Deploy Kubernetes and Kubesphere on Bare Metal

Introduction

In addition to the deployment on cloud, Kube AI Hub can also be installed on bare metal. As the virtualization layer is removed, the infrastructure overhead is drastically reduced, which brings more compute and storage resources to app deployments. As a result, hardware efficiency is improved. Refer to the example below to deploy Kube AI Hub on bare metal.

Prerequisites

• Make sure you already know how to install Kube AI Hub on a multi-node cluster based on the tutorial Multi-node Installation.
• Server and network redundancy in your environment.
• For a production environment, it is recommended that you prepare persistent storage and create a StorageClass in advance. For development and testing, you can use the integrated OpenEBS to provision LocalPV as the storage service directly.

Prepare Linux Hosts

This tutorial uses 3 physical machines of DELL 620 Intel (R) Xeon (R) CPU E5-2640 v2 @ 2.00GHz (32G memory), on which CentOS Linux release 7.6.1810 (Core) will be installed for the minimal deployment of Kube AI Hub.

Install CentOS

Download and install the image first, and CentOS Linux release 7.6.1810 (Core) is recommended. Make sure you allocate at least 200 GB to the root directory where it stores docker images (you can skip this if you are installing Kube AI Hub for testing).

For more information about the supported systems, see System Requirements.

Here is a list of the three hosts for your reference.

NIC settings

1. Clear NIC configurations.

   ifdown em1
   

   ifdown em2
   

   rm -rf /etc/sysconfig/network-scripts/ifcfg-em1
   

   rm -rf /etc/sysconfig/network-scripts/ifcfg-em2
   

2. Create the NIC bonding.

   nmcli con add type bond con-name bond0 ifname bond0 mode 802.3ad ip4 192.168.60.152/24 gw4 192.168.60.254
   

3. Set the bonding mode.

   nmcli con mod id bond0 bond.options mode=802.3ad,miimon=100,lacp_rate=fast,xmit_hash_policy=layer2+3
   

4. Bind the physical NIC.

   nmcli con add type bond-slave ifname em1 con-name em1 master bond0
   

   nmcli con add type bond-slave ifname em2 con-name em2 master bond0
   

5. Change the NIC mode.

   vi /etc/sysconfig/network-scripts/ifcfg-bond0
   BOOTPROTO=static
   

6. Restart Network Manager.

   systemctl restart NetworkManager
   

   nmcli con # Display NIC information
   

7. Change the host name and DNS.

   hostnamectl set-hostname worker-1
   

   vim /etc/resolv.conf
   

Time settings

1. Synchronize time.

   yum install -y chrony
   

   systemctl enable chronyd
   

   systemctl start chronyd
   

   timedatectl set-ntp true
   

2. Set the time zone.

   timedatectl set-timezone Asia/Shanghai
   

3. Check if the ntp-server is available.

   chronyc activity -v
   

Firewall settings

Execute the following commands to stop and disable the FirewallD service:

iptables -F

systemctl status firewalld

systemctl stop firewalld

systemctl disable firewalld

Package updates and dependencies

Execute the following commands to update system packages and install dependencies.

yum update

yum install openssl openssl-devel

yum install socat

yum install epel-release

yum install conntrack-tools

Download KubeKey

Kubekey is the next-gen installer which provides an easy, fast and flexible way to install Kubernetes and Kube AI Hub.

Follow the step below to download KubeKey.

curl -sfL https://get-kk.kubesphere.io | VERSION=v3.0.13 sh -

export KKZONE=cn

curl -sfL https://get-kk.kubesphere.io | VERSION=v3.0.13 sh -

Make kk executable:

chmod +x kk

Create a Multi-node Cluster

With KubeKey, you can install Kubernetes and Kube AI Hub together. You have the option to create a multi-node cluster by customizing parameters in the configuration file.

Create a Kubernetes cluster with Kube AI Hub installed (for example, --with-kubesphere v3.4.1):

./kk create config --with-kubernetes v1.22.12 --with-kubesphere v3.4.1

A default file config-sample.yaml will be created. Modify it according to your environment.

vi config-sample.yaml

apiVersion: kubekey.kubesphere.io/v1alpha1
kind: Cluster
metadata:
  name: config-sample
spec:
  hosts:
  - {name: master1, address: 192.168.60.152, internalAddress: 192.168.60.152, user: root, password: P@ssw0rd}
  - {name: worker1, address: 192.168.60.153, internalAddress: 192.168.60.153, user: root, password: P@ssw0rd}
  - {name: worker2, address: 192.168.60.154, internalAddress: 192.168.60.154, user: root, password: P@ssw0rd}
  roleGroups:
    etcd:
    - master1
    control-plane:
    - master1
    worker:
    - worker1
    - worker2
  controlPlaneEndpoint:
    domain: lb.kubesphere.local
    address: ""                    
    port: 6443

Create a cluster using the configuration file you customized above:

./kk create cluster -f config-sample.yaml

Verify the installation

After the installation finishes, you can inspect the logs of installation by executing the command below:

kubectl logs -n kubesphere-system $(kubectl get pod -n kubesphere-system -l 'app in (ks-install, ks-installer)' -o jsonpath='{.items[0].metadata.name}') -f

If you can see the welcome log return, it means the installation is successful.

**************************************************
#####################################################
###              Welcome to Kube AI Hub!           ###
#####################################################
Console: http://192.168.60.152:30880
Account: admin
Password: P@88w0rd
NOTES：
  1. After you log into the console, please check the
     monitoring status of service components in
     the "Cluster Management". If any service is not
     ready, please wait patiently until all components
     are up and running.
  2. Please change the default password after login.
#####################################################
https://kubesphere.io             20xx-xx-xx xx:xx:xx
#####################################################

Log in to the console

You will be able to use default account and password admin/P@88w0rd to log in to the console http://{$IP}:30880 to take a tour of Kube AI Hub. Please change the default password after login.

Enable pluggable components (Optional)

The example above demonstrates the process of a default minimal installation. To enable other components in Kube AI Hub, see Enable Pluggable Components for more details.

System Improvements

• Update your system.

  yum update
  

• Add the required options to the kernel boot arguments:

  sudo /sbin/grubby --update-kernel=ALL --args='cgroup_enable=memory cgroup.memory=nokmem swapaccount=1'
  

• Enable the overlay2 kernel module.

  echo "overlay2" | sudo tee -a /etc/modules-load.d/overlay.conf
  

• Refresh the dynamically generated grub2 configuration.

  sudo grub2-set-default 0
  

• Adjust kernel parameters and make the change effective.

  cat <<EOF | sudo tee -a /etc/sysctl.conf
  vm.max_map_count = 262144
  fs.may_detach_mounts = 1
  net.ipv4.ip_forward = 1
  vm.swappiness=1
  kernel.pid_max =1000000
  fs.inotify.max_user_instances=524288
  EOF
  sudo sysctl -p
  

• Adjust system limits.

  vim /etc/security/limits.conf
  *                soft    nofile         1024000
  *                hard    nofile         1024000
  *                soft    memlock        unlimited
  *                hard    memlock        unlimited
  root             soft    nofile         1024000
  root             hard    nofile         1024000
  root             soft    memlock        unlimited
  

• Remove the previous limit configuration.

  sudo rm /etc/security/limits.d/20-nproc.conf
  

• Reboot the system.

  reboot