Multicluster with VelaD

Multicluster with VelaD

There are many scenarios that developers may want to deploy their application across multi-clusters. This article will introduce you to how make use of VelaD to try multi-cluster features of KubeVela. If you don't know the basic use of VelaD, you can follow this document.

Prerequisite

Windows/macOS System.

Targets

In this article, there are several targets:

1. Create a cluster with KubeVela with VelaD as control plane of clusters.
2. Create one cluster as managed cluster.
3. Join managed cluster into KubeVela
4. Deliver one multi-cluster application.

Architecture

The three targets above will actually build two clusters within the same Docker network(so they can access each other).

The image below shows how multi-cluster scenario works:

When using VelaD, clusters are K3d which running in Docker container. Note that these clusters are in the same Docker network, which enables the access between them.

Steps

Create control plane

This step has nothing special, velad install is all you need which has benn shown in 01.simple. Here we just skip the output of install.

velad install

Create managed cluster

Create the second cluster, namely managed cluster.

velad install --name sub-cluster --cluster-only

In this command, use --name flag to distinguish it from first cluster which name is default. use --cluster-only to tell VelaD not to install KubeVela in this cluster.

expcted output

...(omit for brevity)
vela CLI is already installed, skip

Keep the token below if you want to restart the control plane
K10967ee649ab9910497b04c91ad9406e439736df46d1fee85e3451afb8d8cc1da3::server:fgwgYIwkmAQKnmyOIVXc

🚀  Successfully install a pure cluster! 
🔗  If you have a cluster with KubeVela, Join this as sub-cluster:
    vela cluster join $(velad kubeconfig --name sub-cluster --internal)
💻  When using gateway trait, you can access with 127.0.0.1:8082
🔑  To access the cluster, set KUBECONFIG:
    export KUBECONFIG=$(velad kubeconfig --name sub-cluster --host)

In the output, wee have seen lots of them. The second one is a new hint which tells how to join this cluster as managed cluster.

Join cluster into KubeVela

As hint above, we can use this command to join the cluster

vela cluster join $(velad kubeconfig --name sub-cluster --internal)

Here velad kubeconfig --name sub-cluster --internal print a kubeconfig file path which vela cluster join need. The --internal flag means this is for docker network.

Note: all cluster created by VelaD are in docker network "k3d-velad"

expected output

Successfully add cluster , endpoint: https://172.31.0.5:6443.

Use vela cluster list to show clusters joined. The second one is the managed cluster.

CLUSTER                         ALIAS   TYPE            ENDPOINT                ACCEPTED        LABELS
local                                   Internal        -                       true                  
k3d-velad-cluster-sub-cluster           X509Certificate https://172.31.0.5:6443 true                  

Deliver multi-cluster application.

After join cluster into KubeVela, we can deliver applications to different clusters.

Here is one multi-cluster app example, run:

cat <<EOF | vela up -f -
apiVersion: core.oam.dev/v1beta1
kind: Application
metadata:
  name: basic-topology
spec:
  components:
    - name: nginx-basic
      type: webservice
      properties:
        image: nginx
      traits:
        - type: expose
          properties:
            port: [80]
  policies:
    - name: topology-sub-cluster
      type: topology
      properties:
        clusters: ["k3d-velad-cluster-sub-cluster"]
EOF

This is the first example of KubeVela website for deploy application to multi-cluster.

You can check the status with vela status

vela status basic-topology

output

About:

  Name:         basic-topology               
  Namespace:    default                      
  Created at:   2022-05-28 18:44:30 +0800 CST
  Status:       running                      

Workflow:

  mode: DAG
  finished: true
  Suspend: false
  Terminated: false
  Steps
  - id:92cn9mrqwm
    name:deploy-topology-sub-cluster
    type:deploy
    phase:succeeded 
    message:

Services:

  - Name: nginx-basic  
    Cluster: k3d-velad-cluster-sub-cluster  Namespace: default
    Type: webservice
    Healthy Ready:1/1
    Traits:
      ✅ expose

There you go, you have deployed an application in multiple clusters with whole process is modeled as a declarative deployment plan. For more usage, check the KubeVela document

Tear down the environment

To tear down the environment, just run velad uninstall to uninstall two cluster.

velad uninstall --name=default
velad uninstall --name=sub-cluster

Summary

With VelaD, testing and developing application across the clusters become easy. No need for caring fee if you are using true clusters or network/kubeconfig trivia if you are using kind/k3d/minikube.