Terway CNI Network Plugin

CNI plugin for Alibaba Cloud VPC/ENI

English | 简体中文

Try It

Install Kubernetes

• Prepare Aliyun ECS instance. The ECS OS we tested is Centos 7.4/7.6.
• Install Kubernetes via kubeadm: create-cluster-kubeadm

After setup kubernetes cluster.

• Change iptables Forward default policy to ACCEPT on every node of cluster: iptables -P FORWARD ACCEPT.
• Check the rp_filter in sysctl parameters, set them to "0" on every node of cluster.

Make sure cluster up and healthy by kubectl get cs.

Install Terway network plugin

• VPC Mode

  VPC Mode, Using Aliyun VPC route table to connect the pods. Can assign dedicated ENI to Pod. Install method:
  Replace Network and access_key/access_secret in terway.yml with your cluster pod subnet and aliyun openapi credentials. Then use kubectl apply -f terway.yml to install Terway into kubernetes cluster.

• ENI Secondary IP Mode

  ENI Secondary IP Mode, Using Aliyun ENI's secondary ip to connect the pods. This mode not limited by VPC route tables quotation. Install method:
  Replace access_key/access_secret and security_group/vswitches in terway-multiip.yml with your aliyun openapi credentials and resources id. Then use kubectl apply -f terway-multiip.yml to install Terway into kubernetes cluster.

Terway requires the access_key have following RAM Permissions

{
  "Version": "1",
  "Statement": [{
      "Action": [
        "ecs:CreateNetworkInterface",
        "ecs:DescribeNetworkInterfaces",
        "ecs:AttachNetworkInterface",
        "ecs:DetachNetworkInterface",
        "ecs:DeleteNetworkInterface",
        "ecs:DescribeInstanceAttribute",
        "ecs:DescribeInstanceTypes",
        "ecs:AssignPrivateIpAddresses",
        "ecs:UnassignPrivateIpAddresses",
        "ecs:DescribeInstances",
        "ecs:ModifyNetworkInterfaceAttribute"
      ],
      "Resource": [
        "*"
      ],
      "Effect": "Allow"
    },
    {
      "Action": [
        "vpc:DescribeVSwitches"
      ],
      "Resource": [
        "*"
      ],
      "Effect": "Allow"
    }
  ]
}

Using kubectl get ds terway -n kube-system to watch plugin launching. Plugin install completed while terway daemonset available pods equal to nodes.

Terway network plugin usage

Vpc network container

On VPC installation mode, Terway will config pod's address using node's podCidr when pod not have any special config. eg:

[[email protected] ~]# kubectl run -it --rm --image busybox busybox
If you don't see a command prompt, try pressing enter.
/ # ip link
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue qlen 1
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
3: [email protected]: <BROADCAST,MULTICAST,UP,LOWER_UP,M-DOWN> mtu 1500 qdisc noqueue
    link/ether 46:02:02:6b:65:1e brd ff:ff:ff:ff:ff:ff
/ # ip addr show
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue qlen 1
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
       valid_lft forever preferred_lft forever
    inet6 ::1/128 scope host
       valid_lft forever preferred_lft forever
3: [email protected]: <BROADCAST,MULTICAST,UP,LOWER_UP,M-DOWN> mtu 1500 qdisc noqueue
    link/ether 46:02:02:6b:65:1e brd ff:ff:ff:ff:ff:ff
    inet 172.30.0.4/24 brd 172.30.0.255 scope global eth0
       valid_lft forever preferred_lft forever
    inet6 fe80::4402:2ff:fe6b:651e/64 scope link
       valid_lft forever preferred_lft forever

Using ENI network interface to get the performance equivalent to the underlying network

On VPC installation mode, Config eni request aliyun/eni: 1 in one container of pod. The following example will create an Nginx Pod and assign an ENI:

apiVersion: v1
kind: Pod
metadata:
  name: nginx
spec:
  containers:
  - name: nginx
    image: nginx
    resources:
      limits:
        aliyun/eni: 1

[[email protected] ~]# kubectl exec -it nginx sh
# ip addr show
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN qlen 1
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
       valid_lft forever preferred_lft forever
    inet6 ::1/128 scope host
       valid_lft forever preferred_lft forever
3: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UNKNOWN qlen 1000
    link/ether 00:16:3e:02:38:05 brd ff:ff:ff:ff:ff:ff
    inet 172.31.80.193/20 brd 172.31.95.255 scope global eth0
       valid_lft forever preferred_lft forever
    inet6 fe80::216:3eff:fe02:3805/64 scope link
       valid_lft forever preferred_lft forever
4: [email protected]: <BROADCAST,MULTICAST,UP,LOWER_UP,M-DOWN> mtu 1500 qdisc noqueue state UP
    link/ether 1e:60:c7:cb:1e:0e brd ff:ff:ff:ff:ff:ff
    inet6 fe80::1c60:c7ff:fecb:1e0e/64 scope link
       valid_lft forever preferred_lft forever

