search

kubeflow / pytorch-operator

PyTorch on Kubernetes
Apache License 2.0
304 stars 143 forks source link

Leverage E2e test framework in tf-operator #108

Open johnugeorge opened 5 years ago

johnugeorge commented 5 years ago

From https://github.com/kubeflow/pytorch-operator/pull/101#issuecomment-440104646

Investigate how we can leverage e2e test framework in tf-operator

Related: #101

Akado2009 commented 5 years ago

/assign Akado2009

johnugeorge commented 5 years ago

In addition to this, we need to add more tests for different job conditions similar to TF operator

johnugeorge commented 5 years ago

@richardsliu Shall we replicate the test framework after modifying for the pytorch operator? Any other options?

richardsliu commented 5 years ago

Can we move the test framework code (test_runner etc) into kubeflow/testing? That way we don't need to replicate the code in every repository.

Akado2009 commented 5 years ago

@jlewi How do you feel about that?

Akado2009 commented 5 years ago

@richardsliu @jlewi so waht do you think about that? Sounds like a great idea

richardsliu commented 5 years ago

The risk of moving test framework code into kubeflow/testing is that if it breaks, all presubmit tests in all repos also break. Fixing the test framework also requires a PR to be merged in kubeflow/testing first, before the changes take effect in kubeflow/tf-operator.

We currently don't have a lot of repos that utilize the Python test framework. Perhaps pytorch can just import the python libraries directly from tf-operator?