https://rancher.com/ logo
Title
e

eager-coat-67039

01/24/2023, 12:22 PM
Hi everyone, I'm trying to create a new cluster using RKE2 (1.24.9+rke2r2) + W2022 LTSC, but I'm stuck on some odd networking/calico issue: windows pods can reach each other and external addresses, but they cannot reach any linux pod, even the cluster DNS IP 10.43.0.10 cannot be reached from windows pods. I have previous experience with a k8s cluster using RKE + Windows 2019, so not a first-time k8s user, but I've exhausted my options debugging it I'd appreciate if someone could give me an insight into whatever it could be.
Well, sometimes asking a question is enough to spark the insight. The problem was that my linux nodes has two dedicated SAN network interfaces, and somehow it's a routing problem, even everything else working fine ... once I disabled the interfaces, the windows pods could reach the linux nodes. Now that I know what is the problem, I have just to figure out how to properly configure rke/calico to use the correct network interface.
­čĹŹ 1
It was easy, just needed to inform which network interface will be used by calico during the cluster installation:
cat >/var/lib/rancher/rke2/server/manifests/rke2-calico-config.yaml <<'EOF'
apiVersion: <http://helm.cattle.io/v1|helm.cattle.io/v1>
kind: HelmChartConfig
metadata:
  name: rke2-calico
  namespace: kube-system
spec:
  valuesContent: |-
    installation:
      calicoNetwork:
        mtu: 1500
        nodeAddressAutodetectionV4:
          interface: eth0
EOF
Or, if the cluster is already installed, just change the
installation
manifest