probelm with running cesm 1.2

[jjakacki@iopan_gda_pl]

Hello,i am on the way of porting cesm 1.2 to the new machine. I have working version, but this is the old one (cesm 1.0.1) on this computer. The problem is that there is error message after starting the job:/users/tmp/cases/NM_B_1850$ more NM_B_1850.o1280025
Warning: no access to tty (Błędny deskryptor pliku).
Thus no job control in this shell.
-------------------------------------------------------------------------
 CESM BUILDNML SCRIPT STARTING
 - To prestage restarts, untar a restart.tar file into /users/tmp/jakacki/cases/NM_B_1850/run
Word too long.
ERROR: cpl.buildnml.csh failed
ERROR from preview namelist - EXITING
/users/tmp/cases/NM_B_1850$ my question is - how to diagnose problem with cpl.buildnml.csh script. I have added some lines inside, but it looks like it does not work. I work on Linux IBM cluster with intel compiler.thank you in advance for any help,regards,jaromir

 

[jedwards]

Try adding the -x option to the cpl.buildnml.csh script - so that the first line of the script is:

#! /bin/csh -fx<br /><br />this will cause the script to echo each line as it is executed and may give you a better idea of what is going on.

  

 

[jjakacki@iopan_gda_pl]

Thank you for the quick response.I added 'x' option to the cpl.buildnml.csh script and it gave me more output. But this output is not enaugh for me so i will be very happy if somebody will give me advice how to make next step. The error message is:setenv MKL_TARGET_ARCH
setenv MKL_LP64_ILP64
setenv MKL_MOD
setenv MKL_VERBOSE
if ( intel64 ==  ) then
while ( intel64 !=  )
if ( intel64 == ia32 ) then
if ( intel64 == intel64 ) then
setenv MKL_TARGET_ARCH intel64
else if ( intel64 == mic ) then
shift
end
while (  !=  )
if ( intel64 ==  ) then
if ( 1 ) then
Word too long.
ERROR: cpl.buildnml.csh failed
ERROR from preview namelist - EXITINGthank you in advance for any help,regards,jaromir

 

[jedwards]

I don't know where this code came from, there is nothing like this in cpl.buildnml.csh.  

 

[santos]

"Word too long" is usually due to an environment variable like PATH being too long for tcsh. I think the limit is 1024 bytes. This is probably not coming from the CESM scripts themselves, but rather the variables in the env_mach_specific script for your machine. At a glance, it looks like env_mach_specific is sourcing an Intel login script.  One of our machines has a similar script from Intel, and I've noticed that it creates an extremely long PATH, which gets longer every time the script is sourced. So I would recommend that you check the length of PATH and other environment variables in env_mach_specific, and maybe edit that file to not source the Intel script if the environment has already been set. 

 

[jjakacki@iopan_gda_pl]

Thank you Sean, It is exactly place which helped me. I commented out line source ~/.cshrc and it has wanted to run. But if i removed this line, i was not able to compile the model (there is configuration for compiler). So what should i change to have version which will be compiled and working without any changes. I think i did something wrong when i tried to adapt the model for this machine (intel cluster). But i am aware that i do not have enough experience and knowledge for such work. I will appreciate if somebody will give me advice what could be wrong in my configuration.  I can temporary compile with this line and run without it. But i have another problem. In my old version if there were not data in the input data directory the model tried to download it from public server (i believe for UCAR or NCAR). But now, when the job started the model crashed because of missing ice file: forrtl: No such file or directory
forrtl: severe (29): file not found, unit 82, file /users/tmp/cesm_input_data/ice/cice/iced.0001-01-01.gx3v7_20080212   Is there any flag which permits to download all needed files? Or should i do it myself?  Thank you in advance for any help and best regards, jaromir   

 

[jedwards]

I think that ice file is a known problem in the script that retreves input data, you should do it by hand.   The .cshrc issue is really a system problem, talk to your system administrators about fixing it or you could try changing your default shell.