While automating the installtion of my K3S cluster with ansible, I struggled to disable traefik (and I’m not the only one, see this), the issue what that the file /etc/systemd/system/k3s.service was looking like this:
ExecStart=/usr/local/bin/k3s \
server \
'--write-kubeconfig-mode' \
' 0644' \
'--node-taint' \
'node-role.kubernetes.io/control-plane:NoSchedule' \
' --disable' \
'traefik' \
' --disable' \
'servicelb' \
' --disable' \
'local-storage' \
' --kubelet-arg' \
'config=/etc/rancher/k3s/kubelet.config' \
' --kube-controller-manager-arg' \
'bind-address=0.0.0.0' \
' --kube-proxy-arg' \
'metrics-bind-address=0.0.0.0' \
' --kube-scheduler-arg' \
'bind-address=0.0.0.0' \
' --kube-controller-manager-arg' \
'terminated-pod-gc-threshold=10' \
This version still ends up installing Traefik v2, bundled with K3S.
The correct k3s.service file look like the following:
ExecStart=/usr/local/bin/k3s \
server \
'--write-kubeconfig-mode' \
'0644' \
'--node-taint' \
'node-role.kubernetes.io/control-plane:NoSchedule' \
'--disable' \
'traefik' \
'--disable' \
'servicelb' \
'--disable' \
'local-storage' \
'--kubelet-arg' \
'config=/etc/rancher/k3s/kubelet.config' \
'--kube-controller-manager-arg' \
'bind-address=0.0.0.0' \
'--kube-proxy-arg' \
'metrics-bind-address=0.0.0.0' \
'--kube-scheduler-arg' \
'bind-address=0.0.0.0' \
'--kube-controller-manager-arg' \
'terminated-pod-gc-threshold=10' \
The fix was to configure yaml to account for end of line properly, check this to dig deeper into yaml multiline options.