Support for MONC?

1

Hi,

I’m a PhD student at Cambridge and I’m currently trying to work with MONC for cloud modelling. Ultimately, my aim is to model marine cloud brightening with it but so far I’m just trying to get it up and running. I’ve been getting help from Adrian Hill, who developed the model, but he no longer works at the Met Office so can only help me in his spare time. Is there anyone in your group who has worked with MONC and might be able to provide some support while I get to grips with it? Any help at all would be appreciated. I am currently still trying to build MONC and keep running into errors that I’m not sure how to solve.

Thank you,
William

2

Hi William,

I have access to some Cambridge systems. On what system are you trying to build MONC? What compilation errors are you seeing?

Simon.

3

Hi Simon,

Thanks for getting back to me. I’m currently just building it on my laptop, I’ll need to figure out what system I can get access to when it comes to running simmulations though. Currently this is the error I’m getting when I run the fcm make command (I think this is all the relevant lines):

[info] target-tree-analysis: elapsed-time=0.1s
[FAIL] mpif90 -obin/monc_driver.exe o/monc_driver.o -Llib -lmonc_driver -L/usr/lib -L/usr/lib/x86_64-linux-gnu/hdf5/mpich -lnetcdff -lnetcdf -lhdf5 -lhdf5_hl -lz -lpthread # rc=1
[FAIL] lto1: fatal error: bytecode stream in file ‘/usr/lib/libnetcdff.a’ generated with LTO version 11.2 instead of the expected 11.3
[FAIL] compilation terminated.

I’ve googled about it a bit but haven’t been able to find anything useful yet.

Kind regards,
William

4

Hi William,

I’m guessing that one of the software packages you installed (likely gcc, mpich or netcdf ) is from a different (Ubuntu?) release from the others. You need to ensure all software package versions are compatible. Unfortunately as this is on your laptop we are unable to offer support beyond general advice.
However, MONC is working on ARCHER2 and we can provide you with the working configs if you have access to the system.

Simon.

5

Just to add - if you are running on your laptop, there is a met office virtual machine with these things installed. Adrian Hill will know which version has been tested with MONC.

6

Hi Simon,

Do you happen to know who I would need to talk to at Cambridge to get access to ARCHER2? I think I will need it when I come to running MONC and if getting on it sooner will skip the slow process of building MONC on my laptop it might be a good idea to start doing it soon.

Thank you,
William

7

We can give you access - please follow Quickstart for users - ARCHER2 User Documentation for access.

Grenville

8

Hi Grenville,

Thanks for that. I have now set up a SAFE account. I don’t yet have a project code though, how could I set one up? My supervisor is Hugh Hunt in engineering. Will he need to set it up? I don’t think he’s worked with ARCHER2 before so he might not know how.

Thank you,
William

9

William

Please apply for access to n02 - I’ll add you to the n02_MONC subproject after that.

Grenville

10

Okay great, thank you. I have done that now.

11

Hi again,

I’ve been making progress with this but have run into a problem at the fcm-make stage.

When I do

fcm make -j4 -f fcm-make/monc-arc2-gnu.cfg
I get the error
[FAIL] /home1/home/n02/n02/wms30/MONC/vn0.8_quickstart/fcm-make/env-arc2.cfg:4: reference to undefined variable
[FAIL] $netcdf_path{?} =
[FAIL] undef($NETCDF_HOME)

I think this is telling me I need to tell it where the netcdf library is. I have it installed because when I enter “module list”, “cray-netcdf/4.8.1.1” is on there. I’m just not sure where that is.

Thank you,
William

12

Hi WIlliam

This worked for me ( I think I ran the gnu build) - maybe follow this to see where your setup differs -

Checkout:

fcm co https://code.metoffice.gov.uk/svn/monc/main/branches/dev/toddjones/r9891_consolidation_202210

…cd r9891_consolidation_202210

Build MONC:

cray: source env/archer_cray_mod ; fcm make -j4 -f fcm-make/monc-cray-cray.cfg -f fcm-make/casim_socrates.cfg --new

gnu: source env/archer_gnu_mod ; fcm make -j4 -f fcm-make/monc-cray-gnu.cfg -f fcm-make/casim_socrates.cfg --new

Grenville

13

Hi Grenville,

The second build with gnu works but the cray one does not. This is the start of the fail messages:

[info] target-tree-analysis: elapsed-time=0.3s
[FAIL] REALTYPE_RD.mod: target not found after an update:
[FAIL] /home1/home/n02/n02/wms30/MONC/r9891_consolidation_202210/build/include/REALTYPE_RD.mod: expect target file
[FAIL] compile+ 0.0 ! REALTYPE_RD.mod ← socrates/src/modules_core/realtype_rd.f90
[FAIL] VARIABLE_PRECISION.mod: target not found after an update:

Do I need to do both? What is the difference between the two commands? When I was running it in my branch I wasn’t writing anything before fcm make, should I have been?

If you have time, would you be available for a meeting to talk a bit about MONC, ARCHER2 and my project? There aren’t a lot of people who can support me in my group so I’m always looking for any external help I can get. I would really appreciate your time but I understand if you can’t.

Thank you,
William

14

William

Sorry for the slow reply – you don’t need two builds; either build with gnu or cray (I’ve not explored why the cray build failed.)

We are not in a position to support a science project as such - our role is limited to computational support.
I’ll be happy to meet to discuss this - I’ll send an invite.

Grenville

15

Thank you for organising that. As I’m new to ARCHER2 any help at all is appreciated.

Speak to you on Thursday,
William