Best practices for configuring nodes with two network interfaces

General Discussions microk8s
1

Hi. I’ve tried microk8s, and it worked like a charm on AWS.
Then I switched to dedicated servers, which have two network interfaces.
When it’s a single node cluster - it all works as it should.
But after adding another node, from time to time, I’ve started getting the errors like below:

  Normal   Scheduled               4m36s  default-scheduler  Successfully assigned default/pod-using-nfs to gphotos-k8s-1
  Warning  FailedCreatePodSandBox  4m5s   kubelet            Failed to create pod sandbox: rpc error: code = Unknown desc = failed to setup network for sandbox "5377466d8cbd8e69533fd050f4e37087791367b044c377599138956e7b9ab1a2": error getting ClusterInformation: Get https://[10.152.183.1]:443/apis/crd.projectcalico.org/v1/clusterinformations/default: dial tcp 10.152.183.1:443: i/o timeout
  Warning  FailedCreatePodSandBox  3m24s  kubelet            Failed to create pod sandbox: rpc error: code = Unknown desc = failed to setup network for sandbox "90494effa859015a65e4ae2937913faa47e200d9cb310ea76cf82970fef45f92": error getting ClusterInformation: Get https://[10.152.183.1]:443/apis/crd.projectcalico.org/v1/clusterinformations/default: dial tcp 10.152.183.1:443: i/o timeout
  Warning  FailedCreatePodSandBox  2m43s  kubelet            Failed to create pod sandbox: rpc error: code = Unknown desc = failed to setup network for sandbox "04bdcc98491e8a8ddcb45f588d470b16d4ac49e244d5cff61fed9079490fb375": error getting ClusterInformation: Get https://[10.152.183.1]:443/apis/crd.projectcalico.org/v1/clusterinformations/default: dial tcp 10.152.183.1:443: i/o timeout
  Warning  FailedCreatePodSandBox  117s   kubelet            Failed to create pod sandbox: rpc error: code = Unknown desc = failed to setup network for sandbox "2f61a1974a74309ce1fc1efba82640af2824ada98503caebb226ecada46645dc": Get https://[10.152.183.1]:443/apis/crd.projectcalico.org/v1/ippools: dial tcp 10.152.183.1:443: i/o timeout
  Normal   Pulling                 105s   kubelet            Pulling image "alpine"
  Normal   Pulled                  105s   kubelet            Successfully pulled image "alpine" in 406.538693ms
  Normal   Created                 105s   kubelet            Created container app
  Normal   Started                 105s   kubelet            Started container app

What can be the case?

2

Did you solve it? What was the microk8s version?

3

I’m also experiencing this. Did someone solve this?