feat: remove unused variable NODE_NAME

[apricote]

This variable is unused. It does not matter where HCCM is running, as long as it can talk to the Kubernetes & Hetzner Cloud APIs.

[mraenker]

Hello there,

I am currently following this [1] tutorial and translating it to Ansible at the same time. Running my existing (and prviously working) playbook today resulted in a error.

I noticed, when trying to deploy my sampleservice today (recreating the cluster from the ground up by creating new servers and so on) using LoadBalancer provided by the Hcloud-Plugin, I never received an External IP, nor was the Loadbalancer created in the Hetzner Cloud Console.

Issueing kubectl get pods -A revealed, the hcloud-cloud-controller-manager was never able to start from the get go of the cluster:

NAMESPACE      NAME                                               READY   STATUS             RESTARTS         AGE
...
kube-system    hcloud-cloud-controller-manager-554c64cd95-g64r9   0/1     CrashLoopBackOff   26 (3m33s ago)   112m
kube-system    hcloud-csi-controller-5868f9d75c-wj4lj             5/5     Running            0                112m
kube-system    hcloud-csi-node-7qmg8                              3/3     Running            0                110m
kube-system    hcloud-csi-node-g6c5d                              3/3     Running            0                110m
kube-system    hcloud-csi-node-mkpdc                              3/3     Running            0                112m
kube-system    hcloud-csi-node-pqrsz                              3/3     Running            0                111m
kube-system    hcloud-csi-node-qwpln                              3/3     Running            0                110m
...

Running kc logs hcloud-cloud-controller-manager-554c64cd95-g64r9 --namespace kube-system revealed this error message:

I1029 10:02:16.787608       1 serving.go:348] Generated self-signed cert in-memory
W1029 10:02:16.788250       1 client_config.go:618] Neither --kubeconfig nor --master was specified.  Using the inClusterConfig.  This might not work.
F1029 10:02:17.524911       1 main.go:67] Cloud provider could not be initialized: could not init cloud provider "hcloud": environment variable "NODE_NAME" is required

For my viewpoint, this variable is at least flagged as "required" by main.go. This flag has to be removed there, or it is indeed required and NODE_NAME should not be removed at all. - I don't know, whats the right decision here.

/edit: Inside Ansible I am using the hcloud dynamic inventory, if that's of any concern.

(Please don't frown upon me due to this rather vague error description. I am still a newbie trying to learn Kubernetes.)

[1] https://community.hetzner.com/tutorials/install-kubernetes-cluster, but with 3 Master Nodes and 2 Worker Nodes.

Thank you for looking into this!

[apricote]

Hey @mraenker,

it looks like the tutorial uses the unstable manifest from the (now renamed) master branch for hcloud-cloud-controller-manager. The referenced container image tag in this manifest is for the last release version, which still requires the environment variable.

I would recommend you to use versioned manifests for deploying. We now also have a Helm Chart, that can make installing & upgrading easier.