What is this?
kubeasz is a set of Ansible playbooks that install a full Kubernetes cluster on your own Linux servers, with docs explaining how the components interact. It is packaged to install cleanly from inside mainland China's network.
So is kubeasz a hosted service, or scripts I run on my own servers?
Scripts you run yourself: Ansible playbooks execute from your admin machine against your servers, and the resulting cluster is fully self-managed on your own hardware.
How do I get from zero to a working cluster?
Follow the numbered install guide: plan the cluster, create certificates, deploy the etcd cluster, the container runtime, master and worker nodes, then the cluster network and plugins. A single-machine All-inOne deployment serves as a quick test environment.
What kind of machines can it install onto?
Target nodes must be Linux distros running systemd: the README lists Ubuntu 22.04/24.04, RHEL 8/9, Debian 12/13, Rocky and Alma 8/9, Fedora, openEuler, Kylin V10 and more, and says most systemd distros work.
Which Kubernetes versions will I end up with?
Current releases (kubeasz 3.6.7 through 3.7.0) target Kubernetes 1.33 to 1.36, so you get recent cluster versions rather than legacy ones.
Do I need to be a Kubernetes expert to drive it?
Linux administration experience is a must; the README recommends further reading such as kubernetes-the-hard-way and feisky's Kubernetes guide, and there is an active WeChat user group for practical questions.
Can I choose the pod network plugin for the cluster?
Yes — kubeasz supports flannel, calico and cilium as CNI options, and the install guide includes a dedicated cluster-network step.
What can it deploy besides the core cluster?
Common plugins like DNS, headlamp, metrics-server, prometheus and efk logging, plus guides for ecosystem tools such as harbor, helm, jenkins, gitlab and argocd.
Does it help with day-2 operations after install?
Yes — the usage guide covers managing node, master and etcd members, upgrading the cluster and backup/restore, plus guided experiments with NetworkPolicy, RollingUpdate and HPA.
Why does the project stress that installs are unaffected by China's network?
Many Kubernetes components and images live on endpoints that are slow or blocked from mainland China; kubeasz is packaged so a cluster install completes without depending on them, which is a core selling point of the project.
What are similar projects?
kubespray, rook, netboot.xyz, ansible, charts, tilt, kubernetes-the-hard-way, kustomize
Which topics is it in?
etcd, jinja, ansible, k8s, infrastructure-as-code, backup, deployment, containers, sysadmin, orchestration, kubernetes, documentation
