Deploy a Production Ready Kubernetes Cluster

If you have questions, check the documentation at kubespray.io and join us on the kubernetes slack, channel #kubespray. You can get your invite here

Can be deployed on AWS, GCE, Azure, OpenStack, vSphere, Equinix Metal (bare metal), Oracle Cloud Infrastructure (Experimental), or Baremetal
Highly available cluster
Composable (Choice of the network plugin for instance)
Supports most popular Linux distributions
Continuous integration tests

Quick Start

Below are several ways to use Kubespray to deploy a Kubernetes cluster.

Ansible

Usage

Install Ansible according to Ansible installation guide then run the following steps:

# Copy ``inventory/sample`` as ``inventory/mycluster``
cp -rfp inventory/sample inventory/mycluster

# Update Ansible inventory file with inventory builder
declare -a IPS=(10.10.1.3 10.10.1.4 10.10.1.5)
CONFIG_FILE=inventory/mycluster/hosts.yaml python3 contrib/inventory_builder/inventory.py ${IPS[@]}

# Review and change parameters under ``inventory/mycluster/group_vars``
cat inventory/mycluster/group_vars/all/all.yml
cat inventory/mycluster/group_vars/k8s_cluster/k8s-cluster.yml

# Clean up old Kubernetes cluster with Ansible Playbook - run the playbook as root
# The option `--become` is required, as for example cleaning up SSL keys in /etc/,
# uninstalling old packages and interacting with various systemd daemons.
# Without --become the playbook will fail to run!
# And be mind it will remove the current kubernetes cluster (if it's running)!
ansible-playbook -i inventory/mycluster/hosts.yaml  --become --become-user=root reset.yml

# Deploy Kubespray with Ansible Playbook - run the playbook as root
# The option `--become` is required, as for example writing SSL keys in /etc/,
# installing packages and interacting with various systemd daemons.
# Without --become the playbook will fail to run!
ansible-playbook -i inventory/mycluster/hosts.yaml  --become --become-user=root cluster.yml

Note: When Ansible is already installed via system packages on the control node, Python packages installed via sudo pip install -r requirements.txt will go to a different directory tree (e.g. /usr/local/lib/python2.7/dist-packages on Ubuntu) from Ansible's (e.g. /usr/lib/python2.7/dist-packages/ansible still on Ubuntu). As a consequence, the ansible-playbook command will fail with:

ERROR! no action detected in task. This often indicates a misspelled module name, or incorrect module path.

This likely indicates that a task depends on a module present in requirements.txt.

One way of addressing this is to uninstall the system Ansible package then reinstall Ansible via pip, but this not always possible and one must take care regarding package versions. A workaround consists of setting the ANSIBLE_LIBRARY and ANSIBLE_MODULE_UTILS environment variables respectively to the ansible/modules and ansible/module_utils subdirectories of the pip installation location, which is the Location shown by running pip show [package] before executing ansible-playbook.

A simple way to ensure you get all the correct version of Ansible is to use the pre-built docker image from Quay. You will then need to use bind mounts to access the inventory and SSH key in the container, like this:

git checkout v2.25.0
docker pull quay.io/kubespray/kubespray:v2.25.0
docker run --rm -it --mount type=bind,source="$(pwd)"/inventory/sample,dst=/inventory \
  --mount type=bind,source="${HOME}"/.ssh/id_rsa,dst=/root/.ssh/id_rsa \
  quay.io/kubespray/kubespray:v2.25.0 bash
# Inside the container you may now run the kubespray playbooks:
ansible-playbook -i /inventory/inventory.ini --private-key /root/.ssh/id_rsa cluster.yml

Collection

See here if you wish to use this repository as an Ansible collection

Vagrant

For Vagrant we need to install Python dependencies for provisioning tasks. Check that Python and pip are installed:

python -V && pip -V

If this returns the version of the software, you're good to go. If not, download and install Python from here https://www.python.org/downloads/source/

Install Ansible according to Ansible installation guide then run the following step:

vagrant up

Documents

Supported Linux Distributions

Flatcar Container Linux by Kinvolk
Debian Bookworm, Bullseye
Ubuntu 20.04, 22.04, 24.04
CentOS/RHEL 7, 8, 9
Fedora 37, 38
Fedora CoreOS (see fcos Note)
openSUSE Leap 15.x/Tumbleweed
Oracle Linux 7, 8, 9
Alma Linux 8, 9
Rocky Linux 8, 9
Kylin Linux Advanced Server V10 (experimental: see kylin linux notes)
Amazon Linux 2 (experimental: see amazon linux notes)
UOS Linux (experimental: see uos linux notes)
openEuler (experimental: see openEuler notes)

Note: Upstart/SysV init based OS types are not supported.

