Upgrade to recipes 0.9.4, and use kerchunk to open inputs

[cisaacstern]

@jordanplanders pangeo-forge/pangeo-forge-recipes#383 was merged this week, so upgrading this feedstock's required pangeo-forge-recipes version may likely resolve the root issue #3 we've been fighting all along. I'll run a test from this PR to see if that's the case. 🤞

[cisaacstern]

/run citrace_tracer

[cisaacstern]

@jordanplanders the test errored because of a bug in the 0.9.3 release. I'm fixing that in pangeo-forge/pangeo-forge-recipes#466, and will retry the test here once that fix makes it up to Pangeo Forge Cloud. Should be today.

[jordanplanders]

@cisaacstern Sounds good. I was surprised that it was still "in progress" this morning, but figured it might just be a lot of bytes to process. Thanks for your persistence!

[cisaacstern]

Actually still waiting for this bug fix to propagate up to Pangeo Forge Cloud, but going to make a few adjustments here in the meantime... I'll explain after committing.

[cisaacstern]

Two changes just made:

• bumped to recipes version 0.9.4 (now that 0.9.3 has been yanked).
• in recipe, swapped subset_inputs option for open_input_with_kerchunk. this is the new way to manage large inputs unlocked by the 0.9.4 release. hopefully I'm understanding this correctly and it will just work!

Now just waiting for the new version of the cloud application to build, after which I'll re-trigger the test here.

[cisaacstern]

Ok I think everything's fixed 🤞 so will re-run test now.

[cisaacstern]

/run citrace_tracer

[pangeo-forge]

The test failed, but I'm sure we can find out why!

Pangeo Forge maintainers are working diligently to provide public logs for contributors.
That feature is not quite ready yet, however, so please reach out on this thread to a
maintainer, and they'll help you diagnose the problem.

[jordanplanders]

@cisaacstern ah, the plot thickens! Let me know what I can probe.

[cisaacstern]

Thanks for the ping, @jordanplanders. I believe this is a versioning issue on the backend, which if I'm correct, should be solved by pangeo-forge/pangeo-forge-orchestrator#210. I'll re-trigger the test once that goes through. As noted in the comments in that PR thread, this is certainly helping to highlight the necessity of some broader infrastructure upgrades we're working on (which should hopefully prevent or greatly minimize this type of problem in the future).

[cisaacstern]

Ok I think this problem is solved... re-testing this recipe now.

[cisaacstern]

/run citrace_tracer

[cisaacstern]

Just going to get out ahead of the upcoming automated failure comment and say... this is going to fail! 😆 Hoping pangeo-forge/pangeo-forge-orchestrator#212 will fix the underlying issue (finally!).

[pangeo-forge]

The test failed, but I'm sure we can find out why!

Pangeo Forge maintainers are working diligently to provide public logs for contributors.
That feature is not quite ready yet, however, so please reach out on this thread to a
maintainer, and they'll help you diagnose the problem.

[cisaacstern]

/run citrace_tracer

[cisaacstern]

pangeo-forge/pangeo-forge-orchestrator#212 did not fix this 😮‍💨 🤦 ... going to try another approach.

[jordanplanders]

@cisaacstern Oh gracious! As always, if there is something I can do (investigate something about the files? I'm not sure how to contribute at this point 😬), let me know! 🤞

[cisaacstern]

@jordanplanders thanks for asking. At this point this is totally on me to fix what has ended up being a far more tangled backend service issue than I'd anticipated. I'll keep truckin' on that and ping you again here when it's (actually!) fixed.