Applying UM branch for a HadGEM3 run?

Hi,

I’m running HadGEM3 on ARCHER2. I am trying to reproduce the output from suite u-ba937, which was run on ARCHER1. My suite u-ck619 uses Rose GUI setting fcm_make_um > env > sources > um_sources=‘branches/dev/mariavittoriaguarino/vn10.7_lig127k_CMIP6@48312’.
This branch was used before on ARCHER1 to run suite u-ba937, and includes very significant changes to the UM orbital parameters in module https://code.metoffice.gov.uk/trac/um/browser/main/branches/dev/mariavittoriaguarino/vn10.7_lig127k_CMIP6/src/atmosphere/radiation_control/orbprm.F90.
If the branch is used correctly, the output from u-ck619 should be very similar to the output from u-ba937.
To test this has worked, I have also run suite u-ck642, which is identical to u-ck619 but has the standard um_sources branch (the same um_sources as used in suite u-as037).
However, the output from u-ck619 is identical to the output from u-ck642, and very different from u-ba937 output. This suggests the branch isn’t used correctly to build the UM when I run u-ck619 on ARCHER2, but I am confused about where this could happen because u-ck619 file fcm_make_2/01/job.out does include the line export um_sources=‘branches/dev/mariavittoriaguarino/vn10.7_lig127k_CMIP6@48312’. What change should I make to make sure the branch is applied correctly?

Many thanks.

Best wishes,

Rachel

Rachel
Please allow us read permission on your ARCHER home and work directories

chmod -R g+rX /home/n02/n02/<your-username>
chmod -R g+rX /work/n02/n02/<your-username>

Grenville

Hi,

I’ve allowed you read permission now.

Rachel

Rachel

Please try again:

grenvill@ln04:~> cd /work/n02/n02/radiam24
-bash: cd: /work/n02/n02/radiam24: Permission denied

Grenville

Hi,

I’ve tried to run the command you suggested several times –

chmod -R g+rX /work/n02/n02/radiam24

I have also run:

chmod -R g+r /work/n02/n02/radiam24

and although both seem run without errors, several directories and files (including directories /work/n02/n02/radiam24 and /work/n02/n02/radiam24/cylc-run) still seem to have permissions drwxr-s— or drwxr-x— rather than drwxr-sr-x or drwxr-xr-x.

Do you have any idea what may be causing this problem?
Otherwise, is there a shared directory I can copy any useful files - e.g. /work/n02/n02/radiam24/cylc-run/u-ck619 and /work/n02/n02/radiam24/archive/u-ck619 - into so that you can answer the original question?

Many thanks,

Best wishes,

Rachel

Hi Rachel,

Can you try running without recursion:

chmod g+rX /work/n02/n02/radiam24 to just get the top-level directory? You shouldn’t need to worry about the last 3 --- on any of the files/directories, as we are trying to set the group permission bits not the world-readable bits.

Cheers,
Ros

Hi Ros,

Sorry I should have mentioned I did already try that (and tried it again just now) but permissions for /work/n02/n02/radiam24 are still drwxr-s—.

I think this is an issue with my laptop setup which I’m consulting IT about, so in the meantime is there a shared directory I can copy any useful files - e.g. /work/n02/n02/radiam24/cylc-run/u-ck619 and /work/n02/n02/radiam24/archive/u-ck619 - into so that you can answer the original question?

Thank you so much.

Best wishes,

Rachel

Hi Rachel

Try copying /work/n02/n02/radiam24/cylc-run/u-ck619 and u-ck642 to /work/n02/n02/shared

Grenville

Hi,

I’ve copied both of those into /work/n02/n02/shared, but both have retained their original permissions (and I can’t change this until chmod works on my laptop). You should definitely be able to access the u-ck619 files, but possibly not the u-ck642 files, apologies.
I hope this is enough to have a look at the original problem? If it isn’t I will let you know when I’ve sorted out read permission on my ARCHER2 home and work directories.

Many thanks.

Best wishes,

Rachel

Rachel

Are you sure you are on the right machine - you have moved u-ck619 u-ck642 to /work/n02/n02/shared on the old 4-cab system?
You need to be using login.archer2.ac.uk now.

Grenville

Hi Grenville,

I have now run both simulations on the ARCHER2 23-cabinet system, and am having the same issue. I have allowed you read permissions on /work/n02/n02/radiam24 and /home/n02/n02/radiam24. The suites I am using are u-ck768 (which should give similar output to u-ba937), and u-ck781 (which should NOT give similar output to u-ck768 or u-ba937).

I am trying to reproduce the output from suite u-ba937, which was run on ARCHER1. My suite u-ck768 uses Rose GUI setting fcm_make_um > env > sources > um_sources=‘branches/dev/mariavittoriaguarino/vn10.7_lig127k_CMIP6@48312’ .
This branch was used before on ARCHER1 to run suite u-ba937, and includes very significant changes to the UM orbital parameters in module https://code.metoffice.gov.uk/trac/um/browser/main/branches/dev/mariavittoriaguarino/vn10.7_lig127k_CMIP6/src/atmosphere/radiation_control/orbprm.F90.
If the branch is used correctly, the output from u-ck768 should be very similar to the output from u-ba937.
To test this has worked, I have also run suite u-ck781, which is identical to u-ck768 but has the standard um_sources branch (the same um_sources as used in suite u-as037).
However, the output from u-ck768 is identical to the output from u-ck781, and very different from u-ba937 output. This suggests the branch isn’t used correctly to build the UM when I run u-ck768 on ARCHER2, but I am confused about where this could happen because u-ck768 file fcm_make_2/01/job.out does include the line export um_sources=‘branches/dev/mariavittoriaguarino/vn10.7_lig127k_CMIP6@48312’ . What change should I make to make sure the branch is applied correctly?

Many thanks.

Best wishes,

Rachel

Rachel

u-ba973 has l_sec_var=.true.
u-ck768 has l_sec_var=.false.

A quick look at branches/dev/mariavittoriaguarino/vn10.7_lig127k_CMIP6@48312 seems to show that Maria’s change only happens if l_sec_var=.true.

Grenville

Thank you so much, I’ve changed it and that worked.

Best wishes,

Rachel