Turtles installed with enabled fleet addon registers cluster twice under Fleet (clusters.fleet.cattle.io)

[thehejik]

What steps did you take and what happened?

After installing turtles:0.10.0 (+capi ui extension 0.5.0) under rancher v2.8-head (with fleet:0.9.6) with enabled Seemless integration with Fleet and CAPI (==fleet addon) and using Fleet for deploying clusterclass cluster from examples I can see that the cluster is actually occurring twice under Fleet/Clusters in fleet-default namespace.

What did you expect to happen?

The cluster will be registered only once under Fleet in fleet-default namespace.

How to reproduce it?

1. deploy rancher v2.8-head
2. Install turtles:0.10.0 and enabled Seemless integration with Fleet and CAPI while installing the turtles addon under rancher
3. enable docker infrastructure provider (+ create kind docker network, disable TLS verification for clusters)
4. enable custom kubeadm bootstrap and control-plane providers
5. go to Fleet, switch to fleet-local namespace and add two gitrepos using this git url https://github.com/rancher-sandbox/rancher-turtles-fleet-example.git branch clusterclass but different paths/names:
   i) /classes
   ii) /clusters
6. once the gitrepos above in fleet-local ns are Active switch to fleet-default namespace under Continuous Delivery (fleet) and go to Clusters.

There are two cluster apparently with the same name (see screenshot above), but there should be just one.

$ kubectl get clusters.fleet.cattle.io -n fleet-default
NAME       BUNDLES-READY   NODES-READY   SAMPLE-NODE            LAST-SEEN              STATUS
cluster1   1/1             6/6           cluster1-75m5m-5mcdb   2024-07-29T14:44:32Z   
c-rwcpd    1/1             6/6           cluster1-75m5m-5mcdb   2024-07-29T14:51:59Z 

Rancher Turtles version

turtles:0.10.0

used CAPI providers:

$ kubectl get capiproviders.turtles-capi.cattle.io -A
NAMESPACE                           NAME                    TYPE             PROVIDERNAME   INSTALLEDVERSION   PHASE
rancher-turtles-system              fleet                   addon                           v0.3.0             Ready
capi-kubeadm-control-plane-system   kubeadm-control-plane   controlPlane     kubeadm        v1.4.6             Ready
rke2-control-plane-system           rke2-control-plane      controlPlane     rke2           v0.4.1             Ready
capi-system                         cluster-api             core             cluster-api    v1.4.6             Ready
rke2-bootstrap-system               rke2-bootstrap          bootstrap        rke2           v0.4.1             Ready
capi-kubeadm-bootstrap-system       kubeadm-bootstrap       bootstrap        kubeadm        v1.4.6             Ready
default                             docker                  infrastructure   docker         v1.4.6             Ready

Anything else you would like to add?

The same clusterclass example cluster deployed without fleet addon enabled is present just once under CD/fleet/Clusters.

It seems that on downstream CAPI clusters with fleet addon enabled there are two instances of fleet-agent running.

> kubectl get deploy -A
NAMESPACE             NAME                   READY   UP-TO-DATE   AVAILABLE   AGE
cattle-fleet-system   fleet-agent            1/1     1            1           17h
fleet-addon-agent     fleet-agent            1/1     1            1           18h

Label(s) to be applied

/kind bug