[Wien] running k-point parallel across nodes

Laurence Marks laurence.marks at gmail.com
Thu Oct 17 02:08:05 CEST 2013


This means that it cannot find the relevant lapw1_1.def, perhaps because it
is working in the wrong directory (a guess). I suspect that you have
something still wrong with the environment.

One thing might be to switch to bash or tcsh to check. Otherwise it would
involve some hacking of lapw1para to try and identify what is going wrong.
If you can find someone who knows ksh well...

Sorry, not so helpful.

---------------------------
Professor Laurence Marks
Department of Materials Science and Engineering
Northwestern University
www.numis.northwestern.edu 1-847-491-3996
"Research is to see what everybody else has seen, and to think what nobody
else has thought"
Albert Szent-Gyorgi
On Oct 16, 2013 6:07 PM, "Oliver Albertini" <ora at georgetown.edu> wrote:

>  I enabled the path through ssh. In /etc/ssh/sshd_config:
>
>  added PermitUserEnvironment = yes
>
>  and added the file ~/.ssh/environment with
> PATH=/usr/bin/WIEN2k/13 in it. Now the following error appears on the
> remote node, the local node runs fine:
>
>
>  Error in LAPW1
>  'INILPW' - can't open definition file
> ^O������^@^@^@^@^@^@^@^A^O������`�^O���^M     ^@^P^@| R^P
>  'LAPW1' - INILPW aborted unsuccessfully.
>
>
>
> On Wed, Oct 16, 2013 at 1:19 PM, Laurence Marks <L-marks at northwestern.edu>wrote:
>
>> It may be that you need to add
>> export PATH=$WIENROOT:$PATH
>>
>> You are going in the right direction, but since I don't have ksh I can
>> only guess. Someone else may know (Europe perhaps asleep now).
>>
>> On Wed, Oct 16, 2013 at 3:15 PM, Oliver Albertini <ora at georgetown.edu>
>> wrote:
>> > userconfig does not do anything for ksh, besides print a to-do list:
>> >
>> > !!!  You are using neither csh, tcsh nor bashrc as your default   !!!
>> >      shell. You will have to configure your system by hand
>> >
>> >      *) add WIEN2k-directory to your seach-path
>> >      *) install aliases
>> >      *) set environment variable EDITOR to your prefered editor
>> >      *) set environment variable SCRATCH to your prefered scratch
>> >         path, to use the working directory set SCRATCH to "./"
>> >      *) set environment variable WIENROOT to your WIEN root directory
>> >      *) set environment variable W2WEB_CASE_BASEDIR to your WIEN cases
>> > directory
>> >      *) set the "stacksize" to unlimited
>> >      *) set LD_LIBRARY_PATH (if necessary)
>> >      *) if you will be using parallel execution using rsh: edit your
>> >         ~/.rhosts file such that rlogin logs you on to the
>> >         selected systems without asking for a password
>> >         using ssh: generate keys (ssh_keygen) and distribute public key
>> >         Eventually you may have to ask your system administrator
>> >
>> > In ksh, the startup script for log-in is .profile, where I had put these
>> > lines:
>> >
>> >
>> PATH=/usr/bin:/etc:/usr/sbin:/usr/ucb:$HOME/bin:/usr/bin/X11:/sbin:/usr/vacpp/bin:/usr/bin/WIEN2k/13:/home/oliver/scripts:.
>> > export WIEN_MPIRUN='poe _EXEC_ -procs _NP_'
>> > export WIENROOT=/usr/bin/WIEN2k/13
>> > export PATH
>> > export W2WEB_CASE_BASEDIR=/home/oliver/data/wiendir
>> > export SCRATCH=./
>> > export EDITOR=vim
>> > if [ -f $HOME/.kshrc -a -r $HOME/.kshrc ]; then
>> >         ENV=$HOME/.kshrc                # set ENV if there is an rc file
>> >         export ENV
>> > fi
>> >
>> > and in .kshrc (for all new shells, not just log-in) I also have the same
>> > variable settings, minus the if statement.
>> >
