`undefined/undefined` is wrong...

[jsoref]

Checklist:

• [x] I've searched in the docs and FAQ for my answer: https://bit.ly/argocd-faq.
• [x] I've included steps to reproduce the bug.
• [x] I've pasted the output of argocd version.

Describe the bug

the details for a not found resource reflected undefined/undefined:

Resource not found in cluster: undefined/undefined:cert-manager-startupapicheck

To Reproduce

1. let argocd manage a cert-manager install
2. let argocd upgrade cert-manager install
3. while argocd is triggering the cert-manager upgrade, a service account for cert-manager-startupapicheck will appear and can be clicked (?)
4. click it
5. depending on (hand waving) the cert-manager-startupapicheck will probably already be gone? so a confusing error will be rendered by argocd

note that for a while longer i could ask argocd to filter for service accounts and see it, but i couldn't click it to actually show the item... and it eventually stopped appearing...

Expected behavior

information about expect objects should be stored in the request pointer so that they can be filled in in cases when they no longer exist instead of asking the ui to imagine them after the fact

Screenshots

Version

{
    "Version": "v2.8.3+77556d9",
    "BuildDate": "2023-09-07T16:05:43Z",
    "GitCommit": "77556d9e64304c27c718bb0794676713628e435e",
    "GitTreeState": "clean",
    "GoVersion": "go1.20.6",
    "Compiler": "gc",
    "Platform": "linux/amd64",
    "KustomizeVersion": "v5.1.0 2023-06-19T16:58:18Z",
    "HelmVersion": "v3.12.1+gf32a527",
    "KubectlVersion": "v0.24.2",
    "JsonnetVersion": "v0.20.0"
}

Logs

Paste any relevant application logs here.

[faust64]

Not sure it's related ... but as of upgrading ArgoCD (from 2.7.14 to 2.8.4), I'm seeing similar messages. For just any object, eg:

Resource not found in cluster: v1/ConfigMap:cronjob-tektonruns-garbage-collect-7h57hctc55 Please update your resource specification to use the latest Kubernetes API resources supported by the target cluster. The recommended syntax is v1/ConfigMap:cronjob-tektonruns-garbage-collect-7h57hctc55

Object is present in cluster though. Application shows healthy, it's not OutOfSync. I would just get that message when looking at individual resources "Live Manifest" in ArgoCD console.

This affects all objects in my applications: Services, ConfigMaps, NetworkPolicies ... Although I don't have that "undefined/undefined" thing ...

Still testing 2.8 / first installation I'm upgrading. I can confirm this is a not an issue, with exact same applications, running 2.7.14 Using that OpenShift GitOps operator lifecycling my ArgoCD -- ArgoCD object did not change in a while, as we did proceed with sso/dex reconfiguration when upgrading from 2.6 to 2.7.

[jsoref]

Sorry, that's a different issue. If I were you, I'd join slack to debug and then file a ticket (feel free to reference it here so people can find it if they stumble on your note here instead of the new ticket).

[faust64]

I am. Still I'm wondering: does the live manifest view works at all? You may be having two problems here.

[jsoref]

Yes, live manifest works fine for everything else.

[ku524]

I am also experiencing the same issue. My version of ArgoCD is 2.9.1. In my case, it was a pod, and I couldn't delete this pod through any method supported by ArgoCD. Note that the actual pod had already been deleted. However, removing the owner allowed me to delete this pod.

[pargupta12]

@jsoref: is your issue resolved? I am also facing same issue and getting error message something like "Resource not found in cluster: undefined/undefined:ly-job-a67f890-postsync-1707682950"

if you have some workaround. Please suggest.Thanks

[jsoref]

No, it isn't resolved. I just deal with it.

[bofaalsarah]

We just upgraded our ArgoCD from 2.7.6 to 2.10.1 and we are facing a similar issue

Resource not found in cluster: v1/ConfigMap:bitbucket-runner-configs-59t96976td Please update your resource specification to use the latest Kubernetes API resources supported by the target cluster. The recommended syntax is v1/ConfigMap:bitbucket-runner-configs-59t96976td

[jsoref]

Sorry, that's a different issue. If I were you, I'd join slack to debug and then file a ticket (feel free to reference it here so people can find it if they stumble on your note here instead of the new ticket).

[bofaalsarah]

I think it's still related because when I look into the live manifest tab of the pod, I am getting

Resource not found in cluster: undefined/undefined:bitbucket-runner-7c87665b77-m886p

[jsoref]

Oh, sorry, I didn't see that in what you pasted earlier. Yes, that's this problem :).

[jsoref]

Fwiw, this is more about error reporting than fixing the underlying problem causing any given error.

[jsoref]

Fwiw, this appears to have been introduced by @ashutosh16 in #13031 -- it's odd, because there are screenshots of it working there, so, presumably the feature works some of the time.

