Skip to content

Latest commit

 

History

History
92 lines (54 loc) · 4.74 KB

README.md

File metadata and controls

92 lines (54 loc) · 4.74 KB
Raw

k8s_kubeadm_cluster

Prerequisites

If you have an M-series (Apple Silicon) Mac, you cannot run VirtualBox. You could use VMWare.

VM Hardware Requirements

  • 8 GB of RAM (16 preferred)
  • 8-core/4-core hyperthreaded or better CPU, e.g. Core-i7/Core-i9 (will be slow otherwise)
  • 50 GB Disk space

git

Required to download the repo. It is normally pre-installed on Mac, but not on Windows. If you need to install it, see here.

VirtualBox

Download and Install VirtualBox on any one of the supported platforms:

  • Windows hosts
  • MacOS hosts x86 only.
  • Linux distributions

This lab was last tested with VirtualBox 7.0.12, though newer versions should be ok.

Vagrant

Once VirtualBox is installed you may choose to deploy virtual machines manually on it. Vagrant provides an easier way to deploy multiple virtual machines on VirtualBox more consistently.

Download and Install Vagrant on your platform.

This lab was last tested with Vagrant 2.3.7, though newer versions should be ok.

Lab Defaults

The labs have been configured with the following networking defaults. If you change any of these after you have deployed any of the lab, you'll need to completely reset it and start again from the beginning:

vagrant destroy -f
vagrant up

If you do change any of these, please consider that a personal preference and don't submit a PR for it.

Virtual Machine Network

Due to how VirtualBox/Vagrant works, the networking for each VM requires two network adapters; one NAT (enp0s3) to communicate with the outside world, and one internal (enp0s8) which is attached to the VirtualBox network mentioned above. By default, Kubernetes components will connect to the default network adapter - the NAT one, which is not what we want, therefore we have pre-set an environment variable PRIMARY_IP on all VMs which is the IP address that Kubernetes components should be using. In the coming labs you will see this environment variable being used to ensure Kubernetes components bind to the correct network interface.

PRIMARY_IP is defined as the IP address of the network interface on the node that is connected to the network having the default gateway, and is the interface that a node will use to talk to the other nodes. For those interested, this variable is assigned the result of the following command

ip route | grep default | awk '{ print $9 }'

Bridge Networking

The default configuration in this lab is to bring the VMs up on bridged interfaces. What this means is that your Kubernetes nodes will appear as additional machines on your local network, their IP addresses being provided dynamically by your broadband router. This facilitates the use of your browser to connect to any NodePort services you deploy.

Should you have issues deploying bridge networking, please raise a bug report and include all details including the output of vagrant up.

NAT Networking

In NAT configuration, the network on which the VMs run is isolated from your broadband router's network by a NAT gateway managed by the hypervisor. This means that VMs can see out (and connect to Internet), but you can't see in (i.e. use browser to connect to NodePorts) without setting up individual port forwarding rules for every NodePort using the VirtualBox UI.

The network used by the Virtual Box virtual machines is 192.168.56.0/24.

To change this, edit the Vagrantfile in your cloned copy (do not edit directly in github), and set the new value for the network prefix at line 9. This should not overlap any of the other network settings.

Note that you do not need to edit any of the other scripts to make the above change. It is all managed by shell variable computations based on the assigned VM IP addresses and the values in the hosts file (also computed).

It is recommended that you leave the pod and service networks as the defaults. If you change them then you will also need to edit the Weave networking manifests to accommodate your change.

If you do decide to change any of these, please treat as personal preference and do not raise a pull request.

Pod Network

The network used to assign IP addresses to pods is 10.244.0.0/16. This should not overlap any of the other network settings.

Service Network

The network used to assign IP addresses to Cluster IP services is 10.96.0.0/16. This should not overlap any of the other network settings.

CNI

On a newly created Kubernetes cluster, the Weave Net CNI pluging can be installed by running the following command:

kubectl apply -f https://reweave.azurewebsites.net/k8s/v1.29/net.yaml