Main menu

Navigation

Using a restart file

19 posts / 0 new
Last post
jerodri3@...
Using a restart file

Greetings, 

I successfully built and ran a CESM version 1.2.0 B_2000_WACCM_CN model for one year, archived monthly. I also would like to make two other runs for 120 hours each, archived hourly. One of these will be run using a restart file, while the other one won't be. Will I need to change the CONTINUE_RUN and RESUBMIT variables for the env_run.xml file in order to use the restart file correctly? What would I need to change to create the run without the restart file? Thank you.

 

santos

CONTINUE_RUN is used to signify that a case has already been started, and you want to continue it. If you have a brand new case and you want it to start from the restart from a different case, you should set the RUN_TYPE to "branch", and set RUN_REFCASE and other "RUN_REF" variables appropriately. You may also have to move or link the restart files to your run directory by hand, if they are from your own run and not our inputdata.

The default values of STOP_OPTION and STOP_N for a new case will give you a 5 day run, which is 120 hours.

Assuming that you are primarily interested in CAM's output, the namelist options "nhtfreq" and "mfilt" are the main controls for the output frequency:

http://www.cesm.ucar.edu/models/cesm1.2/cesm/doc/modelnl/nl_cam.html

Sean Patrick Santos

CESM Software Engineering Group

jerodri3@...

When I have a job running on the background of our supported machine, it runs and produces the output properly until I logoff and check the output the following day. Even when I use the 'top' function to check for cesm.exe, it will not show up, making me feel the job is not running properly. What should I do in order to check for this? 

santos

Is there an error in any log, or does the model just stop? What script/command are you using to run the model?

Sean Patrick Santos

CESM Software Engineering Group

jerodri3@...

The model just stops. I'm using ./$CASE.run & command to run my case in the background. 

santos

This is not related to CESM. It's a normal attribute of UNIX systems to send a signal to close processes started by a terminal in this case:

http://en.wikipedia.org/wiki/SIGHUP#SIGHUP

The CESM scripts are designed to work with a machine's batch system. If you have a supported machine, you should run the "submit" script to submit a job for you, rather than running the "run" script directly.

If you use the run script from a terminal, you'll have to either leave the terminal window open or find a way to "daemonize" CESM. Bash has a "disown" command for this.

Sean Patrick Santos

CESM Software Engineering Group

jerodri3@...

Whenever I call the $CASE.submit script I receive an error message which states USERDEFINED_optional_run: Command not found; ERROR: cesm_submit problem sourcing templar. I'm not sure what this means.

santos

This means that the port to your machine is incomplete. Assuming that you are running this on a cluster with a batch control system, the BATCHSUBMIT variable needs to be set to use the batch system on that machine. Otherwise, you will have to find your own method to run the model in a way appropriate for that machine.

Sean Patrick Santos

CESM Software Engineering Group

jerodri3@...

When I'm prestaging the restart data into my new case's run directory, the hsi command on p. 58 of the User's Manual is unrecognizable when I type it in my system. Can I just manually move my files to the new run directory? 

santos

Yes, you can just copy it by hand. The hsi command is just an example of what you would do to get data from NCAR 's HPSS.

Sean Patrick Santos

CESM Software Engineering Group

jerodri3@...

Can I use the ./$CASE.run instead of ./$CASE.submit to run a branch/hybrid run? 

hannay

You can use $CASE.run to submit your job to the batch queue.

The command to use will depend on the computer:

for a Cray computer, you would use:

qsub $CASE.run

jerodri3@...

I successfully ran a case for one year archived monthly, and had the RESUBMIT value set to 9 (so the model could continue this startup run for another 10 years). How come I received this message below once the run finished? 

Tue Nov  5 11:11:04 CST 2013 -- CSM EXECUTION BEGINS HERE

Fri Nov 15 21:48:48 CST 2013 -- CSM EXECUTION HAS FINISHED

(seq_mct_drv): ===============       SUCCESSFUL TERMINATION OF CPL7-CCSM ===============

cp: No match.

RESUBMIT is now 9

USERDEFINED_optional_run: Command not found.

 

ccsm_postrun error: problem sourcing tempres

jerodri3@...

