Use OpenELB in Layer 2 Mode

This document demonstrates how to use OpenELB in Layer 2 mode to expose a Service backed by two Pods. The Eip, Deployment and Service described in this document are examples only and you need to customize the commands and YAML configurations based on your requirements.

Prerequisites

  • You need to prepare a Kubernetes cluster where OpenELB has been installed. All Kubernetes cluster nodes must be on the same Layer 2 network (under the same router).
  • You need to prepare a client machine, which is used to verify whether OpenELB functions properly in Layer 2 mode. The client machine needs to be on the same network as the Kubernetes cluster nodes.
  • The Layer 2 mode requires your infrastructure environment to allow anonymous ARP/NDP packets. If OpenELB is installed in a cloud-based Kubernetes cluster for testing, you need to confirm with your cloud vendor whether anonymous ARP/NDP packets are allowed. If not, the Layer 2 mode cannot be used.

This document uses the following devices as an example:

Device Name IP Address MAC Address Description
master1 192.168.0.2 52:54:22:a3:9a:d9 Kubernetes cluster master
worker-p001 192.168.0.3 52:54:22:3a:e6:6e Kubernetes cluster worker 1
worker-p002 192.168.0.4 52:54:22:37:6c:7b Kubernetes cluster worker 2
i-f3fozos0 192.168.0.5 52:54:22:fa:b9:3b Client machine

Step 1: Enable strictARP for kube-proxy

In Layer 2 mode, you need to enable strictARP for kube-proxy so that all NICs in the Kubernetes cluster stop answering ARP requests from other NICs and OpenELB handles ARP requests instead.

  1. Log in to the Kubernetes cluster and run the following command to edit the kube-proxy ConfigMap:

    kubectl edit configmap kube-proxy -n kube-system
  2. In the kube-proxy ConfigMap YAML configuration, set data.config.conf.ipvs.strictARP to true.

    ipvs:
     strictARP: true
  3. Run the following command to restart kube-proxy:

    kubectl rollout restart daemonset kube-proxy -n kube-system

Step 2: Specify the NIC Used for OpenELB

If the node where OpenELB is installed has multiple NICs, you need to specify the NIC used for OpenELB in Layer 2 mode. You can skip this step if the node has only one NIC.

In this example, the master1 node where OpenELB is installed has two NICs (eth0 192.168.0.2 and eth1 192.168.1.2), and eth0 192.168.0.2 will be used for OpenELB.

Run the following command to annotate master1 to specify the NIC:

kubectl annotate nodes master1 layer2.openelb.kubesphere.io/v1alpha1="192.168.0.2"

Step 3: Create an Eip Object

The Eip object functions as an IP address pool for OpenELB.

  1. Run the following command to create a YAML file for the Eip object:

    vi layer2-eip.yaml
  2. Add the following information to the YAML file:

    apiVersion: network.kubesphere.io/v1alpha2
    kind: Eip
    metadata:
     name: layer2-eip
    spec:
     address: 192.168.0.91-192.168.0.100
     interface: eth0
     protocol: layer2

NOTE

  • The IP addresses specified in spec:address must be on the same network segment as the Kubernetes cluster nodes.

  • For details about the fields in the Eip YAML configuration, see Configure IP Address Pools Using Eip.

  1. Run the following command to create the Eip object:

    kubectl apply -f layer2-eip.yaml

Step 4: Create a Deployment

The following creates a Deployment of two Pods using the luksa/kubia image. Each Pod returns its own Pod name to external requests.

  1. Run the following command to create a YAML file for the Deployment:

    vi layer2-openelb.yaml
  2. Add the following information to the YAML file:

    apiVersion: apps/v1
    kind: Deployment
    metadata:
     name: layer2-openelb
    spec:
     replicas: 2
     selector:
       matchLabels:
         app: layer2-openelb
     template:
       metadata:
         labels:
           app: layer2-openelb
       spec:
         containers:
           - image: luksa/kubia
             name: kubia
             ports:
               - containerPort: 8080
  3. Run the following command to create the Deployment:

    kubectl apply -f layer2-openelb.yaml

Step 5: Create a Service

  1. Run the following command to create a YAML file for the Service:

    vi layer2-svc.yaml
  2. Add the following information to the YAML file:

    kind: Service
    apiVersion: v1
    metadata:
     name: layer2-svc
     annotations:
       lb.kubesphere.io/v1alpha1: openelb
       protocol.openelb.kubesphere.io/v1alpha1: layer2
       eip.openelb.kubesphere.io/v1alpha2: layer2-eip
    spec:
     selector:
       app: layer2-openelb
     type: LoadBalancer
     ports:
       - name: http
         port: 80
         targetPort: 8080
     externalTrafficPolicy: Cluster

NOTE

  • You must set spec:type to LoadBalancer.
  • The lb.kubesphere.io/v1alpha1: openelb annotation specifies that the Service uses OpenELB.
  • The protocol.openelb.kubesphere.io/v1alpha1: layer2 annotation specifies that OpenELB is used in Layer 2 mode.
  • The eip.openelb.kubesphere.io/v1alpha2: layer2-eip annotation specifies the Eip object used by OpenELB. If this annotation is not configured, OpenELB automatically uses the first available Eip object that matches the protocol. You can also delete this annotation and add the spec:loadBalancerIP field (for example, spec:loadBalancerIP: 192.168.0.91) to assign a specific IP address to the Service.
  • If spec:externalTrafficPolicy is set to Cluster (default value), OpenELB randomly selects a node from all Kubernetes cluster nodes to handle Service requests. Pods on other nodes can also be reached over kube-proxy.
  • If spec:externalTrafficPolicy is set to Local, OpenELB randomly selects a node that contains a Pod in the Kubernetes cluster to handle Service requests. Only Pods on the selected node can be reached.
  1. Run the following command to create the Service:

      kubectl apply -f layer2-svc.yaml

Step 6: Verify OpenELB in Layer 2 Mode

The following verifies whether OpenELB functions properly.

