We need to have E2E tests to verify the mnist example works. Tests should ensure

1. Training works
2. Deploying model works
3. Sending predictions works

This is P1 because we see lots of customer demand for xgboost so we need to know that its working against the latest Kubeflow release.

I'm working on adding tests to the GitHub issue summarization example. We can use that as a model for adding tests to this example.

In addition here are some other resources:

• General instructions for writing an Argo workflow to be used as an E2E test

• Guide To Writing Tests for E2E examples - This is in an in progress doc to come up with instructions for writing E2E tests.

I'm bumping this to priority P1 because we currently use mnist as part of our getting started guides; e.g https://www.kubeflow.org/docs/guides/gke/gcp-e2e/