The relevant code is: https://github.com/argoproj/argo-cd/blob/891d08930470675c8851909254b0b5bd45eb79f0/ui/src/app/applications/components/application-node-info/application-node-info.tsx#L240-L241

[LS80]

Are you missing the cluster rbac for the server component?

[cmarasescu]

I had the same issues myself after install ArgoCD v2.10.1. I've followed the official install instructions (https://argo-cd.readthedocs.io/en/stable/getting_started/, with the only difference that I've use a custom namespace for installation (argocd-test)

kubectl create ns argocd-test
kubectl apply -n argocd-test -f https://raw.githubusercontent.com/argoproj/argo-cd/stable/manifests/core-install.yaml

Errors:

1. Issue 1: live manifest was showing errors for all resources, even tough the resources were deployed successfuly. In example, the errors for a configMap:

Resource not found in cluster: v1/ConfigMap:public-ingress-controller-test 
Please update your resource specification to use the latest Kubernetes API resources supported by the target cluster. 
The recommended syntax is v1/ConfigMap:public-ingress-controller-test

2. Issue 2: I was not able to delete resources from ArgoCD GUI - in example, while trying to delete a confgiMap, I got the following error:

   Unable to delete resource: error deleting resource: configmaps "public-ingress-controller-test" is forbidden: 
   User "system:serviceaccount:argocd-test:argocd-server" cannot delete resource "configmaps" in API group "" in the namespace "lu-ingress-controller"

Resolution:

• It seems that even tough I've installed ArgoCD in a custom namespace, the installation file used above - https://raw.githubusercontent.com/argoproj/argo-cd/stable/manifests/core-install.yaml - had hardcoded the argocd namespace for the argocd-server ClusterRoleBinding

  
  apiVersion: rbac.authorization.k8s.io/v1
  kind: ClusterRoleBinding
  metadata:
  labels:
  app.kubernetes.io/component: server
  app.kubernetes.io/name: argocd-server
  app.kubernetes.io/part-of: argocd
  name: argocd-server
  roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: ClusterRole
  name: argocd-server
  subjects:

• kind: ServiceAccount name: argocd-server namespace: argocd

The solution in my case was to adapt the argocd-server ClusterRoleBinding to also include the argocd-server serviceAccount from my custom namespace argocd-test. Basically, under subjects I've added the following 2 lines:

subjects:
- kind: ServiceAccount
  name: argocd-server
  namespace: argocd-test

[ppodevlabs]

i'm facing the same issue. I've deployed argocd 2.10.2 in cluster setup with the applications in any namespace, i've made the proper configuration and deployed https://github.com/argoproj/argo-cd/tree/master/manifests/cluster-rbac too.

But my application keeps syncing

Actually the deployment/ingres/networkpolicy are deployed and healthy but argo does not seems to detect them.

Any idea what am i missing here?

[cooervo]

I was having this issue in our helm values file we had some custom config for server.clusterRoleRules.rules and had server.clusterRoleRules.enabled: true that was overriding default config notice:

{{- if .Values.createClusterRoles }}
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
  name: {{ include "argo-cd.server.fullname" . }}
  labels:
    {{- include "argo-cd.labels" (dict "context" . "component" .Values.server.name "name" .Values.server.name) | nindent 4 }}
rules:
  {{- if .Values.server.clusterRoleRules.enabled }}
    {{- toYaml .Values.server.clusterRoleRules.rules | nindent 2 }}
  {{- else }}
  - apiGroups:
      - '*'

had to remove the custom rules and server.clusterRoleRules.enabled: false same as in default argocd values file. and now I can see all resources just fine

[andrii-korotkov-verkada]

ArgoCD versions 2.10 and below have reached EOL. Can you upgrade and let us know if the issue is still present, please?

[jsoref]

Unless someone has fixed this, it's almost certainly still present. Reproducing it requires a resource to be visible long enough to try to click on it but to be gone by the time argocd tries to ask questions about it (effectively a race). I don't really have the energy to try to produce this race at this time.

Further, if you're going to comment on 100 of my issues asking me to reproduce, I'll get even grumpier. Please be nice to contributors who have filed lots of tickets. Your comment is cheap, the work required for me to respond to each of them is not.

[andrii-korotkov-verkada]

There were ~1600 open bugs, so at least some managing is needed. I know my comment is less than ideal, but I don't see a better way to prioritize at this point. Also, there's no easy way to tell if the issue got fixed. For example, perf improvements in 2.13 for getting application tree can fix a whole class of refresh-related issues, but there's no 100% way to know.

I'll try to improve the situation with issues that don't get as much attention, but it'd take time and require multiple steps. For now sure, I won't ping your issues asking to reproduce.

[jsoref]

Because we happened to trip on it today, here's a screenshot from v2.12.4+27d1e64:

[ap0phi5]

Same here. We are also seeing this exact issue. v2.12.3+6b9cd82