[maker-devel] Maker-Error when started with IMPI : CORRECTED MAIL : SEE THIS ONE

[Carson Holt]

This error —> slurmstepd: error: couldn't chdir to `/tmp/kn_pop235844/maker-job.uc1.11658244.170324_043356': No such file or directory: going to /tmp instead

It is from SLURM and not from MAKER. It occurs before your job even started. It’s from the SLURM initialization of one of the nodes you are using.

Note /tmp is not shared. It is independent on each node. So /tmp/kn_pop235844/maker-job.uc1.11658244.170324_043356 may exist on one node, but not on the others. Since you are somehow setting this before you launch the job, SLURM is complaining because it doesn’t exist on one of the other nodes during initialization. So you need to review how you are launching things.

—Carson


> On Mar 24, 2017, at 3:10 AM, Rainer Rutka <rainer.rutka at uni-konstanz.de> wrote:
> 
> HI!
> First of all thank your for previous help.
> Running Maker 2.31.9 with MPI (Intel) is running fine, if we
> use ONE node only.
> 
> But, if we try to concatenate more than one node (e.g. 2 node a´ 8
> cores) we get this error:
> 
> [...]
> ### Running Maker example
> 
> MOAB_PROCCOUNT: 16
> slurmstepd: error: couldn't chdir to `/tmp/kn_pop235844/maker-job.uc1.11658244.170324_043356': No such file or directory: going to /tmp instead
> STATUS: Parsing control files...
> Argument "ALRM" isn't numeric in exit at /pfs/data1/software_uc1/bwhpc/common/bio/maker/2.31.9/bin/../perl/lib/forks.pm line 2184.
> [...]
> 
> /tmp/kn_pop235844/maker-job.uc1.11658244.170324_043356
> was created before and is EXISTING during the period of the
> job continuance.
> 
> I attached the complete log to this e-mail.
> 
> Again: THANK YOU VERY MUCH.
> 
> All the best.
> 
> -- 
> Rainer Rutka
> Universität Konstanz
> Kommunikations-, Informations-, Medienzentrum (KIM)
> * KIM Ausbildung
> * Wissenschaftliches Rechnen/bwHPC-C5
> * KIM Basisdienste, KIM Support
> Raum: V511
> 78457 Konstanz
> +49 7531 88-5413
> <bwunicluster-maker-example.moab><maker-job.11658541>