Loadbalancer on "master" is "red"

[exocode]

Hi there,

• I deployed a three node k8s cluster (1 x master, 2 x worker),
• the cluster has cloud_provider.name=external
• I have 1 Loadbalancer
• And 1 Network
• I apply the current CMM with Network https://github.com/hetznercloud/hcloud-cloud-controller-manager/releases/latest/download/ccm-networks.yaml

I then deploy a Workload and a Loadbalancer 👍

---
apiVersion: "v1"
kind: Service
metadata:
  name: "nginx-hello-world"
  labels:
    app: "hello-world"
  annotations:
    load-balancer.hetzner.cloud/name: "lb-staging"
    load-balancer.hetzner.cloud/health-check-port: "10254" #You can find it in the ingress-nginx manifest.
spec:
  ports:
    - name: "http"
      port: 80
      targetPort: 80
  selector:
    app: "hello-world"
  type: LoadBalancer
---
apiVersion: "apps/v1"
kind: "Deployment"
metadata:
  name: "nginx-hello-world"
  labels:
    app: "hello-world"
spec:
  selector:
    matchLabels:
      app: "hello-world"
  strategy:
    type: "Recreate"
  template:
    metadata:
      labels:
        app: "hello-world"
    spec:
      containers:
        - image: "nginx"
          name: "nginx-hello-world"
          imagePullPolicy: "Always"
          ports:
            - containerPort: 80
              name: "http

Everything fine so far: the loadbalancer balances the requests to the two worker nodes and is reachable through its public IP.

As you can see the "Master" is "not healthy".

And the taints...

# taints
node-role.kubernetes.io/controlplane=true:NoSchedule 
node-role.kubernetes.io/etcd=true:NoExecute

# labels
role=master
topology.kubernetes.io/region=fsn1
topology.kubernetes.io/zone=fsn1-dc14

...seems normal for me: the controlpane should not be loadbalanced.

Do I have to simply "remove" the Master node from the loadbalancers target list in the Hetzner Dashboard? or is there "cleaner" way to get everything "green"?

Thank you in advance

[LKaemmerling]

Hey @exocode,

please have a look at this comment: https://github.com/hetznercloud/hcloud-cloud-controller-manager/issues/182#issuecomment-830577642 The CCM is not in charge of selecting the nodes that should be behind the LoadBalancer. The CCM gets a list of nodes from Kubernetes and simply adds those nodes to the LoadBalancer.

[exocode]

very nice: WORKED. Thank you