Argo CD can't delete an app if it cannot generate manifests. You need to either:

1. Reinstate/fix your repo.
2. Delete the app using --cascade=false and then manually deleting the resources.

See Diffing documentation for reasons resources can be OutOfSync, and ways to configure Argo CD to ignore fields when differences are expected.

Argo CD provides health for several standard Kubernetes types. The Ingress and StatefulSet types have known issues which might cause health check to return Progressing state instead of Healthy.

• Ingress is considered healthy if status.loadBalancer.ingress list is non-empty, with at least one value for hostname or IP. Some ingress controllers (contour, traefik) don't update status.loadBalancer.ingress field which causes Ingress to stuck in Progressing state forever.

• StatefulSet is considered healthy if value of status.updatedReplicas field matches to spec.replicas field. Due to Kubernetes bug kubernetes/kubernetes#68573 the status.updatedReplicas is not populated. So unless you run Kubernetes version which include the fix kubernetes/kubernetes#67570 StatefulSet might stay in Progressing state.

• Your StatefulSet or DaemonSet is using OnDelete instead of RollingUpdate strategy. See #1881.

As workaround Argo CD allows providing health check customization which overrides default behavior.

By default the password is set to the name of the server pod, as per the getting started guide.

To change the password, edit the argocd-secret secret and update the admin.password field with a new bcrypt hash. You can use a site like https://www.browserling.com/tools/bcrypt to generate a new hash. For example:

# bcrypt(password)=$2a$10$rRyBsGSHK6.uc8fntPwVIuLVHgsAhAX7TcdrqW/RADU0uh7CaChLa
kubectl -n argocd patch secret argocd-secret \
  -p '{"stringData": {
    "admin.password": "$2a$10$rRyBsGSHK6.uc8fntPwVIuLVHgsAhAX7TcdrqW/RADU0uh7CaChLa",
    "admin.passwordMtime": "'$(date +%FT%T%Z)'"
  }}'

Another option is to delete both the admin.password and admin.passwordMtime keys and restart argocd-server. This will set the password back to the pod name as per the getting started guide.

Add admin.enabled: "false" to the argocd-cm ConfigMap (see user management).

Argo CD might fail to generate Helm chart manifests if the chart has dependencies located in external repositories. To solve the problem you need to make sure that requirements.yaml uses only internally available Helm repositories. Even if the chart uses only dependencies from internal repos Helm might decide to refresh stable repo. As workaround override stable repo URL in argocd-cm config map:

data:
  # v1.2 or earlier use `helm.repositories`
  helm.repositories: |
    - url: http://<internal-helm-repo-host>:8080
      name: stable
  # v1.3 or later use `repositories` with `type: helm`
  repositories: |
    - type: helm
      url: http://<internal-helm-repo-host>:8080
      name: stable

Check if cluster secret has argocd.argoproj.io/secret-type: cluster label. If secret has the label but the cluster is still not visible then make sure it might be a permission issue. Try to list clusters using admin user (e.g. argocd login --username admin && argocd cluster list).

Use the following steps to reconstruct configured cluster config and connect to your cluster manually using kubectl:

kubectl exec -it <argocd-pod-name> bash # ssh into any argocd server pod
argocd-util kubeconfig https://<cluster-url> /tmp/config --namespace argocd # generate your cluster config
KUBECONFIG=/tmp/config kubectl get pods # test connection manually

Now you can manually verify that cluster is accessible from the Argo CD pod.

To terminate the sync, click on the "synchronisation" then "terminate":

Is some cases, the tool you use may conflict with Argo CD by adding the app.kubernetes.io/instance label. E.g. using Kustomize common labels feature.

Argo CD automatically sets the app.kubernetes.io/instance label and uses it to determine which resources form the app. If the tool does this too, this causes confusion. You can change this label by setting the application.instanceLabelKey value in the argocd-cm. We recommend that you use argocd.argoproj.io/instance.

!!! note When you make this change your applications will become out of sync and will need re-syncing.

See #1482.

Kubernetes has normalized your resource limits when they are applied, and then Argo CD has then compared the version in your generated manifests to the normalized one is Kubernetes - they won't match.

E.g.

• '1000m' normalized to '1'
• '0.1' normalized to '100m'
• '3072Mi' normalized to '3Gi'
• 3072 normalized to '3072' (quotes added)

To fix this use diffing customizations settings.

Argo CD uses a JWT as the auth token. You likely are part of many groups and have gone over the 4KB limit which is set for cookies. You can get the list of groups by opening "developer tools -> network"

• Click log in
• Find the call to <argocd_instance>/auth/callback?code=<random_string>

Decode the token at https://jwt.io/. That will provide the list of teams that you can remove yourself from.

See #2165.

Maybe you're behind a proxy that does not support HTTP 2? Try the --grpc-web flag.:

argocd ... --grpc-web

Your not running your server with correct certs.

If you're not running in a production system (e.g. you're testing Argo CD out), try the --insecure flag:

argocd ... --insecure

!!! warning "Do not use --insecure in production"

Most likely you forgot to set the url in argocd-cm to point to your ArgoCD as well. See also the docs