Hi! I set up Rancher in high availability mode with an RKE cluster, although with one node for now. For some reason the node has the taint node.kubernetes.io/network-unavailable:NoSchedule and even if I remove it manually with kubectl, it is added back right away, automatically, and no new pods can be scheduled. Rancher still works fine even after rebooting but I would prefer having the Rancher cluster in a normal state before committing to use this cluster to manage other clusters. Now, I could just reinstall the Rancher cluster, but it’s already managing another cluster and I don’t want to set up both from scratch again if I can avoid that.
I tried creating an nginx pod and it stayed in scheduling phase until I removed the taint with kubectl, then the pod was scheduled but the taint as said is added automatically all the time. My understanding is that this taint means that there is a problem with the network plugin, Canal in this case, but the nginx pod with an ingress is working fine so I would think that the network plugin is working? Or is it not related?
Any help on how to permanently remove that taint would be much appreciated. Thanks!