>> > But it seems that the non-interactive shell receives a different path:
>> >
>> > $ ssh stblade02 exec /usr/bin/ksh < echo_path.sh
>> >
>> /usr/bin:/etc:/usr/sbin:/usr/ucb:/usr/bin/X11:/sbin:/usr/java5/jre/bin:/usr/java5/bin
>> >
>> >
>> >
>> >
>> > On Wed, Oct 16, 2013 at 11:00 AM, Laurence Marks <
>> L-marks at northwestern.edu>
>> > wrote:
>> >>
>>   >> You have to make sure that the executables path is known on the
>> other
>> >> nodes, It looks like you are using ksh (I am not very familiar with
>> >> it) so you need to have appropriate lines in the initialization files
>> >> ( .kshrc ??? as a guess) to set this up. The script .userconfig does
>> >> this for bash/csh, not sure about ksh (never tried).
>> >>
>> >> On Wed, Oct 16, 2013 at 12:51 PM, Oliver Albertini <ora at georgetown.edu
>> >
>> >> wrote:
>> >> > Hello,
>> >> >
>> >> > To run k-point parallel across different nodes, is it enough to
>> simply
>> >> > have
>> >> > w2k installed on all the nodes along with pw-less ssh? I have pw-less
>> >> > ssh
>> >> > working among the nodes, but when I try to run another kpoint on
>> another
>> >> > node, the shell cannot find the executables:
>> >> >
>> >> > $ x lapw1 -p
>> >> > starting parallel lapw1 at Wed Oct 16 10:48:53 PDT 2013
>> >> > ->  starting parallel LAPW1 jobs at Wed Oct 16 10:48:53 PDT 2013
>> >> > running LAPW1 in parallel mode (using .machines)
>> >> > 2 number_of_parallel_jobs
>> >> > [1] 4653272
>> >> > [2] 3605022
>> >> > ksh: lapw1c:  not found.
>> >> > ksh: fixerror_lapw:  not found.
>> >> > ksh: /home/oliver/data/wiendir/benchmark/test_case:  not found.
>> >> > ksh: lapw1c:  not found.
>> >> > ksh: fixerror_lapw:  not found.
>> >> > [2]  - Done                 ( ( $remote $machine[$p]  ...
>> >> > [1]  + Done                 ( ( $remote $machine[$p]  ...
>> >> >      stblade01(1) 0.000u 0.000s 0.1 0.00%      0+0k 0+0io 0pf+0w
>> >> >      stblade02(1) 0.000u 0.000s 0 0.00%      0+0k 0+0io 0pf+0w
>> >> > test_case.scf1_1: A file or directory in the path name does not
>> exist.
>> >> >    Summary of lapw1para:
>> >> >    stblade01     k=1     user=0  wallclock=6
>> >> >    stblade02     k=1     user=0  wallclock=0
>> >> > 0.1u 0.1s 0:02 8% 0+0k 0+0io 0pf+0w
>> >> >
>> >> >
>> >> > .machines:
>> >> > 1:stblade01
>> >> > 1:stblade02
>> >> >
>> >> > Sincerely,
>> >> >
>> >> > Oliver
>> >> >
>> >> >
>> >>
>> >>
>> >>
>> >> --
>> >> Professor Laurence Marks
>> >> Department of Materials Science and Engineering
>> >> Northwestern University
>> >> www.numis.northwestern.edu 1-847-491-3996
>> >> "Research is to see what everybody else has seen, and to think what
>> >> nobody else has thought"
>> >> Albert Szent-Gyorgi
>> >> _______________________________________________
>> >> Wien mailing list
>> >> Wien at zeus.theochem.tuwien.ac.at
>> >> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>> >> SEARCH the MAILING-LIST at:
>> >> http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html
>> >
>> >
>>
>>
>>
>> --
>> Professor Laurence Marks
>> Department of Materials Science and Engineering
>> Northwestern University
>> www.numis.northwestern.edu 1-847-491-3996
>> "Research is to see what everybody else has seen, and to think what
>> nobody else has thought"
>> Albert Szent-Gyorgi
>> _______________________________________________
>> Wien mailing list
>> Wien at zeus.theochem.tuwien.ac.at
>> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>> SEARCH the MAILING-LIST at:
>> http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20131016/2c2bd734/attachment-0001.htm>


More information about the Wien mailing list