Ice_read_nc_xy: Cannot find variable licefw_io

COSIMA
1

Hi Folks,

I am trying to setup a new 01deg_jra55_ryf for someone. I am following these quick start instructions:

The model runs fine from initial conditions. However, when I grab restarts from here:
/g/data/ik11/outputs/access-om2-01/01deg_jra55v13_ryf9091/restart995

the model crashes with:
ice_read_nc_xy: Cannot find variable licefw_io

MPI_ABORT was invoked on rank 4359 in communicator MPI_COMM_WORLD

with errorcode 1.

What am I doing wrong?

Thanks!

Paul
PS - I am not sure if this is an appropriate hive topic. Is COSIMA slack dead now?

2

Hi Paul,

If you’re using the latest master of GitHub - COSIMA/01deg_jra55_ryf: 0.1 degree ACCESS-OM2 experiment with JRA55 RYF atmospheric forcing., then the problem is probably the forcing versions - you’re trying to intialize a JRA55-do v1.4 from a JRA55-do v1.3 restart. If you want to use that restart, you’d want to use the v1.3 forcing, which I think is in this branch: GitHub - COSIMA/01deg_jra55_ryf at 01deg_jra55v13_ryf9091.

We have collected some information on standard COSIMA control runs on this forum: ACCESS-OM2 Control Runs

3

Thanks Ryan for the quick response :slight_smile:

I thought I was cloning GitHub - COSIMA/01deg_jra55_ryf at 01deg_jra55v13_ryf9091
but when I clone that branch e.g.:
git clone GitHub - COSIMA/01deg_jra55_ryf: 0.1 degree ACCESS-OM2 experiment with JRA55 RYF atmospheric forcing.

the config.yaml still has:

name: atmosphere
model: yatm
exe: /g/data/ik11/inputs/access-om2/bin/yatm_d750b4b.exe
input:
- /g/data/ik11/inputs/access-om2/input_20201102/yatm_01deg
- /g/data/ik11/inputs/JRA-55/RYF/v1-4

So … I guess, which is different from the config.yaml shown on the github page I am cloning from that shows:

name: atmosphere
model: yatm
exe: /short/public/access-om2/bin/yatm_b77aecd2.exe
input: /short/public/access-om2/input_08022019/yatm_01deg
ncpus: 1

How do I clone the v13 branch? Hmmm … feeling stupid :slight_smile:

Thank you,
Paul

4

Are their 01-RYF JRA v1.4 restarts anywhere?

I note also that the config.yaml for 01deg_jra55_ryf/config.yaml at 01deg_jra55v13_ryf9091 · COSIMA/01deg_jra55_ryf · GitHub

Has old links to raijin short:
name: atmosphere
model: yatm
exe: /short/public/access-om2/bin/yatm_b77aecd2.exe
input: /short/public/access-om2/input_08022019/yatm_01deg
ncpus: 1

5

FYI: I have working version of 01-RYF-JRAv13 here:
/home/561/pas561/access-om2/01deg_jra55_9091ryf

But I am struggling to get it going from restarts via github clones.

6

You clone the repository as a whole. If you want a branch other than the master one then you have to do something like

git clone --branch <branchname> <remote-repo-url>.

Alternatively, I usually clone the repository and then do a checkout with

git checkout -t origin/01deg_jra55v13_ryf9091

I can’t see any in /g/data/ik11/outputs/access-om2-01/

Yes that looks like a problem. I’ve asked @AndyHoggANU to push up the latest configurations for that simulation.

@mauricehuguenin perhaps you can help here? What configuration are you using for your 1/10th-degree RYF control run for the ENSO work?

This kind of issue is exactly why we want to put together the wiki at ACCESS-OM2 Control Experiments.

2 Likes
7

I think we’ve run into this exact issue before. I can’t find the github issue in which it’s discussed. However, clearly in this repo: Commits · rmholmes/01deg_jra55_ryf · GitHub, I’ve spent some time updating the configurations so that the RYF v1.3 01deg run can be continued on Gadi.

1 Like
8

Here’s a verison of the configs that should work on Gadi: GitHub - rmholmes/01deg_jra55_ryf at 08596d93ecb5296012c1cf5a4c58249c20ff1a8f. You could try that. Otherwise, wait for Andy to push up the latest version of the configuration files that he used for that run.

2 Likes
9

Not dead, but asking questions (and having them answered!) here helps anyone who might have the same problem in the future. The slack has time limited history and isn’t publicly searchable, so isn’t the best place for useful information to reside.

10

I am running the 01deg_jra55_ryf model with configurations I cloned from Matt Pudig.

Maybe you could clone my ENFull branch that works?

I just pushed this branch with the latest changes to my github repo again and should be good to go.

ENFull stands for my El Niño run with full forcing active in all variables.

I am branching off my perturbations from
/g/data/ik11/outputs/access-om2-01/01deg_jra55v13_ryf9091/output975 using JRA55-do-1-3.

I remember there were a few inconsistencies in the naming of the forcing files and naming of the variables inside the forcing files (e.g., psl vs. slp or prrn vs. rain) that Matt Pudig also fixed before I cloned his branch.

1 Like
11

Thanks @mauricehuguenin. I’m guessing Paul just wants your control simulation. Is that just GitHub - mauricehuguenin/01deg_jra55_ryf at Control?

12

Yes, that is the correct one. I just pushed it up again and the config files are now the latest ones.

13

BTW, if you want my original RYF config, it is now here:

1 Like