Open aekiss opened 1 year ago
@micaeljtoliveira MOM6 has a REPRO build option https://youtu.be/94m3CMTwJ1E?t=415
Is this equivalent to what we do in our build system?
@CodeGat and @jo-basevi have built a very nice model repro CI framework https://github.com/ACCESS-NRI/access-om2-configs/pull/2
This will be put in a separate GitHub actions repro so it can be reused for repro testing of other model configs, including ours.
Hello! Yes, we have been working on testing out the rest of the repro-ci tests for ACCESS-OM2
and putting more generic actions into https://github.com/ACCESS-NRI/reproducibility. There will more generic actions to port from access-om2-configs
(I can already see some good candidate steps and actions) but it would be great to see how it goes with ACCESS-OM3
as well! We will be finishing up and putting it through its paces around next week :)
I think it could be a good idea to prioritise this. We've now had a few instances where config-breaking PRs have been merged.
From a very quick peruse, I think the main thing stopping this from just working for us is that we need some ACCESS-OM3-specific code for extracting checksums from the output of a model run (e.g. https://github.com/ACCESS-NRI/access-om2-configs/tree/main/test). This should be pretty straightforward
It would be good to have a CI reproducibility test, as this is fundamental to regression testing but isn't guaranteed https://github.com/COSIMA/access-om3/issues/40
The test would be to cold-start two identical runs and see if the restarts are identical. The runs would need to be multi-processor but could otherwise be cheap (e.g. 1 day at 1° resolution), with restart saved at the end of each run.