noresmhub / cime Goto Github PK

Because the hostname of a Betzy login node does not match the machine name (betzy), certain scripts (e.g., compare_test_results) do not work on Betzy.

Building the model fails if the archiver is not called "ar" but for example "x86_64-cos6-linux-gnu-ar", as is the case when using conda cesm.

Currently, even if the user has loaded a valid Python python version (3.8 or higher), the build phase can fail because gmake can call back into the CIME CCS (./case.build) and it inherits the Python from the user's .bashrc .
Because Python 2.7 is still(!) the default (nearly four full years since it has been declared 'end of life'), that can cause the build to fail.

Would it be possible to identify the banches which we are not using and remove them from the NorESMhub/cime fork?

Fram has been updated which requires an update to its entry in the config_machines.xml file.

An error message in test_scheduler.py has a formatting issue so a test failure results in an exception rather than showing you the failure messages.

Port modifications that have been introduced to cime for NorESM 2.0 (zipping and long-term archiving) into the workflows in cime6 for the development version.

In the NorESM2 branch of CIME (cime5.6.10_cesm2_1_rel_06-Nor_v1.0.5), access to the develop queue was removed in 2021 (commit b4e36e7). Because the devel queue now works (as evidenced by its use in ccs_config_noresm).
The format of config_batch.xml is different, however, the directive element with the queue attribute might allow the correct flags to be passed (e.g., --qos=devel).

@mvertens @gold2718

To introduce bromoform into NorESM2.3, also CIME has to be modified. Currently NorESM2.3 refers in Externals.cfg for CIME to the tag cime5.6.10_NorESM2_3_r2.

Can I start from that tag cime5.6.10_NorESM2_3_r2 a branch called noresm2_3_develop?

@tto061 Hi, is N2dev_port_from_CESM2.2.0 a new pure CIME branch based on CESM2.2 (cime5.8.32) or is it a old branch on that new CIME code is merged? Because log message "adaptation of ESMCI/cime tag cime5.8.32 to NorESM2 science code showing "
956 changed files with 96,838 additions and 53,248 deletions. For me, it seems like as on old CIME tag we are having new CIME tag.

At line 2904 replace F0I (which corresponds to "(2A,I10)") by F0L (which corresponds to "(2A,L3)") because alb_cosz_avg is defined as a logical and not an integer

Some grids, e.g., f19_f19_mtn14, are allowed for any compset without an active ocean (not_compset="_POP|_BLOM").
Other grids, e.g., f09_f09_mtn14, are only allowed for compsets with a data ocean (compset="_DOCN").
The second set is then not usable for any compset with no ocean (compset="_SOCN").

The (suggested) solution is to change both types of restriction to be compset="_DOCN|_SOCN" which fulfills the original goal (not usable for compsets with an active ocean) but with minimal restrictions and no need to name the set of possible active ocean models.

Hi,
I get an error when I run the script
<NorESM_root>/cime/scripts/query_config --machine betzy

Traceback (most recent call last):
File "./cime/scripts/query_config", line 350, in
_main_func(doc)
File "./cime/scripts/query_config", line 346, in _main_func
query_machines(machine_name=args.machines)
File "./cime/scripts/query_config", line 47, in query_machines
machines.print_values(machine_name=machine_name)
File "./cime/scripts/query_config", line 318, in print_values
print(" {} : {} ".format(name, desc))
UnicodeEncodeError: 'ascii' codec can't encode character u'\xae' in position 23: ordinal not in range(128)

It seems that the script has trouble processing non-ascii characters.
$ grep -i betzy -A 1 cime/config/cesm/machines/config_machines.xml

BullSequana XH2000 AMD® Epyc™ "Rome" 2.2GHz, 128-way nodes, os is Linux, batch system is SLURM

Removing the "®" and "™" characters seem to fix the problem.

In order for a compset / resolution combination to not require the --run-unsupported option to create_newcase, each supported test category must be included in the 'magic' list inside CIME. Add (and possibly edit) this list to match NorESM needs.

This issue is related to the issue 205 in NorESM repository and issue 5 in CTSM repository.

The new CIME branch will be used to keep the version of CIME that works with the most recent version of CTSM and related model component. But whether it works with the atmosphere and ocean model components of NorESM is not tested.

The new CIME branch will also contain a lot of setup for running single-cell simulations over observation sites.

The configuration for running the model on different supercomputers (e.g., SAGA, FRAM) will be the same as the branch for running global NorESM simulation.

The new CIME branch is ready at my local repository. I can make a pull request if we decide on the name of the new branch.

What would be a proper name for the new CIME branch? @annefou, @kjetilaas, @sunnivin, @DirkOlivie, @matsbn, @MichaelSchulzMETNO

There are grids in config_grids.xml that are incompatible with BLOM but are not marked as such.

g17_g17
TL319_g17
T62_g17
f02_g17
f05_g17
f09_g17
f09_g17_gl4
f09_g17_gl20
f09_g17_gl5
f19_g17
f19_g17_r01
f19_g17_gl4
f19_g17_gl5
f19_g17_rx1
ne30_g17_rx1

Update minimum required version of Python to 3.8.
Note: All older versions of Python have been retired ("End of Life" in Python terminology).