SpinApp does not respond to `kubectl scale`

[bacongobbler]

Given the following SpinApp:

><> kubectl get spinapp hello-rust -o yaml
apiVersion: core.spinoperator.dev/v1
kind: SpinApp
metadata:
  annotations:
    kubectl.kubernetes.io/last-applied-configuration: |
      {"apiVersion":"core.spinoperator.dev/v1","kind":"SpinApp","metadata":{"annotations":{},"name":"hello-rust","namespace":"default"},"spec":{"executor":"containerd-shim-spin","image":"bacongobbler/hello-rust:latest","replicas":2}}
  creationTimestamp: "2024-02-20T18:47:21Z"
  generation: 1
  name: hello-rust
  namespace: default
  resourceVersion: "1716"
  uid: 3a4565a9-e0ab-4933-a3ce-eb01b7c908e9
spec:
  checks: {}
  enableAutoscaling: false
  executor: containerd-shim-spin
  image: bacongobbler/hello-rust:latest
  replicas: 2
  resources: {}
  runtimeConfig: {}
status:
  activeScheduler: containerd-shim-spin
  conditions:
  - lastTransitionTime: "2024-02-20T18:49:31Z"
    message: Deployment has minimum availability.
    reason: MinimumReplicasAvailable
    status: "True"
    type: Available
  - lastTransitionTime: "2024-02-20T18:47:21Z"
    message: ReplicaSet "hello-rust-77496795b6" has successfully progressed.
    reason: NewReplicaSetAvailable
    status: "True"
    type: Progressing
  readyReplicas: 2

kubectl get spinapp hello-rust works:

><> kubectl get spinapp hello-rust
NAME         READY REPLICAS   EXECUTOR
hello-rust   2                containerd-shim-spin

But kubectl scale spinapp hello-rust does not.

><> k scale spinapp hello-rust --replicas=3
Error from server (NotFound): spinapps.core.spinoperator.dev "hello-rust" not found

[bacongobbler]

I think this might've been due to the fact that I'm using an outdated fork... Recreating my dev environment to reproduce the issue.

[bacongobbler]

Nope. I was able to reproduce the same issue on main.

><> spin k8s scaffold --from bacongobbler/hello-rust:latest | kubectl apply -f -
spinapp.core.spinoperator.dev/hello-rust created
><> k get spinapps hello-rust -o yaml
apiVersion: core.spinoperator.dev/v1
kind: SpinApp
metadata:
  annotations:
    kubectl.kubernetes.io/last-applied-configuration: |
      {"apiVersion":"core.spinoperator.dev/v1","kind":"SpinApp","metadata":{"annotations":{},"name":"hello-rust","namespace":"default"},"spec":{"executor":"containerd-shim-spin","image":"bacongobbler/hello-rust:latest","replicas":2}}
  creationTimestamp: "2024-02-20T19:00:11Z"
  generation: 1
  name: hello-rust
  namespace: default
  resourceVersion: "1107"
  uid: c8a32ddf-3ace-4a05-a7ea-c106fd153fa2
spec:
  checks: {}
  enableAutoscaling: false
  executor: containerd-shim-spin
  image: bacongobbler/hello-rust:latest
  replicas: 2
  resources: {}
  runtimeConfig: {}
status:
  activeScheduler: containerd-shim-spin
  conditions:
  - lastTransitionTime: "2024-02-20T19:00:14Z"
    message: Deployment has minimum availability.
    reason: MinimumReplicasAvailable
    status: "True"
    type: Available
  - lastTransitionTime: "2024-02-20T19:00:11Z"
    message: ReplicaSet "hello-rust-77496795b6" has successfully progressed.
    reason: NewReplicaSetAvailable
    status: "True"
    type: Progressing
  readyReplicas: 2
><> k scale spinapp hello-rust --replicas 3
Error from server (NotFound): spinapps.core.spinoperator.dev "hello-rust" not found

[endocrimes]

We don't implement a /scale subresource - we currently have you disable our replicas if you want to integrate with things like HPA.

We could make the SpinApp CR scalable with something like:

// +kubebuilder:subresource:scale:specpath=.spec.replicas,statuspath=.status.readyReplicas,selectorpath=.status.selector

if we published the pod selector in our status too. But this isn't something we've discussed or thought through - I'm not sure how integrating with HPA works metrics wise if you're not a first party k8s component

[bacongobbler]

To be clear, there is an existing workaround. You can change the number of replicas in the manifest and call kubectl apply a second time.

><> spin k8s scaffold --from bacongobbler/hello-rust:latest | kubectl apply -f -
spinapp.core.spinoperator.dev/hello-rust created

><> spin k8s list
NAMESPACE       NAME            EXECUTOR                READY
default         hello-rust      containerd-shim-spin    2/2

><> spin k8s scaffold --from bacongobbler/hello-rust:latest --replicas 3 | kubectl apply -f -
spinapp.core.spinoperator.dev/hello-rust configured

><> spin k8s list
NAMESPACE       NAME            EXECUTOR                READY
default         hello-rust      containerd-shim-spin    3/3

[endocrimes]

Yep! (you can also kubectl edit and update the replicas there)

[bacongobbler]

Can we remove this from the Initial release project board? Doesn't seem totally necessary to solve this issue given that we have several documented workarounds in this thread.

[bacongobbler]

closing as duplicate of https://github.com/spinkube/spin-operator/issues/24.