Back to UEMS. Old problems

Questions and other topics related to UEMS 18.
NoNickName
Posts: 37
Joined: Thu Jan 06, 2011 9:10 pm

Back to UEMS. Old problems

Post by NoNickName » Wed Jul 04, 2018 10:20 am

Hi, everybody. i'm in the process of setting up a new workstation, and decided to start from scratch, with the newest UEMS release.
New machine, new Zorin OS, new installation, new domain. Old problem.
After the definition of the domain, localisation fails because unable to find geogrid.log .
Same problem it always had since dwiz was released. E.g. viewtopic.php?f=12&t=1195&p=5663&hilit= ... tion#p5663

Anyone has got a solution or workaround?

meteoadriatic
Posts: 1567
Joined: Wed Aug 19, 2009 10:05 am

Re: Back to UEMS. Old problems

Post by meteoadriatic » Wed Jul 04, 2018 10:54 am

Workaround is probably using procedure described in Chapter 5.4:
http://strc.comet.ucar.edu/software/uem ... pter05.pdf

NoNickName
Posts: 37
Joined: Thu Jan 06, 2011 9:10 pm

Re: Back to UEMS. Old problems

Post by NoNickName » Wed Jul 04, 2018 12:04 pm

Gotcha, but why does dwiz always and consistently fails on me?
Anyone is experiencing the same problem?

andres
Posts: 19
Joined: Wed Apr 25, 2018 12:34 pm
Location: Chile

Re: Back to UEMS. Old problems

Post by andres » Wed Jul 04, 2018 1:12 pm

The problem seems that it doesn't find the GEOGRID:TBL file. Is not copied automatically.

Copy the file to the directory or make sure is correctly linked in

/<your path>/uems/data/tables/wps/
---
Geophysics Department
University of Concepcion, Chile

NoNickName
Posts: 37
Joined: Thu Jan 06, 2011 9:10 pm

Re: Back to UEMS. Old problems

Post by NoNickName » Wed Jul 04, 2018 5:23 pm

Doesn't seem to be the case. See image:

Image

smartie
Posts: 97
Joined: Sat May 21, 2011 7:34 am

Re: Back to UEMS. Old problems

Post by smartie » Wed Jul 04, 2018 7:15 pm

What is in

Code: Select all

/uems/logs/domain_wizard.log
If there is an error of this nature

Code: Select all

mpiexec.gf: relocation error: /lib/libnss_files.so.2: symbol __rawmemchr, version GLIBC_2.2.5 not defined in file libc.so.6 with link time reference
see here
viewtopic.php?f=4&t=215&start=10
I believe Zorin is a Ubuntu based system and that may be the source of the issue.

NoNickName
Posts: 37
Joined: Thu Jan 06, 2011 9:10 pm

Re: Back to UEMS. Old problems

Post by NoNickName » Thu Jul 05, 2018 8:53 am

Replying to both:
1) symlinking GEOGRID.TBL TO GEOGRID.TBL.ARW doesn't make any difference whatsoever.
2) this is what I got in domain_wizard.log

Image

I suspect $EMS_UTIL is not exported properly...

smartie
Posts: 97
Joined: Sat May 21, 2011 7:34 am

Re: Back to UEMS. Old problems

Post by smartie » Thu Jul 05, 2018 10:04 am

No wonder you have problems!
You are using the wrong shell- install csh or tcsh and use that.

NoNickName
Posts: 37
Joined: Thu Jan 06, 2011 9:10 pm

Re: Back to UEMS. Old problems

Post by NoNickName » Thu Jul 05, 2018 5:00 pm

Makes no difference whatsoever what shell I'm using, but I'm making progress.

Code: Select all

ems_domain.pl exit code is: 0
Local
INFO: ( $EMS_UTIL/mpich2/bin/mpiexec.gf -n $OMP_NUM_THREADS /home/alex/uems/bin/geogrid >& /home/alex/uems/runs/lombardia3/static/geogrid.log ; echo $? > /home/alex/uems/runs/lombardia3/static/geogrid.log.exitCode) &
Local
Local
geogrid.log still not found, so I did from csh:

Code: Select all

% $EMS_UTIL/mpich2/bin/mpiexec.gf -n $OMP_NUM_THREADS /home/alex/uems/bin/geogrid >& /home/alex/uems/runs/lombardia3/static/geogrid.log 
Now geogrid.log is where it's supposed to be in ./static, but it contains:

Code: Select all

ERROR: Error opening file namelist.wps
ERROR: Error opening file namelist.wps
ERROR: Error opening file namelist.wps
ERROR: Error opening file namelist.wps
application called MPI_Abort(MPI_COMM_WORLD, 127) - process 3
application called MPI_Abort(MPI_COMM_WORLD, 127) - process 0
I've got namelist.wps in ./static together with geogrid.log
What's next?

smartie
Posts: 97
Joined: Sat May 21, 2011 7:34 am

Re: Back to UEMS. Old problems

Post by smartie » Thu Jul 05, 2018 5:28 pm

There is very little to go on here... except the scripts are not finding files which are present (I presume). The WRF and WRF-EMS documentation specifically states that the software must be run in a csh/tcsh environment. WRF-EMS docs also warn about library issues with Ubuntu, therefore do you want to contnue to struggle against the stream? I would advise working in Fedora/Centos (or Debian/SuSe-that works fine) and do a clean install with csh/tcsh.

Post Reply