Really basic BSHIST branch run question

[TriD]

Hi CESM forum!

I'm hoping this is an easy one, but ... I still don't know the answer.

I'm trying to run a really basic BHIST run (nothing fancy at all) branching off of 2000 (specifically: b.e21.BHIST.f09_g17.CMIP6-historical.004) in derecho. I've got the restart files for 2000-01-01, but notably not the case files (e.g. env_case.xml). Do you know where they live?

The command:
/glade/work/rtdatta/$src/cime/scripts/create_clone --case /glade/derecho/scratch/$USER/$caze --clone /glade/derecho/scratch/rtdatta/b.e21.BHIST.f09_g17.CMIP6-historical.004

The error:
ERROR: Command: '/usr/bin/xmllint --noout --schema /glade/work/rtdatta/my_cesm_sandbox/cime/config/xml_schemas/env_entry_id.xsd /glade/derecho/scratch/rtdatta/b.e21.BHIST.f09_g17.CMIP6-historical.004/env_case.xml' failed with error 'b"/glade/derecho/scratch/rtdatta/b.e21.BHIST.f09_g17.CMIP6-historical.004/env_case.xml:2: element file: Schemas validity error : Element 'file': Missing child element(s). Expected is ( group ).\n/glade/derecho/scratch/rtdatta/b.e21.BHIST.f09_g17.CMIP6-historical.004/env_case.xml fails to validate"' from dir '/glade/work/rtdatta/grids/BHIST_ocean'


Thanks very much for any thoughts.

R. Tri Datta

 

[peverley]

Hello!

My guess is you have your arguments swapped in the command. Which case are you trying to clone (e.g. the one that already exists)? That is the one that should be the argument to "--clone".

the command should be /glade/work/rtdatta/$src/cime/scripts/create_clone --case /path/to/new/case/directory --clone /path/to/existing/case/directory

Courtney

 

[TriD]

Hi Courtney,

Thanks for the response! Actually, that's exactly the order that I have.

My main problem is that I'm trying to clone a BHIST run and start it on derecho at year 2000. I have restart files in 2001-01-01 from one bhist CMIP6 run (b.e21.BHIST.f09_g17.CMIP6-historical.004), but the restart files (e.g. env_batch.xml) are missing, so I've ripped them off someone else's BHIST run (which uses. cesm2.1.0. ), so I am using the release-cesm2.1.0 codebase.

It looks like where it's failing to create the newcase is now probably a bit more to do with something not getting ported over when cheyenne/etc changed over. Is this something you'd recognize? This looks like cime_config files are deliberately coming from a common directory which no longer exists (for cesm2.1.0) ...

FileNotFoundError: [Errno 2] No such file or directory: '/gpfs/fs1/p/cesm/tutorial/release-cesm2.1.0-CU/cime_config/config_files.xml'

Is there a way to change this (and point to the right files for cesm2.1.0 without referencing a tutorial which no longer exists?

Thanks in advances!

R. Tri Datta

 

[TriD]

More generally, if there's a BHIST run with BOTH restart files ~2000 AND the initial setup files somewhere, this would be really helpful.
Are there BHIST runs which have been done on derecho yet? I htink the problems I'm seeing has to do with the major changes to computing in the last few months..

Thanks so much,

R. Tri Datta

 

[peverley]

Ah, ok. Yes, it looks like the case you've cobbled together doesn't have everything you need to clone.

I would suggest that, instead of cloning the case, to set up a branch run. I find this documentation helpful: Hybrid, Branch and Startup — CESM Tutorial

I believe you'll want to set $RUN_REFCASE to /glade/derecho/scratch/rtdatta/b.e21.BHIST.f09_g17.CMIP6-historical.004 (or /glade/campaign/cesm/cesmdata/inputdata/cesm2_init/b.e21.BHIST.f09_g17.CMIP6-historical.004 which I assume is the same) and then $RUN_REFDATE will be 2000-01-01

Hope that helps.

Courtney