ENI Secondary IP Pod：

On ENI secondary IP installation mode, Terway will create & allocate ENI secondary IP for pod. The IP of pod will in same IP Range:

[[email protected] ~]# kubectl get pod -o wide
NAME                     READY   STATUS    RESTARTS   AGE   IP              NODE                                 NOMINATED NODE
nginx-64f497f8fd-ckpdm   1/1     Running   0          4d    192.168.0.191   cn-hangzhou.i-j6c86lmr8k9rk78ju0nc   <none>
[[email protected] ~]# kubectl get node -o wide cn-hangzhou.i-j6c86lmr8k9rk78ju0nc
NAME                                 STATUS   ROLES    AGE   VERSION   INTERNAL-IP     EXTERNAL-IP   OS-IMAGE                KERNEL-VERSION              CONTAINER-RUNTIME
cn-hangzhou.i-j6c86lmr8k9rk78ju0nc   Ready    <none>   12d   v1.11.5   192.168.0.154   <none>        CentOS Linux 7 (Core)   3.10.0-693.2.2.el7.x86_64   docker://17.6.2
[[email protected] ~]# kubectl exec -it nginx-64f497f8fd-ckpdm bash
[email protected]:/# ip addr show
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
       valid_lft forever preferred_lft forever
3: [email protected]: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default
    link/ether 4a:60:eb:97:f4:07 brd ff:ff:ff:ff:ff:ff link-netnsid 0
    inet 192.168.0.191/32 brd 192.168.0.191 scope global eth0
       valid_lft forever preferred_lft forever

Using network policy to limit accessible between containers

The Terway plugin is compatible with NetworkPolicy in the standard K8S to control access between containers, for example:

1. Create and expose an deployment for test

   [[email protected] ~]# kubectl run nginx --image=nginx --replicas=2
   deployment "nginx" created
   [[email protected] ~]# kubectl expose deployment nginx --port=80
   service "nginx" exposed
   

2. Run busybox to test connection to deployment:

   [[email protected] ~]# kubectl run busybox --rm -ti --image=busybox /bin/sh
   If you don't see a command prompt, try pressing enter.
   / # wget --spider --timeout=1 nginx
   Connecting to nginx (172.21.0.225:80)
   / #
   

3. Config network policy，only allow pod access which have run: nginx label:

   kind: NetworkPolicy
   apiVersion: networking.k8s.io/v1
   metadata:
     name: access-nginx
   spec:
     podSelector:
       matchLabels:
         run: nginx
     ingress:
     - from:
       - podSelector:
           matchLabels:
             access: "true"
   

4. The Pod access service without the specified label is rejected, and the container of the specified label can be accessed normally.

   [[email protected] ~]# kubectl run busybox --rm -ti --image=busybox /bin/sh
   If you don't see a command prompt, try pressing enter.
   / # wget --spider --timeout=1 nginx
   Connecting to nginx (172.21.0.225:80)
   wget: download timed out
   / #
   
   [

apiVersion: v1
kind: Pod
metadata:
  name: nginx
  annotations:
    k8s.aliyun.com/ingress-bandwidth: 1m
    k8s.aliyun.com/egress-bandwidth: 1m
spec:
  nodeSelector:
    kubernetes.io/hostname: cn-shanghai.i-uf63p6s96kf4jfh8wpwn
  containers:
  - name: nginx
    image: nginx:1.7.9
    ports:
    - containerPort: 80

docker build -t acs/terway:latest .

git clone https://github.com/AliyunContainerService/terway.git
docker run -i --rm \
  -v $(pwd)/terway:/go/src/github.com/AliyunContainerService/terway \
  -w /go/src/github.com/AliyunContainerService/terway \
  sunyuan3/gometalinter:v1 bash -c "go test -race ./..."

cd terway/tests
./test.sh --cluster-id ${clusterid} \
    --access-key ${ACCESS_KEY_ID} --access-secret ${ACCESS_KEY_SECRET} \
    --region ${region} --category vpc \
    --image ${terwayimage}

./test.sh --cluster-id c05ef31ec40754f6c99c995963e2e01ed \
    --access-key ******** --access-secret ******** \
    --region cn-huhehaote --category vpc \
    --image registry.cn-hongkong.aliyuncs.com/sunyuan/terway:45