I also created a new case for a hybrid run, using a startup case from a previous run as a reference. When I tried to run that case, I get this error message: 

-------------------------------------------------------------------------

 CESM PRESTAGE SCRIPT STARTING

 - Case input data directory, DIN_LOC_ROOT, is /tng4/users/ldiao/CESM/data

 - Checking the existence of input datasets in DIN_LOC_ROOT

 

The following files were not found, this is informational only

Input Data List Files Found:

/tng1/users/jrodrigu/CESM/cesm1_2_0/scripts/11_18_hybrid/Buildconf/cam.input_data_list

/tng1/users/jrodrigu/CESM/cesm1_2_0/scripts/11_18_hybrid/Buildconf/cpl.input_data_list

/tng1/users/jrodrigu/CESM/cesm1_2_0/scripts/11_18_hybrid/Buildconf/pop2.input_data_list

/tng1/users/jrodrigu/CESM/cesm1_2_0/scripts/11_18_hybrid/Buildconf/cice.input_data_list

/tng1/users/jrodrigu/CESM/cesm1_2_0/scripts/11_18_hybrid/Buildconf/rtm.input_data_list

/tng1/users/jrodrigu/CESM/cesm1_2_0/scripts/11_18_hybrid/Buildconf/clm.input_data_list

File status unknown: B_2000_WACCM_CN_11_05_test_on_slave_node_n11.cam.i.0003-01-01-00000.

nc 

File status unknown: B_2000_WACCM_CN_11_05_test_on_slave_node_n11.rtm.r.0003-01-01-00000.

nc 

File status unknown: B_2000_WACCM_CN_11_05_test_on_slave_node_n11.clm2.r.0003-01-01-00000

.nc 

 

 CESM PRESTAGE SCRIPT HAS FINISHED SUCCESSFULLY

-------------------------------------------------------------------------

Mon Nov 18 12:03:02 CST 2013 -- CSM EXECUTION BEGINS HERE

Mon Nov 18 12:06:34 CST 2013 -- CSM EXECUTION HAS FINISHED

Model did not complete - see /tng1/users/jrodrigu/CESM/cesm1_2_0/exe_11_18/run/cesm.log.1

31118-120233

How come when I prestage those restart files that the model claims to be an "unkown file status" I get this error message? Attached is the cesm.log.131118-120233.
Attachment: 
jedwards
PGFIO-F-209/OPEN/unit=93/'OLD' specified for file which does not exist.
 File name = rpointer.ocn.ovf

CESM Software Engineer

jerodri3@...

When I set up a new case for a branch run, I received this error message:

 

-------------------------------------------------------------------------

 CESM BUILDNML SCRIPT STARTING

 - To prestage restarts, untar a restart.tar file into /tng1/users/jrodrigu/CESM/cesm1_2_0/ex

e_11_25_a/run

 infile is /tng1/users/jrodrigu/CESM/cesm1_2_0/scripts/11_25_a/Buildconf/cplconf/cesm_namelis

CAM writing dry deposition namelist to drv_flds_in 

CAM writing namelist to atm_in 

CLM configure done.

CLM adding use_case 2000_control defaults for var sim_year with val 2000 

CLM adding use_case 2000_control defaults for var sim_year_range with val constant 

CLM adding use_case 2000_control defaults for var stream_year_first_ndep with val 2000 

CLM adding use_case 2000_control defaults for var stream_year_last_ndep with val 2000 

CLM adding use_case 2000_control defaults for var use_case_desc with val Conditions to simula

te 2000 land-use 

CICE configure done.

POP2 build-namelist: ocn_grid is gx1v6 

POP2 build-namelist: ocn_tracer_modules are  iage 

ERROR(Build::Namelist::_parse_next): expect a variable instead got:  11_25.rtm.r.0001-01-03-0

0000.nc

 

ERROR: rtm.buildnml.csh failed

ERROR from preview namelist - EXITING

  I don't understand how the river runoff component for this case inhibits the model's ability to run. Would I need to edit a specific file in the rtm? Thank you. 
hannay

There is a problem with setting up a branch simulation when using RTM. The problem occurs when ./cesm_setup is invoked.

Try to replace:
cesm1_2_0/models/rof/rtm/bld/rtm.buildnml.csh

with the file attached:
rtm.buildnml.csh
(You need to unzip first: )

See also: https://bb.cgd.ucar.edu/bug-cesm120-when-using-branch-brnchretaincasenam...

Attachment: 
jerodri3@...

If I have a year's worth of data archived monthly and I want to create a plot using NCL to make a year's worth of that data, can I use a function to merge the monthly archived data into one file, so I can upload it to NCL and mark the months onto one axis? 

hannay

You can concatenate the files using the nco operator.

ncrcat allows to concatenate files along the record dimension, and store the results ina new file taht you can open with ncl.

For isnatnce,

 

Concatenate files85.nc,86.nc, … 89.ncalong the record dimension, and store the results in8589.nc:

ncrcat 85.nc 86.nc 87.nc 88.nc 89.nc 8589.nc

 

See more about ncrcat at:

http://nco.sourceforge.net/nco.html#ncrcat-netCDF-Record-Concatenator

Log in or register to post comments

Who's new

  • jwolff
  • tinna.gunnarsdo...
  • sarthak2235@...
  • eolivares@...
  • shubham.gandhi@...