test(ws): add e2e tests

[Adembc]

This PR makes the following changes to the workspace controller:

• Remove the +kubebuilder:validation:XValidation:rule="self == oldSelf" annotations for image config and pod config.
• Comment out probes as they prevent pod creation when empty:

  probes:
  # startupProbe: {}
  # livenessProbe: {}
  # readinessProbe: {}

• Fix typo in extraEnv NB_PREFIX value from "juptyerlab" to "jupyterlab".
• Remove installation and uninstallation of Prometheus Operator and Cert Manager in e2e tests - before and after each test.
• Create the default-editor service account and delete it after testing.
• Create home PVC and data PVC, and delete them afterwards.
• Apply some WorkspaceKind samples.
• Apply some Workspace samples.
• Check if the workspace pod is running.
• Test if the pod is exposed correctly through the service and path generated from the NB_PREFIX environment variable.

[Adembc]

Now the unit tests are failing because I removed the validation that prevented the WorkspaceKind fields .Spec.PodTemplate.Options.ImageConfig and .Spec.PodTemplate.Options.PodConfig from being mutable. These validations were causing errors in the workspace reconciliation loop. As discussed with @thesuperzapper, these validations should be removed and replaced by validation webhooks. Maybe we should merge the validation webhook PR first to ensure the unit tests pass.

[thesuperzapper]

@Adembc I have done the following changes in https://github.com/kubeflow/notebooks/pull/30/commits/61c85fa0c88149706f7a255f69ad6b4ed6723552

• Auto detect the current kind cluster name for the kind load docker-image ... command.

[thesuperzapper]

@Adembc I have done the following changes in https://github.com/kubeflow/notebooks/pull/30/commits/16886cededb63db5f815f554fc85de69089597e5

• Removed the tests which were failing because +kubebuilder:validation:XValidation:rule="self == oldSelf" was removed
• Made the default image ghcr.io/kubeflow/notebooks/workspace-controller:latest (as this will be the image name once we release)
• Grouped the samples under ./workspaces/controller/config/samples better to separate the "common" ones like ServiceAccount/default-editor and PVC/workspace-home-pvc from our actual custom CRDs.
• Introduced a system to get Warning type Events from the StatefulSet and use them to pouplate the Workspace status.stateMessage when the Pod is missing.
  • note, this required adding a new Kubebuilder index on the .involvedObject.uid field of Events
• Introduced new Workspace status.stateMessage information for cases when the Pod is unschedulable (e.g. due to no node meeting the CPU/RAM request)
• Refactored the ./workspaces/controller/test/e2e/e2e_test.go:
  • to not run tests in the default namespace
  • to check that the Workspace status.state is Running
  • to properly run a curl pod that checks the service is up
• Removed the now unused methods from ./workspaces/controller/test/utils/utils.go:
  • InstallPrometheusOperator()
  • UninstallPrometheusOperator()
  • UninstallCertManager()
  • InstallCertManager()

[thesuperzapper]

@Adembc thanks for your work on this! We can do follow ups in future PRs.

/lgtm /approve

[google-oss-prow[bot]]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: thesuperzapper

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files: - ~~[OWNERS](https://github.com/kubeflow/notebooks/blob/notebooks-v2/OWNERS)~~ [thesuperzapper] Approvers can indicate their approval by writing `/approve` in a comment Approvers can cancel approval by writing `/approve cancel` in a comment