alicloud-controller-manager

README

Kubernetes Cloud Controller Manager for Alibaba Cloud

cloud-controller-manager is the external Kubernetes cloud controller manager implementation for AliCloud(Alibaba Cloud). Running cloud-controller-manager allows you build your kubernetes clusters leverage on many cloud services on AliCloud. You can read more about Kubernetes cloud controller manager here.

WARNING: This project is still work in progress, be careful using it in production environment.

Requirements

Version

Kubernetes version 1.7.2 or higher is required to get a stable running.

AliCloud ECS

Only VPC network is supported.

Getting started

To deploy cloud-controller-manager in kubernetes cluster, we need to a few things:

• Get an cloud-controller-manager image.
• Prepare your kubernetes cluster with some requirements.
• Prepare and deploy cloud-controller-manager.
• Try it!

Get an cloud-controller-manager image

You can either get an image from official release by image name registry.cn-hangzhou.aliyuncs.com/acs/cloud-controller-manager:<RELEASE_VERSION>

Or build it from source which require docker has been installed:

```bash
# for example. export REGISTRY=registry.cn-hangzhou.aliyuncs.com/acs
$ export REGISTRY=<YOUR_REGISTRY_NAME>
# This will build cloud-controller-manager from source code and build an docker image from binary and push to your specified registry.
# You can also use `make binary && make build` if you don't want push this image to your registry.
$ make image
$ docker images |grep cloud-controller-manager
```

Prepare your kubernetes cluster with some requirements

--cloud-provider=external

In order to external cloud provider feature, we need to deploy or reconfigure kube-apiserver/kube-controller-manager/kubelet component with extra flag --cloud-provider=external, which means cloud provider functionality will hand to out of tree external cloud provider, here we use cloud-controller-manager.

How and where to set this flag depends on how you deploy your cluster, we will give a detail kubeadm way to deploy cluster with cloud-controller-manager later.

hostname and provider id

By default, the kubelet will name nodes based on the node's hostname. But in cloud-controller-manager, we use <REGION_ID>.<ECS_ID> format to build a unique node id to identity one node. In order to elimite these difference, we need to set extra flags --hostname_override and --provider-id to <REGION_ID>.<ECS_ID>.

If you are not sure how to find your ECS instance's ID and region id, try to run these command in your ECS instance:

```bash
$ META_EP=http://100.100.100.200/latest/meta-data
$ echo `curl -s $META_EP/region-id`.`curl -s $META_EP/instance-id`
```

Prepare and deploy cloud-controller-manager

1. Prepare AliCloud access key id and secret

apiVersion: v1
kind: Secret
metadata:
  name: cloud-config
  namespace: kube-system
data:
  # insert your base64 encoded AliCloud access id and key here, ensure there's no trailing newline:
  # to base64 encode your token run:
  #      echo -n "abc123abc123doaccesstoken" | base64
  access-key-id: "<ACCESS_KEY_ID>"
  access-key-secret: "<ACCESS_KEY_SECRET>"

1. Prepare cloud-controller-manager deployment yaml

apiVersion: extensions/v1beta1
kind: Deployment
metadata:
  name: cloud-controller-manager
  namespace: kube-system
spec:
  replicas: 1
  revisionHistoryLimit: 2
  template:
    metadata:
      labels:
        app: cloud-controller-manager
    spec:
      dnsPolicy: Default
      tolerations:
        # this taint is set by all kubelets running `--cloud-provider=external`
        - key: "node.cloudprovider.kubernetes.io/uninitialized"
          value: "true"
          effect: "NoSchedule"
      containers:
      - image: registry.cn-hangzhou.aliyuncs.com/acs/cloud-controller-manager:v1.8.4-14-gca3c1da
        name: cloud-controller-manager
        command:
          - /cloud-controller-manager
          # set leader-elect=true if you have more that one replicas
          - --leader-elect=false
          - --allocate-node-cidrs=true
          # set this to what you set to controller-manager or kube-proxy
          - --cluster-cidr=192.168.0.0/20
          # if you want to use a secure endpoint or deploy in a kubeadm deployed cluster, you need to use a kubeconfig instead.
          - --master=<YOUR_MASTER_INSECURE_ENDPOINT>
        env:
          - name: ACCESS_KEY_ID
            valueFrom:
              secretKeyRef:
                name: cloud-config
                key: access-key-id
          - name: ACCESS_KEY_SECRET
            valueFrom:
              secretKeyRef:
                name: cloud-config
                key: access-key-secret

Mare sure container image, --cluster-cidr and --master field match your needs.

1. Deploy cloud-controller-manager

$ kubectl create -f cloud-controller-manager.yaml

Try it!

Once cloud-controller-manager is up and running, run a sample nginx deployment:

$ cat <<EOF >nginx.yaml
apiVersion: extensions/v1beta1
kind: Deployment
metadata:
  name: nginx-example
spec:
  replicas: 1
  revisionHistoryLimit: 2
  template:
    metadata:
      labels:
        app: nginx-example
    spec:
      containers:
      - image: nginx:latest
        name: nginx
        ports:
          - containerPort: 80
EOF

$ kubectl create -f nginx.yaml

Then create service with type: LoadBalancer:

$ kubectl expose deployment nginx-example --name=nginx-example --type=LoadBalancer
$ kubectl get svc
NAME            CLUSTER-IP        EXTERNAL-IP     PORT(S)        AGE
nginx-example   192.168.250.19    106.xx.xx.xxx   80:31205/TCP   5s