knative-serving: Disable scale-to-zero feature

[antoineco]

Description

Changes proposed in this pull request:

• Disable Knative Serving's scale-to-zero feature

Side effect: forces every Revision to have at least one replica running. Revisions multiply over time, leaving running Pods around.

$ kubectl -n kyma-integration get revision
NAME               CONFIG NAME   K8S SERVICE NAME   GENERATION   READY
stuff-h49dg        stuff         stuff-h49dg        6            True
stuff-j69tt        stuff         stuff-j69tt        4            True
stuff-kvx5b        stuff         stuff-kvx5b        7            True
stuff-tpv6n        stuff         stuff-tpv6n        5            True

$ kubectl -n kyma-integration get pod -l  serving.knative.dev/service=stuff
NAME                                      READY   STATUS    RESTARTS   AGE
stuff-h49dg-deployment-fd66946f9-p27qz    3/3     Running   0          15m
stuff-j69tt-deployment-5bdfd76b8c-4qzzr   3/3     Running   0          24m
stuff-kvx5b-deployment-66569865c7-vlztn   3/3     Running   0          15m
stuff-tpv6n-deployment-84979b8ddc-qvwqq   3/3     Running   0          15m

Related issue(s)

#5352
#6619

[k15r]

I don't see why we should do this. We can disable scale to zero per service using autoscaling.knative.dev/minScale: "1". Your solution would prevent scale-to-zero for all ksvcs.