Bug 1711043

fixed by https://github.com/openshift/cluster-svcat-apiserver-operator/pull/56

This applies only to the Service Catalog API Server Operator - - Service Catalog Controller Manager is independent and continues to work as it was before.

When cluster admin changes the ServiceCatalogAPIServer resource to have a `managementState: Removed` the operator will log a new condition:

  message: Unable to automatically remove Service Catalog once installed. Request to remove will be ignored. See documentation for more details.
  reason: Unable to automatically remove Service Catalog once installed.
  status: "True"
  type: RemovalRequestIgnored

and will *not* delete the namespace.

The admin can then either change the managementStage back to Managed (which would clear this RemoveRequestIgnored condition) or cluster admin can consult doc where they will be told if they proceed, gc will delete all secrets that were created by service catalog bindings.  Alternatively, we may provide a script/oc commands they can run that would remove the owenerRef from these secrets.  If they want to proceed with removing Service Catalog, cluster admin should execute `oc delete namespace openshift-service-catalog`.  This will trigger finalizer code in the Operator and will result in the Service Catalog API Server being completely removed.  Last step is the cluster admin must `oc delete APIService v1beta1.servicecatalog.k8s.io`.   At this point the operator reflects that service catalog api server has been removed - admin could re-install if desired.

This will be better addressed in 4.1.z.

That fixed PR wasn't merged in the payload: 4.1.0-0.nightly-2019-05-16-223922, but 4.1.0-0.nightly-2019-05-17-041605.

1, Enable Service Catalog by changing `managementState: Removed` to `managementState: Managed`.
mac:~ jianzhang$ oc get pods -n openshift-service-catalog-apiserver 
NAME              READY   STATUS    RESTARTS   AGE
apiserver-45tbw   1/1     Running   0          36m
apiserver-cn5sm   1/1     Running   0          36m
apiserver-wxhnh   1/1     Running   0          36m
mac:~ jianzhang$ oc get pods -n openshift-service-catalog-controller-manager
NAME                       READY   STATUS    RESTARTS   AGE
controller-manager-g9srl   1/1     Running   0          36m
controller-manager-hkkgc   1/1     Running   0          36m
controller-manager-l4s2s   1/1     Running   0          36m

2, Deploy a sample broker.
mac:~ jianzhang$ oc get clusterservicebroker
NAME         URL                                                        STATUS   AGE
ups-broker   http://ups-broker.kube-service-catalog.svc.cluster.local   Ready    1m
mac:~ jianzhang$ oc get clusterserviceclass
NAME                                   EXTERNAL-NAME                        BROKER       AGE
4f6e6cf6-ffdd-425f-a2c7-3c9258ad2468   user-provided-service                ups-broker   1m
5f6e6cf6-ffdd-425f-a2c7-3c9258ad2468   user-provided-service-single-plan    ups-broker   1m
8a6229d4-239e-4790-ba1f-8367004d0473   user-provided-service-with-schemas   ups-broker   1m

mac:~ jianzhang$ oc get pods -n kube-service-catalog
NAME                          READY   STATUS    RESTARTS   AGE
ups-broker-5f8568bc95-fzp9q   1/1     Running   0          4m16s

3, Disable Service Catalog by changing `managementState: Managed` to `managementState: Removed`.
The APIServer ignore the `Removed` status, and the controller-manager worked as before.
And the `openshift-service-catalog-controller-manager` project was deleted as expected.

mac:~ jianzhang$ oc get servicecatalogapiserver cluster -o yaml
apiVersion: operator.openshift.io/v1
kind: ServiceCatalogAPIServer
metadata:
  annotations:
    release.openshift.io/create-only: "true"
  creationTimestamp: "2019-05-17T08:51:17Z"
  generation: 3
  name: cluster
  resourceVersion: "29211"
  selfLink: /apis/operator.openshift.io/v1/servicecatalogapiservers/cluster
  uid: ef5a050b-7880-11e9-aad8-02aeb8a97e12
spec:
  logLevel: Normal
  managementState: Removed
status:
  conditions:
  - lastTransitionTime: "2019-05-17T09:19:13Z"
    status: "True"
    type: Available
  - lastTransitionTime: "2019-05-17T09:18:02Z"
    status: "False"
    type: Progressing
  - lastTransitionTime: "2019-05-17T08:57:55Z"
    reason: Removed
    status: "False"
    type: Degraded
  - lastTransitionTime: "2019-05-17T09:17:53Z"
    reason: NoUnsupportedConfigOverrides
    status: "True"
    type: UnsupportedConfigOverridesUpgradeable
  - lastTransitionTime: "2019-05-17T09:17:53Z"
    status: "False"
    type: ResourceSyncControllerDegraded
  - lastTransitionTime: "2019-05-17T09:17:59Z"
    status: "False"
    type: WorkloadDegraded
  - lastTransitionTime: "2019-05-17T09:57:53Z"
    message: Unable to automatically remove Service Catalog once installed. Request
      to remove will be ignored. See documentation for more details.
    reason: Unable to automatically remove Service Catalog once installed.
    status: "True"
    type: RemovalRequestIgnored
  generations:
  - group: apps
    hash: ""
    lastGeneration: 2
    name: apiserver
    namespace: openshift-service-catalog-apiserver
    resource: daemonsets
  observedGeneration: 2
  readyReplicas: 0

mac:~ jianzhang$ oc get servicecatalogcontrollermanager cluster -o yaml
apiVersion: operator.openshift.io/v1
kind: ServiceCatalogControllerManager
metadata:
  creationTimestamp: "2019-05-17T08:57:55Z"
  generation: 3
  name: cluster
  resourceVersion: "29341"
  selfLink: /apis/operator.openshift.io/v1/servicecatalogcontrollermanagers/cluster
  uid: dc6d898d-7881-11e9-952c-022716466838
spec:
  logLevel: Normal
  managementState: Removed
status:
  conditions:
  - lastTransitionTime: "2019-05-17T09:19:20Z"
    reason: Removed
    status: "True"
    type: Available
  - lastTransitionTime: "2019-05-17T09:18:25Z"
    reason: Removed
    status: "False"
    type: Progressing
  - lastTransitionTime: "2019-05-17T08:57:55Z"
    reason: Removed
    status: "False"
    type: Degraded
  - lastTransitionTime: "2019-05-17T09:18:22Z"
    status: "False"
    type: WorkloadDegraded
  generations:
  - group: apps
    hash: ""
    lastGeneration: 2
    name: controller-manager
    namespace: openshift-service-catalog-controller-manager
    resource: daemonsets
  observedGeneration: 2
  readyReplicas: 0
  version: 4.1.0-0.nightly-2019-05-17-041605

LGTM for the current fix. Verify it.

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory, and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

https://access.redhat.com/errata/RHBA-2019:0758