only has MVAPICH2. It  does not have OpenMPI (even though it has been requested several times). OpenFabrics libraries (used by MVAPICH2) have a known issue that restricts programs from making system calls while running under MPI.  A system call is when one program launches another (i.e. MAKER launching BLAST).  For this reason MAKER does not work with MVAPICH2.  It only works with OpenMPI.

You can still get it to work on MVAPICH2, but only on a single node.  If you request more than one node then it will fail.  The solution would be for TACC to install OpenMPI as an option on Stampede (like they have on Lonestar), but until that happens you can only run MAKER on a single node.

Thanks,
Carson


> On Jan 22, 2015, at 10:51 PM, Won C Yim <wyim at unr.edu> wrote:
> 
> Dear anyone whom may it concern,
> 
> Hi!
> 
> My name is Won Cheol Yim in University of Nevada, Reno.
> 
> I try to run MAKER on TACC Stampede.
> 
> It looks everything installed properly.
> 
> ==============================================================================
> STATUS MAKER v2.31.8
> ==============================================================================
> PERL Dependencies:
> VERIFIED
> External Programs:
> VERIFIED
> External C Libraries:
> VERIFIED
> MPI SUPPORT:
> ENABLED
> MWAS Web Interface:
> DISABLED
> MAKER PACKAGE:
> CONFIGURATION OK
> 
> And I installed Perl 5.18.4 with threads option.
> 
> But I try to run it with MPI, it generated error.
> 
> I assumed this problem came from ibrun in Stampede.
> 
> Is there anyway to run it on Stampede?
> 
> Here is my log.
> 
> TACC: Starting up job
> TACC: Setting up parallel environment for MVAPICH ssh-based mpirun.
> cat: /home1/02908/wyim/.sge/job..hostlist.kUm5vXw9: No such file or directory
> sort: open failed: /home1/02908/wyim/.sge/job..hostlist.kUm5vXw9: No such file or directory
> TACC: Setup complete. Running job script.
> TACC: starting parallel tasks...
> [c404-703.stampede.tacc.utexas.edu:mpirun_rsh][read_hostfile] Can't open hostfile `/home1/02908/wyim/.sge/job..hostlist.kUm5vXw9':  (2)
> TACC: MPI job exited with code: 1
> TACC: Shutting down parallel environment.
> TACC: Shutdown complete. Exiting.
> 
> 
> Regards,
> 
> Won
> -- 
> Yim, Won Cheol
> Sent with Airmail

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://yandell-lab.org/pipermail/maker-devel_yandell-lab.org/attachments/20150123/8314b69f/attachment-0002.html>