Add a dockerfile to build CAM-SIMA

[K20shores]

• Adds a dockerfile to build an example case that uses CAM-SIMA
• I based much of this work off of the ESCOMP-Containers repository, just as an FYI

This will allow the fine-grain testing of newly-added components to CAM-SIMA.

[K20shores]

This is in draft because CAM-SIMA doesn't run successfully; I get a SIGILL. I am trying to debug why

[K20shores]

@nusbaume If you follow the instructions in the readme for building first esmf, and then cam sima, you should be able to start the container and run ./case.submit and see a run of cam-sima happen in docker.

[nusbaume]

@K20shores It looks like your new test-musica PR (#225) contains many of the same container files. Does that PR supersede this one (and so we can go-ahead and close this), or did you still want this to come in separately?

[K20shores]

@nusbaume I branched off of this branch to do the work in #225. I suppose that specific work in #225 really only tests a build of CAM-SIMA with Musica. If you don't care to have a base image that doesn't include Musica, we can abandon this PR.

[nusbaume]

I think since the vast majority of our eventual model configurations will want some type of chemistry, and because it would be good to test even seemingly independent physics changes with MICM linked in to make sure we aren't breaking that link, I would probably vote to only have one container (at least for now) that includes Musica.

@mattldawson @peverwhee @mwaxmonsky @boulderdaze Would any of you disagree with this thought process? Happy to be over-ruled here.

[mattldawson]

I think since the vast majority of our eventual model configurations will want some type of chemistry, and because it would be good to test even seemingly independent physics changes with MICM linked in to make sure we aren't breaking that link, I would probably vote to only have one container (at least for now) that includes Musica.

@mattldawson @peverwhee @mwaxmonsky @boulderdaze Would any of you disagree with this thought process? Happy to be over-ruled here.

Hi @nusbaume,

I have no strong feelings here, although if forced to choose, I would probably opt to be able to build CAM-SIMA images with and without MUSICA. Even if most of the time people will use MUSICA with CAM-SIMA, it could still be useful to be able to build and test them independently to ensure that they don't become too entangled. This could make it easier when someone wants to use a different chemistry (like GEOS-Chem). But, I'm fine with either option.

[nusbaume]

@mattldawson That's a good point, I hadn't thought about things like GEOS-Chem. I am fine to keep both containers then. I'll still plan on getting to the Test musica PR first as I imagine it is required for any future chemistry work, but will then put this PR as second on my to-do list, unless I hear otherwise. Thanks!

[K20shores]

@cacraigucar I have added some installation instructions for docker desktop to the readme. Do they seem comprehensive enough to you?

[mwaxmonsky]

@K20shores Is there anyway to verify the case ran successfully in the container beyond that it generated log without error?

[K20shores]

@K20shores Is there anyway to verify the case ran successfully in the container beyond that it generated log without error?

not that I'm aware of. @nusbaume would know better

[nusbaume]

@K20shores Feel free to merge this PR whenever you're ready. Thanks!