Naming resources for Aggregated APIs

[kubernetes]

Naming resources for Aggregated APIs

[nilebox]

Having some way of short scopes as an alternative to fully-qualified namespaced resource names would be useful for Aggregated APIs. For example,

kubectl get servicecatalog:binding my-binding

where servicecatalog is an alias for the API server owning the binding resource type.

[MHBauer]

Example of fully qualified specification looks like

k get binding.servicecatalog.k8s.io
k get bindings.v1alpha1.servicecatalog.k8s.io --all-namespaces

Long and repetitive to type.

Even had we created a type that was unique at the time, there is no guarantee that a future type would not be created in the core with the same name later. (i.e. we created Binding type first, but later core apiserver comes along and creates a Binding type)
We basically need ICANN/IANA for types in the kube namespace.

[nilebox]

We basically need ICANN/IANA for types in the kube namespace.

it is not realistic to have a centralized registry of unique resource names in the decentralized Aggregated APIs world.

The only reliable solution provided out of the box is fully qualified namespace, but as you mentioned it's too long and repetitive for manual typing in kubectl. Thus, I would consider an alternative scope approach, and discuss what could be the generic scope which works for everyone:

• API server alias/key - simple and directly suits the API Aggregation which is essentially a collection of API servers under proxy umbrella
• hierarchical name resolution (e.g. bindings, servicecatalog.bindings, v1alpha1.servicecatalog.bindings etc) with prioritized list of API servers/namespace for conflict naming resolution (i.e. bindings would be resolved to Kubernetes core API once it is introduced)

[seans3]

I believe this same conflict dynamic is why TPR does not work with aggregation. The TPR names "binding.servicecatalog.k8s.io", "broker.servicecatalog.k8s.io", etc. overlap with the aggregated API of the same names. While we don't care about TPR, this same problem might also affect CRD.

[deads2k]

Partially qualified names are found too, so kubectl get binding.se works. You only need to specify enough to unambiguous. Scripts should always fully qualify.

[fejta-bot]

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

Prevent issues from auto-closing with an /lifecycle frozen comment.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or @fejta.
/lifecycle stale

[fejta-bot]

Stale issues rot after 30d of inactivity.
Mark the issue as fresh with /remove-lifecycle rotten.
Rotten issues close after an additional 30d of inactivity.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle rotten
/remove-lifecycle stale

[fejta-bot]

Rotten issues close after 30d of inactivity.
Reopen the issue with /reopen.
Mark the issue as fresh with /remove-lifecycle rotten.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/close