Reloader keeps logging warnings about DeploymentConfigs

[Stef16Robbe]

Describe the bug
Reloader operator produces dozens of logs per second warning about DeploymentConfigs (DC) being deprecated after OpenShift 4.14:

...
W1031 08:20:26.998706 1 warnings.go:70] apps.openshift.io/v1 DeploymentConfig is deprecated in v4.14+, unavailable in v4.10000+
W1031 08:20:27.157885 1 warnings.go:70] apps.openshift.io/v1 DeploymentConfig is deprecated in v4.14+, unavailable in v4.10000+
W1031 08:20:27.299301 1 warnings.go:70] apps.openshift.io/v1 DeploymentConfig is deprecated in v4.14+, unavailable in v4.10000+
...

To Reproduce

1. Install on OpenShift 4.14/4.15
2. Leave reloader.isOpenshift to it's default false, haven't tried setting it to true but we don't use DC's

Expected behavior
No messages about DC's

Screenshots
Log message should say enough

Environment

• Operator Version: v1.1.0
• Kubernetes/OpenShift Version: Both on 4.14 and 4.15

Additional context

• This looks to be coming from the K8S api, it's shown regularly when using it, e.g.:

$ oc status
Warning: apps.openshift.io/v1 DeploymentConfig is deprecated in v4.14+, unavailable in v4.10000+

• My Deployment manifest:

---
apiVersion: apps/v1
kind: Deployment
metadata:
  name: reloader-reloader
spec:
  replicas: 1
  revisionHistoryLimit: 2
  selector:
    matchLabels:
      app: reloader-reloader
  template:
    metadata:
      labels:
        app: reloader-reloader
    spec:
      containers:
        - image: "ghcr.io/stakater/reloader:v1.1.0"
          args:
            - --reload-strategy=annotations
          imagePullPolicy: IfNotPresent
          name: reloader-reloader
          ports:
            - name: http
              containerPort: 9090
          livenessProbe:
            httpGet:
              path: /live
              port: http
            timeoutSeconds: 5
            failureThreshold: 5
            periodSeconds: 10
            successThreshold: 1
            initialDelaySeconds: 10
          readinessProbe:
            httpGet:
              path: /metrics
              port: http
            timeoutSeconds: 5
            failureThreshold: 5
            periodSeconds: 10
            successThreshold: 1
            initialDelaySeconds: 10
          securityContext: {}
      securityContext:
        runAsNonRoot: true
        seccompProfile:
          type: RuntimeDefault
      serviceAccountName: reloader-reloader

[karl-johan-grahn]

We will attempt to reproduce it on Kind cluster to verify it's an OpenShift specific issue

[rasheedamir]

Yes @karl-johan-grahn this is OpenShift specific issue; as DC is OpenShift concept and NOT a Kubernetes concept; and DCs are going away soon.

[MuneebAijaz]

@Stef16Robbe code is particularly checking for an API resource to exist to determine if its OCP or not, that causes this issue https://github.com/stakater/Reloader/blob/master/pkg/kube/client.go#L71

[Stef16Robbe]

@MuneebAijaz That makes sense. Do you have a fix in mind? 😄

[juv]

This deprecation warning is literally spamming in our logs 24/7 and sometimes even cause CPU throttling on the reloader pod

[sachinavina]

Same issue has been seen in OCP 4.16 as well.
In a span of 5hrs it's printed 342 times.

$ oc logs reloader-8c4df5f54-z9b9d | grep -i "apps.openshift.io/v1 DeploymentConfig is deprecated" | wc -l
$ 342