More thorough automated dask testing

[sgibson91]

Description

We wanted to make sure that the options listed in the following docs were all working on the Pangeo deployment before we go live with the prod hub: https://pangeo.io/cloud.html#dask Since these are quite generic tests of dask (e.g. don't require requester pays access to data, etc), it would be nice to add them to our daskhub tests so we can verify the scaling of all our daskhubs and be alerted to any issues. Our current test only tests local dask right now.

Value / benefit

• We can be confident in the performance of our daskhubs with little human effort
• We can be alerted to issues with our daskhubs with little human effort

Implementation details

I would expect there to be a check that the dask scheduler and worker pods were correctly created in the namespace and I'm not sure how to write that test

Tasks to complete

No response

Updates

No response

[damianavila]

I would expect there to be a check that the dask scheduler and worker pods were correctly created in the namespace and I'm not sure how to write that test

What about looking for some end result in the notebook that assures us the scheduler and the worked pods were created and worked as expected? Would not be that an easier test to write, perhaps?

[sgibson91]

This is Pangeo's current testing infrastructure:

https://github.com/pangeo-data/pangeo-cloud-federation/blob/staging/test.py

https://github.com/pangeo-data/pangeo-cloud-federation/blob/staging/run_tests.sh

[damianavila]

We already have the machinery to run tests so I guess it is a matter to include/adapt test.py into a test notebook, maybe?

[sgibson91]

Yep. I'll look into it once I've babysat the Auth0 stuff :)