Supported Components

Core
- kubernetes v1.30.4
- etcd v3.5.12
- docker v26.1
- containerd v1.7.20
- cri-o v1.30.3 (experimental: see CRI-O Note. Only on fedora, ubuntu and centos based OS)
Network Plugin
- cni-plugins v1.2.0
- calico v3.27.3
- cilium v1.15.4
- flannel v0.22.0
- kube-ovn v1.12.21
- kube-router v2.0.0
- multus v3.8
- weave v2.8.7
- kube-vip v0.8.0
Application
- cert-manager v1.14.7
- coredns v1.11.1
- ingress-nginx v1.11.2
- krew v0.4.4
- argocd v2.11.0
- helm v3.14.2
- metallb v0.13.9
- registry v2.8.1
Storage Plugin
- cephfs-provisioner v2.1.0-k8s1.11
- rbd-provisioner v2.1.1-k8s1.11
- aws-ebs-csi-plugin v0.5.0
- azure-csi-plugin v1.10.0
- cinder-csi-plugin v1.30.0
- gcp-pd-csi-plugin v1.9.2
- local-path-provisioner v0.0.24
- local-volume-provisioner v2.5.0
- node-feature-discovery v0.16.4

Container Runtime Notes

The cri-o version should be aligned with the respective kubernetes version (i.e. kube_version=1.20.x, crio_version=1.20)

Requirements

Minimum required version of Kubernetes is v1.28
Ansible v2.14+, Jinja 2.11+ and python-netaddr is installed on the machine that will run Ansible commands
The target servers must have access to the Internet in order to pull docker images. Otherwise, additional configuration is required (See Offline Environment)
The target servers are configured to allow IPv4 forwarding.
If using IPv6 for pods and services, the target servers are configured to allow IPv6 forwarding.
The firewalls are not managed, you'll need to implement your own rules the way you used to. in order to avoid any issue during deployment you should disable your firewall.
If kubespray is run from non-root user account, correct privilege escalation method should be configured in the target servers. Then the ansible_become flag or command parameters --become or -b should be specified.

Hardware: These limits are safeguarded by Kubespray. Actual requirements for your workload can differ. For a sizing guide go to the Building Large Clusters guide.

Master
- Memory: 1500 MB
Node
- Memory: 1024 MB

Network Plugins

You can choose among ten network plugins. (default: calico, except Vagrant uses flannel)

flannel: gre/vxlan (layer 2) networking.
Calico is a networking and network policy provider. Calico supports a flexible set of networking options designed to give you the most efficient networking across a range of situations, including non-overlay and overlay networks, with or without BGP. Calico uses the same engine to enforce network policy for hosts, pods, and (if using Istio and Envoy) applications at the service mesh layer.
cilium: layer 3/4 networking (as well as layer 7 to protect and secure application protocols), supports dynamic insertion of BPF bytecode into the Linux kernel to implement security services, networking and visibility logic.
weave: Weave is a lightweight container overlay network that doesn't require an external K/V database cluster. (Please refer to weave troubleshooting documentation).
kube-ovn: Kube-OVN integrates the OVN-based Network Virtualization with Kubernetes. It offers an advanced Container Network Fabric for Enterprises.
kube-router: Kube-router is a L3 CNI for Kubernetes networking aiming to provide operational simplicity and high performance: it uses IPVS to provide Kube Services Proxy (if setup to replace kube-proxy), iptables for network policies, and BGP for ods L3 networking (with optionally BGP peering with out-of-cluster BGP peers). It can also optionally advertise routes to Kubernetes cluster Pods CIDRs, ClusterIPs, ExternalIPs and LoadBalancerIPs.
macvlan: Macvlan is a Linux network driver. Pods have their own unique Mac and Ip address, connected directly the physical (layer 2) network.
multus: Multus is a meta CNI plugin that provides multiple network interface support to pods. For each interface Multus delegates CNI calls to secondary CNI plugins such as Calico, macvlan, etc.
custom_cni : You can specify some manifests that will be applied to the clusters to bring you own CNI and use non-supported ones by Kubespray. See tests/files/custom_cni/README.md and tests/files/custom_cni/values.yamlfor an example with a CNI provided by a Helm Chart.

The network plugin to use is defined by the variable kube_network_plugin. There is also an option to leverage built-in cloud provider networking instead. See also Network checker.