Various fixes needed to improve installation experience

[nichannah]

Thanks to @AndyHoggANU and others for these.

• the last line in .gitmodules should be a https:// address, otherwise it won’t work for gumbies like myself & Kial who don’t have ssh keys set up. I put an issue on this one in github.
• get_input_data.py doesn’t work without sh, which has to be installed by the user, and requires python 2.7.11 … would help if that was more generic.
• making worked well, but we would suggest putting the pytest instructions first, and making it more explicit that there are two ways of doing this.
• the suggest commands to copy the executables should be “cp” not “ls”. Also, include optional commands for the other resolutions of cice.

[adele-morrison]

I was able to follow the installation instructions without issue (seemed to compile fine, still waiting to run). The only thing I noted to change was that the path given in the 'Run' instructions should be:
cd $ACCESS_OM_DIR/control/1deg_jra55_ryf/

[adele-morrison]

Also, it would have been useful to know in the 'Run' instructions that access-om2 requires payu/dev, not payu/0.7.

And the naming instructions for the cice executable are for the 1deg model. For the 025 model it would be useful to know that it needs to be: bin/cice_auscom_1440x1080_480p_fa5b0bcf.exe

[marshallward]

I can do a payu 0.8 system update today if that helps. AFAIK payu/dev can only be accessed inside v45.

[nichannah]

• Better documentation about getting the CICE resolution right.

[nichannah]

• It would be good to automate moving/renaming the executables and updating the config.yaml with the correct name. This step would need to:
  1. check the repo HEAD hash
  2. check a hash stored in the exe
  3. do the copy to the right filename (based on the above hashes)
  4. update the config.yaml

[FanghuaWu]

Following the installation instructions, I have already run the model. However, when collating the output files, I got the following error message:

Traceback (most recent call last):
File "/jobfs/local/pbs/mom_priv/jobs/7765834.r-man2.SC", line 9, in
collate_cmd.runscript()
File "/projects/v45/apps/payu/dev/lib/payu/subcommands/collate_cmd.py", line 98, in runscript
expt.collate()
File "/projects/v45/apps/payu/dev/lib/payu/experiment.py", line 625, in collate
model.collate()
File "/projects/v45/apps/payu/dev/lib/payu/models/fms.py", line 85, in collate
assert mppnc_path
AssertionError

According to the suggestion from Aidan, cp mppnccombine to $ACCESS_OM_DIR/bin/ and then works.

Do I miss something? Or where to copy mppnccombine file should be mentioned in the instructions.

[AndyHoggANU]

Hi Fanghua,
No, you didn't miss anything - I had this problem as well and forget to tell Nic when he did the re-release. I agree we should add this to the instructions when we can.

[aekiss]

In the course of following these instructions I've also worked on my clone of the instructions (README.md) to address the issues I came across and also deal with those listed here. I'll circulate tomorrow for comment.

[aekiss]

Are the build instructions intended to just create JRA versions only or also CORE2?

I found python -m pytest test/test_build.py creates src/matm/build_nt62/matm_nt62.exe (the CORE2 version) instead of src/matm/build_jra55/matm_jra55.exe so is missing JRA.

In contrast, the non-pytest build creates create $ACCESS_OM_DIR/control/1deg_core_nyf but does not build the executable $ACCESS_OM_DIR/src/matm/build_nt62/matm_nt62.exe

[aekiss]

some more things: after going through the compilation

• control/025deg_jra55_ryf/ocean/input.nml needs io_layout = 6,5 added to the &ocean_model_nml group.
• control/*/ocean/input.nml need

     debug_this_module=.false.
     riverspread_diffusion=.true.
     riverspread_diffusion_passes = 2
     vel_micom_smooth = 0.2

added to &ocean_riverspread_nml group.

[nichannah]

Thank you @FanghuaWu and @aekiss this is very useful feedback. Let's plan to do another release soon to incorporate all of these fixes.

[aekiss]

OK here's my edit to README.md
README.txt
(named README.txt as github weirdly won't allow .md upload - should I do this as a pull request instead? - sorry, very new to github!)

I think it covers all the points raised above (other than the first two bullets in the opening post which I didn't really follow and weren't a problem for me) but let me know if I've missed anything. I tried to present it in a way that could be followed by a beginning grad student with basic unix skills but no other experience - but let me know if anything is more confusing than it should be.

A number of points (labelled TODO and BUG) would probably be good to follow up as part of the next release.

I've worked through these instructions (the non-pytest version) to check and have 1deg_jra55 and 025deg_jra queued but I'm yet to see whether they run successfully.

[aekiss]

I recently updated README.md with these commits: 457d216 cab127b 350bcfe 849103f

However the documentation is still out of date and inconsistent between README.md (https://github.com/OceansAus/access-om2/blob/master/README.md) and the wiki (https://github.com/OceansAus/access-om2/wiki). I think some of us should get together and redesign things to make it clearer and more coordinated.

Some suggestions:

• Make README.md into a much briefer overview, with detailed instructions merged and consolidated into the wiki
• Update wiki, e.g
  • https://github.com/OceansAus/access-om2/wiki/Basic-Model-Timestepping should use Payu
• Organise and partition documentation based on the needs of different users, e.g.
  • brief, clear, simple instructions for new users who want to obtain and analyse output from existing model runs but not do any runs themselves (i.e. a link to the cookbook)
  • brief, clear, simple instructions for new users who want to run model binaries (/short/public/access-om2/bin) without recompiling (but perhaps with changes to diags, namelists, forcing) - e.g. how to spinup or branch from an existing experiment and contribute config (https://github.com/OceansAus/access-om2/wiki/Contributing-to-model-configurations)
  • detailed, complete instructions for recompiling (i.e. most of what's in existing README.md) for more advanced users
  • instructions and notes by/for developers to remind us of how to we did things

[aekiss]

Looks like the binaries in /short/public/access-om2/bin are incomplete and out of date:

ls -l /short/public/access-om2/bin/
total 62284
-rwxr-xr-x 1 nah599 v45 7811595 Sep 20 21:45 cice_auscom_3600x2700_1200p_fe730022.exe*
-rwxr-xr-x 1 nah599 v45 7894078 Aug 31 15:02 cice_auscom_360x300_24p_fe730022.exe*
-rwxr-xr-x 1 nah599 v45 38198723 Aug 31 17:35 fms_ACCESS-OM_030fb1f2.x*
-rwxr-xr-x 1 nah599 v45 4884714 Aug 31 15:02 matm_jra55_2318e909.exe*
-rwxr-xr-x 1 nah599 v45 4883056 Sep 12 13:44 matm_nt62_b1f482c3.exe*
-rwxr-xr-x 1 nah599 v45 86952 Aug 31 15:02 mppnccombine*

cice_auscom_1440x1080_480p_fe730022.exe is missing

matm_jra55_2318e909.exe is 4 commits behind current (b1f482c3): https://github.com/OceansAus/matm/commits/master

Ideally /short/public/access-om2/bin should be automatically updated by a script as per #40.

[aekiss]

config submodules need better explanation: #42 (comment)

[nichannah]

May of the points in this issue have or are being addressed. The conversation is being referenced in the wiki: https://github.com/OceansAus/access-om2/wiki/Getting-started