At least on our SLURM system, mpirun is not supported and you have to
use srun instead.
In siteconfig, there is even an option "ifort + slurm"; why not trying
to use these defaults ?
Am 29.06.2018 um 00:23 schrieb venkatesh chandragiri:
Dear Wien2k users,
I have forwarded the suggestions
[renwei@ln3%th2 ~]$ which mpirun
/usr/local/mpi3/bin/mpirun
*31 mpirun: Command not found.*
Similar to the "manpath: command not found" and "libmpi.so.12 => not
found" errors that you say are gone now, the mpirun seems to be
installed okay on one of the nodes but it seems like it is not
You will need to talk more to your sysadmin (maybe show this email).
The simple one is the line
*31 mpirun: Command not found.*
This means what it says -- mpirun is not in your PATH and/or some other
command is needed. This is being set in $WIENROOT/mpirun.
The other one
*32 setrlimit():
Dear Wien2k users,
I have forwarded the suggestions given by Prof. Gavin as well as Prof.
Marks to the cluster administrator and now it seems that those earlier
errors was rectified. However, there are still more errors coming out when
I am submitting my job into SLURM based queuing process and
The "ssh cn308 ldd $WIENROOT/lapw0_mpi" is finding files for your ifort
installation like
/THFS/opt/intel/composer_xe_2013_sp1.3.174/mkl/lib/intel64/libmkl_scalapack_lp64.so
just fine. So your environmental variables seem to be setup and working
fine on both nodes. It looks like the
Dear Prof. Marks,
I did "ssh othernode ldd $WIENROOT/lapw0_mpi".
=
[renwei@ln3 ~]$ ssh cn308 ldd $WIENROOT/lapw0_mpi
/THFS/opt/intel/composer_xe_2013_sp1.3.174/mkl/bin/mklvars.sh: line 118:
manpath: command not found
linux-vdso.so.1 => (0x7fffd8fff000)
cd $WIENROOT
edit parallele_options and setUSE_REMOTE and MPI_REMOTE to zero.
Then there is no ssh anymore. (But you can use only one node for k-parallel)
Regards
Am 16.06.2018 um 12:02 schrieb venkatesh chandragiri:
Dear Prof. Gavin,
I am using slurm based environment for running the
Dear Prof. Gavin,
I am using slurm based environment for running the jobs. I have attached
the typical script I made to submit the job. Although, I kept export &
source of LD_LIBRARY_PATH and path to the compilervars.sh, I have also
source them again by keeping them in separate "myenev" file.
Gavin has already answered in large part, but let me add a little. You have
two questions/issues which are probably unrelated:
***
1) Is my ssh working?
I suggest that you test this by itself, for instance by running "ssh
othernode" from your head node, and also execute some simple commands that
As pointed in my mail I kept all the variable paths to bashrc file as
well as in the jobscript file.
You didn't mention you were using a job script. I assumed you weren't.
That may be important. Some queue systems require a flag in the job
script to export the environmental variables
Dear Prof. Laurence Marks,
thanks for your reply. As pointed in my mail I kept all the variable paths
to bashrc file as well as in the jobscript file. Also I did the "ldd
lapw1c_mpi"
the output is ==
[renwei@ln3 ~/venky/soft/wien2k]$ ldd lapw1c_mpi
linux-vdso.so.1 =>
If I understand you email correctly, the lines
"/THFS/home/renwei/venky/soft/wien2k/lapw0_mpi: error while loading shared
libraries: libmpifort.so.12: cannot open shared object file: No such file
or directory
/THFS/home/renwei/venky/soft/wien2k/lapw0_mpi: error while loading shared
libraries:
Dear wien2k users,
I forgot to mention in my earlier mail that I have already have .ssh folder
in the server where wien2k installed and already copied the key in the
id_rsa.pub file to authorized_keys file . But I don't know why the second
error (when I used .machines file without lapw0 line)
13 matches
Mail list logo
