I installed k8s 1.18 bare metal cluster with kubeadm:

• 3 external etcd
• 2 control plane
• 2 workers

I tested disaster recovery, I restored etcd from night backup. When I type

kubectl get pods

I see state of etcd, I can't see pods which were created before etcd restore. If create deployment, pods stuck with "Pending" status, plus I can't terminate pods, they are stuck in "Terminating" status. The workaround is to stop kubelet on all nodes including control plane, delete all docker containers and start kubelets on all nodes simultaneously.

How can I resync state in etcd and pods on my nodes? Thank you.

after creation of new deployment, pods appeared in "Pending" status,

kubectl logs

shows:

Error from server (NotFound): the server could not find the requested resource ( pods/log nginx-deployment-5c4689cfc4-5zkq9), kubectl describe shows error: the server doesn't have a resource type "nginx-deployment-5c4689cfc4-5zkq9"

kubectl describe

shows

error: the server doesn't have a resource type "nginx-deployment-5c4689cfc4-5zkq9"