Problem with restarting model

[EmilyVanDeKoot]

Hi, I am trying to do a two-year test run which restarts at the end of the first year. I am using CESM2.1 with the land component replaced by SLIM (Home · marysa/SimpleLand Wiki), a slab ocean and CAM5. I am running the model on the ARCHER2 supercomputer. The model successfully completes the first year but then fails to restart (note I am able to run much longer simulations if I don't restart the model).

From the cesm.log, I can see the model fails to restart because it is unable to find the last monthly output file of the first year (I output both monthly and daily data): "ERROR: GETFIL: FAILED to get two_years_updated_model.cam.h0.0001-12.nc". I have had a look, and this file exists in the directory "work/n02/n02/emilyvdk/cesm_data/archive/two_years_updated_model/atm/hist/" along with all the other output data. The restart file is in ".../cesm_data/runs/two_years_updated_model/run/" and is called "two_years_updated_model.cam.r.0002-01-01-00000.nc".

I have attached my cesm.log, env_run.xml, user_nl_cam, user_nl_clm, user_docn.streams.txt.som, submission script and .xml files in case they are helpful. I would be very grateful for any advice.

 

[erik]

The problem is that file needs to be copied to the run directory. The short term archiver should leave a copy in place, but something must not be working right in your port and it doesn't leave a copy around. So a simple work around would be to copy it back in by hand.

 

[EmilyVanDeKoot]

Hi Erik,

Thanks for getting back to me! I copied the files by hand and this allowed the model to restart. However I would like the model to be able to restart automatically as there is a limit on the run time for an individual job on the supercomputer I am using. Therefore I was wondering you know what might be wrong with my port?

Since posting I have changed some lines in my config_machines.xml file (shown below) but this has not resolved the issue. I have also made sure that these change are reflected in the env_run.xml file.

<CIME_OUTPUT_ROOT>/work/n02/n02/emilyvdk/cesm_data/scratch/$USER</CIME_OUTPUT_ROOT>
<DOUT_S_ROOT>/work/n02/n02/emilyvdk/cesm_data/scratch/$USER/archive/$CASE</DOUT_S_ROOT>

Many thanks,
Emily

 

[zhangmeixin]

hi！have you solved the probem？I meet the same error too...

 

[slevis]

In your case's env_run.xml, change DOUT_S_SAVE_INTERIM_RESTART_FILES to TRUE.

This will not hurt anything and may solve your problem. Let us know if it does.

 

[zhangmeixin]

I have set DOUT_S_SAVE_INTERIM_RESTART_FILES = TRUE, but it doesnt work.
The restart files are in the run directory, but the h0 files are moved to the /scratch/FCHIST/atm/hist directory and no copy to the run directory.
Sometimes this happens, but sometimes it doesn't ...Not sure why...
CAM history file from previous year missing from run directory have mentioned the same error, but I have set the path for netcdf.

 

[zhangmeixin]

I think I have solved the problem, actually, the reason is same as CAM history file from previous year missing from run directory.
must load the netcdf before running.