How to Add Heterogeneous Nodes to a Worker Cluster¶

This page explains how to add ARM architecture nodes with Kylin v10 sp2 operating system to an AMD architecture worker cluster with CentOS 7.9 operating system.

Prerequisites¶

• A DCE 5.0 Full Mode deployment has been successfully completed, and the bootstrap node is still alive. Refer to the documentation Offline Installation of DCE 5.0 Enterprise for the deployment process.
• A worker cluster with AMD architecture and CentOS 7.9 operating system has been created through the DCE 5.0 platform. Refer to the documentation Creating a Worker Cluster for the creation process.

Procedure¶

Download and Import Offline Packages¶

Take ARM architecture and Kylin v10 sp2 operating system as examples.

Make sure you are logged into the bootstrap node! Also, make sure the clusterConfig.yaml file used during the DCE 5.0 deployment is available.

Offline Image Package¶

After downloading, extract the offline package:

tar -xvf offline-v0.18.0-arm64.tar

ISO Offline Package (Kylin v10 sp2)¶

osPackage Offline Package (Kylin v10 sp2)¶

The Kubean project provides osPackage offline packages for different operating systems. Visit https://github.com/kubean-io/kubean/releases to view the available packages.

Import Offline Packages to the Bootstrap Node¶

Run the import-artifact command:

./offline/dce5-installer import-artifact -c clusterConfig.yaml \
    --offline-path=/root/offline \
    --iso-path=/root/Kylin-Server-10-SP2-aarch64-Release-Build09-20210524.iso \
    --os-pkgs-path=/root/os-pkgs-kylinv10-v0.7.4.tar.gz

After a successful import command execution, the offline package will be uploaded to Minio on the bootstrap node.

Add Heterogeneous Worker Nodes¶

Make sure you are logged into the management node of the DCE 5.0 Global Service Cluster.

Modify the Host Manifest¶

Here is an example of host manifest:

apiVersion: v1
kind: ConfigMap
metadata:
  name: ${cluster-name}-hosts-conf
  namespace: kubean-system
data:
  hosts.yml: |
    all:
      children:
        etcd:
          hosts:
            centos-master:
        k8s_cluster:
          children:
            kube_control_plane:
            kube_node:
        kube_control_plane:
          hosts:
            centos-master:
        kube_node:
          hosts:
            centos-master:
    hosts:
      centos-master:
        ip: 10.5.14.122
        access_ip: 10.5.14.122
        ansible_host: 10.5.14.122
        ansible_connection: ssh
        ansible_user: root
        ansible_ssh_pass: ******

apiVersion: v1
kind: ConfigMap
metadata:
  name: ${cluster-name}-hosts-conf
  namespace: kubean-system
data:
  hosts.yml: |
    all:
      hosts:
        centos-master:
          ip: 10.5.14.122
          access_ip: 10.5.14.122
          ansible_host: 10.5.14.122
          ansible_connection: ssh
          ansible_user: root
          ansible_ssh_pass: ******
          # Add heterogeneous nodes
        kylin-worker:
          ip: 10.5.10.220
          access_ip: 10.5.10.220
          ansible_host: 10.5.10.220
          ansible_connection: ssh
          ansible_user: root
          ansible_ssh_pass: dangerous@2022
        children:
          kube_control_plane:
            hosts:
              - centos-master
          kube_node:
            hosts:
              - centos-master
              - kylin-worker  # Add the name of heterogeneous node
          etcd:
            hosts:
              - centos-master
          k8s_cluster:
            children:
              - kube_control_plane
              - kube_node

To add information about the newly added worker nodes according to the above comments:

kubectl edit cm ${cluster-name}-hosts-conf -n kubean-system

Add Expansion Tasks through ClusterOperation.yml¶

Example:

apiVersion: kubean.io/v1alpha1
kind: ClusterOperation
metadata:
  name: add-worker-node
spec:
  cluster: ${cluster-name} # Specify cluster name
  image: ghcr.m.daocloud.io/kubean-io/spray-job:v0.5.0
  backoffLimit: 0
  actionType: playbook
  action: scale.yml
  extraArgs: --limit=kylin-worker
  preHook:
    - actionType: playbook
      action: ping.yml
    - actionType: playbook
      action: disable-firewalld.yml
    - actionType: playbook
      action: enable-repo.yml
      extraArgs: |
        -e "{repo_list: ["http://10.5.14.30:9000/kubean/kylin-iso/\$releasever/os/\$basearch","http://10.5.14.30:9000/kubean/kylin/\$releasever/os/\$basearch"]}"
  postHook:
    - actionType: playbook
      action: cluster-info.yml

To create and deploy join-node-ops.yaml according to the above configuration:

vi join-node-ops.yaml
kubectl apply -f join-node-ops.yaml -n kubean-system

Check the status of the task execution¶

kubectl -n kubean-system get pod | grep add-worker-node

To check the progress of the scaling task, you can view the logs of the corresponding pod.

Verify in UI¶

1. Go to Container Management -> Clusters -> Nodes .

2. Click the newly added node to view details.

Comments