This issue is related to this provisioning workflow: https://github.com/kubernetes/kubernetes/tree/master/docs/getting-started-guides/coreos/azure

The provisioning works fine but as soon as I login into kube-00 (i.e. "ssh -F ./output/kube_7d10e66148c601_ssh_conf kube-00") I constantly see this message:

CoreOS stable (647.2.0)
Update Strategy: No Reboots
Failed Units: 2
docker.service
weave-create-bridge.service

If I login in into kube-01 and kube-02 there are more services that had failures:

CoreOS stable (647.2.0)
Update Strategy: No Reboots
Failed Units: 4
docker.service
kubectl-create-minion.service
weave-create-bridge.service
docker.socket

If I login into the etcd nodes they don't show any error (Failed Units).

I have re-deployed the solution at least 10 times. The infrastructure provisioning always goes smooth but the issue above surfaces constantly inside the instances.