TELCODOCS-1788: OpenShift 4.15 secrets disabled

[mikemckiernan]

Authored by @StephenJamesSmith.

Review HTML:

Special Considerations section: https://nvidia.github.io/cloud-native-docs/review/pr-27/openshift/latest/steps-overview.html#special-considerations-for-openshift-4-15

23.9.2 RN: https://nvidia.github.io/cloud-native-docs/review/pr-27/gpu-operator/latest/release-notes.html#known-limitations

[github-actions]

Documentation preview

https://nvidia.github.io/cloud-native-docs/review/pr-27

[mikemckiernan]

@StephenJamesSmith , I thought that I submitted this feedback before I lost internet connectivity. PTAL.

[mikemckiernan]

I honestly got lost about whether the issue was the registry is disabled or if it was that storage for the registry wasn't configured.

[mikemckiernan]

Because this is prep for installation, before checking pod status, I think Erwan suggested that customers could check if storage was configured ("Managed"?) for the image registry.

I've lost the output at this point, but on a bare-metal deployment, the following command did not show the storage as "Managed" (IIRC):

oc describe configs.imageregistry.operator.openshift.io cluster

On the cluster that Erwan started for me (AWS, not BM) I was noodling around with something like this to see if we could cook up a "If your output looks like (or does not look like), then the NVIDIA GPU Operator installation will be able to generate secrets and no further action is required. Proceed to "

oc get configs.imageregistry.operator.openshift.io cluster -o jsonpath='{.spec.storage}'

...but, I had an AWS cluster and the problem can't be repped on AWS.

[mikemckiernan]

While this is true, the customer won't have installed the Operator yet, I don't think. So, I don't think we need from here to the end. We could include a sample command that indicates the installation will succeed. I think it was the oc describe configs.imageregistry.operator.openshift.io cluster that Erwan suggested, with the key piece being the section under the Storage: field. As long as storage was configured, my understanding is that the secrets could be made and the Operator installation would succeed.

[mikemckiernan]

Stephen, I put a few notes at the end of Erwan's document. Unless someone knows of a better way, I'm thinking that the oc get configs.imageregistry.operator.openshift.io cluster -o jsonpath='{.spec.storage}' command is the best way to predict failure and the need to configure storage for the image registry.

[StephenJamesSmith]

@mikemckiernan Saw your "sloppy notes". Will you be making these changes to the PR?

[StephenJamesSmith]

s/instatllation/installation

[mikemckiernan]

I need to find a different codespell hook.

pre-commit run --files openshift/steps-overview.rst 
mixed line ending........................................................Passed
trim trailing whitespace.................................................Passed
check yaml...........................................(no files to check)Skipped
codespell................................................................Passed

Thank you for pointing it out to me.

[StephenJamesSmith]

Found just 1 typo.

[egallen]

@mikemckiernan except the typo, the content is good.
/lgtm