I have been trying to setup a Kubernetes cluster where deployed containers should get their IP address from the DHCP server on the network.

When deploying these containers manually, I run the following commands :

INTERFACE="eno2" # Host interface
CONTAINER_ID=$(docker run -d --rm --name="$CONTAINER_NAME" <snip>)
PID=$(docker inspect --format='{{.State.Pid}}' "$CONTAINER_ID")
ip link add link "$INTERFACE" eth1 netns "$PID" type macvlan mode bridge

Now if I do :

dhclient eth1

on the container, I can reach the container by going to the IP address on eth1. Essentially, the container behaves as if it was a physical machine ( or a virtual machine ) connected directly to the network.

I am trying to manage this container using Kubernetes, and I am trying to use CNI plugins. Here is my /etc/cni/net.d/10-multus.conf :

{
    "cniVersion": "0.2.0",
    "name": "macvlan-dhcp",
    "type": "macvlan",
    "master": "eno2",
    "isGateway": true,
    "ipMasq": true,
    "ipam": {
         "type": "dhcp"
    }
}

But containers are not starting up. I did a test deployment for hello-node application and tried :

kubectl get deployment hello-node -o yaml

and I am getting :

apiVersion: extensions/v1beta1
kind: Deployment
<snip>
status:
  conditions:
  - lastTransitionTime: 2018-07-23T04:04:25Z
    lastUpdateTime: 2018-07-23T04:04:25Z
    message: Deployment does not have minimum availability.
    reason: MinimumReplicasUnavailable
    status: "False"
    type: Available
  - lastTransitionTime: 2018-07-23T04:14:25Z
    lastUpdateTime: 2018-07-23T04:14:25Z
    message: ReplicaSet "hello-node-7b788668d8" has timed out progressing.
    reason: ProgressDeadlineExceeded
    status: "False"
    type: Progressing
  observedGeneration: 1
  replicas: 1
  unavailableReplicas: 1
  updatedReplicas: 1

What am I missing here ?