Forum Discussion
kylec_251298
Jun 30, 2017Nimbostratus
Hmmm yeah the problem is the way rancher deploys kube-proxy. It deploys all of kubernetes as basically a docker-compose file and runs everything from containers.
This is how the Rancher kube-proxy is started.
proxy:
privileged: true
image: rancher/k8s:v1.5.4-rancher1-4
network_mode: host
command:
- kube-proxy
- --master=http://kubernetes.kubernetes.rancher.internal
- --v=2
- --healthz-bind-address=0.0.0.0
labels:
io.rancher.container.dns: 'true'
io.rancher.scheduler.global: 'true'
So adding a volume to the host won't do any good because
/etc/kubernetes/kubelet.conf
doesnt exist. That is running from another container.
I might need to contact Rancher to see how this could be done.