Strimzi ArgoCD out of sync

[miroslavmasaryk3221]

Hello,

We are using strimzi kafka operator along with argoCD.

ArgoCD is reporting out of sync ClusterRoleBinding.

The desired manifest is empty

We are using onap-community strimzi application chart.
https://github.com/onap/oom

We have tried to ignore it like this:
|

---
apiVersion: argoproj.io/v1alpha1
kind: Application
metadata:
  name: onap-strimzi
  namespace: argocd
  annotations:
    argocd.argoproj.io/sync-wave: "-95"
spec:
  project: argo-management
  ignoreDifferences:
  - group: rbac.authorization.k8s.io
    kind: ClusterRoleBinding"
    name: strimzi-onap-onap-strimzi-kafka-init
    jsonPointers:
    - /apiVersion
    - /kind
    - /metadata
    - /managedFields
    - /name
    - /resourceVersion
    - /uid
    - /roleRef
    - /subjects
  syncPolicy:
    syncOptions:
    - RespectIgnoreDifferences=true
    automated:
      prune: false
      selfHeal: true
  destination:
    server: https://kubernetes.default.svc
    namespace: onap
  source:
    repoURL: <companyrepo>
    targetRevision: v10.0.0
    path: strimzi
    helm:
      valueFiles: 
      - ../onap/values-global.yaml
      - ../onap/values-strimzi.yaml

Any ideas?

Thanks

[Frawless]

Hi @miroslavmasaryk3221, have you tried force-sync if it produces some errors? I am using ArgoCD with our examples from main (https://github.com/strimzi/strimzi-kafka-operator/tree/main/packaging/install/cluster-operator) and I didn't see any errors with CRBs. Sounds to me more like a question for Onap community.

[scholzj]

The examples do not have Rack awareness enabled, so they do not created Cluster Role Bindings. I'm not using ArgoCD, but IIRC there were some older discussions about it with some solutions which might be worth looking through. The problem is I think that Cluster Role Bindings as a global-scoped resources cannot have owner reference and that is causing issues to Argo and some other methods need to be used there.

[ryuseongryong]

I've got the same issue.
Using strimzi-cluster-operator-0.31.1 and kafka-3.2.3 with argocd.
some cluster is good but the other one is got the same issue.
It seems to be created when strimzi operator pod check the creation or modification.
I think good cluster and bad cluster strimzi config is same but argocd version is different.
so need more research.

Is there anything new this issue?

[dbaarorapr]

we have same issue,do you have any workarounds

[MohamedRhimii]

Same issue.. Any solution/workaround?