  1. In the Kubernetes cluster, run the following command to obtain the external IP address of the Service:

    root@master1:~# kubectl get svc
    NAME         TYPE           CLUSTER-IP      EXTERNAL-IP    PORT(S)        AGE
    kubernetes   ClusterIP      10.233.0.1      <none>         443/TCP        20h
    layer2-svc   LoadBalancer   10.233.13.139   192.168.0.91   80:32658/TCP   14s
  2. In the Kubernetes cluster, run the following command to obtain the IP addresses of the cluster nodes:

    root@master1:~# kubectl get nodes -o wide
    NAME    		STATUS   ROLES		AGE		VERSION   INTERNAL-IP     EXTERNAL-IP   OS-IMAGE             KERNEL-VERSION       CONTAINER-RUNTIME
    master1   		Ready    master		20h		v1.17.9   192.168.0.2     <none>        Ubuntu 18.04.3 LTS   4.15.0-55-generic    docker://19.3.11
    worker-p001   	Ready    worker		20h		v1.17.9   192.168.0.3     <none>        Ubuntu 18.04.3 LTS   4.15.0-55-generic    docker://19.3.11
    worker-p002   	Ready    worker		20h		v1.17.9   192.168.0.4     <none>        Ubuntu 18.04.3 LTS   4.15.0-55-generic    docker://19.3.11
  3. In the Kubernetes cluster, run the following command to check the nodes of the Pods:

    root@master1:~# kubectl get pod -o wide
    NAME                              READY   STATUS    RESTARTS   AGE     IP             NODE    		NOMINATED NODE   READINESS GATES
    layer2-openelb-7b4fdf6f85-mnw5k   1/1     Running   0          3m27s   10.233.92.38   worker-p001   <none>           <none>
    layer2-openelb-7b4fdf6f85-px4sm   1/1     Running   0          3m26s   10.233.90.31   worker-p002   <none>           <none>

NOTE

In this example, the Pods are automatically assigned to different nodes. You can manually assign Pods to different nodes.
  1. On the client machine, run the following commands to ping the Service IP address and check the IP neighbors:

    root@i-f3fozos0:~# ping 192.168.0.91 -c 4
    PING 192.168.0.91 (192.168.0.91) 56(84) bytes of data.
    64 bytes from 192.168.0.91: icmp_seq=1 ttl=64 time=0.162 ms
    64 bytes from 192.168.0.91: icmp_seq=2 ttl=64 time=0.119 ms
    64 bytes from 192.168.0.91: icmp_seq=3 ttl=64 time=0.145 ms
    64 bytes from 192.168.0.91: icmp_seq=4 ttl=64 time=0.123 ms
       
    --- 192.168.0.91 ping statistics ---
    4 packets transmitted, 4 received, 0% packet loss, time 3076ms
    rtt min/avg/max/mdev = 0.119/0.137/0.162/0.019 ms
    root@i-f3fozos0:~# ip neigh
    192.168.0.1 dev eth0 lladdr 02:54:22:99:ae:5d STALE
    192.168.0.2 dev eth0 lladdr 52:54:22:a3:9a:d9 STALE
    192.168.0.3 dev eth0 lladdr 52:54:22:3a:e6:6e STALE
    192.168.0.4 dev eth0 lladdr 52:54:22:37:6c:7b STALE
    192.168.0.91 dev eth0 lladdr 52:54:22:3a:e6:6e REACHABLE

In the output of the ip neigh command, the MAC address of the Service IP address 192.168.0.91 is the same as that of worker-p001 192.168.0.3. Therefore, OpenELB has mapped the Service IP address to the MAC address of worker-p001.

  1. On the client machine, run the curlcommand to access the Service:

    curl 192.168.0.91

If spec:externalTrafficPolicy in the Service YAML configuration is set to Cluster, both Pods can be reached.

   root@i-f3fozos0:~# curl 192.168.0.91
   You've hit layer2-openelb-7b4fdf6f85-px4sm
   root@i-f3fozos0:~# curl 192.168.0.91
   You've hit layer2-openelb-7b4fdf6f85-mnw5k
   
   root@i-f3fozos0:~# curl 192.168.0.91
   You've hit layer2-openelb-7b4fdf6f85-px4sm
   root@i-f3fozos0:~# curl 192.168.0.91
   You've hit layer2-openelb-7b4fdf6f85-mnw5k

If spec:externalTrafficPolicy in the Service YAML configuration is set to Local, only the Pod on the node selected by OpenELB can be reached.

   root@i-f3fozos0:~# curl 192.168.0.91
   You've hit layer2-openelb-7b4fdf6f85-mnw5k
   
   root@i-f3fozos0:~# curl 192.168.0.91
   You've hit layer2-openelb-7b4fdf6f85-mnw5k
   
   root@i-f3fozos0:~# curl 192.168.0.91
   You've hit layer2-openelb-7b4fdf6f85-mnw5k

Last modified March 14, 2022: Add link for vercel (44c8460)