This will set up a highly available etcd cluster, and a single-master Kubernetes cluster using kubeadm
- 3 etcd nodes, each etcd has 2 CPUs and 2 GBs RAM
- 1 master has 2 CPUs and 2 GBs RAM
- 1 worker has 2 CPUs and 2 GBs RAM
Before you want to try this on your local, here are requirements
- Directly download Virtualbox and install, or use homebrew
brew install --cask virtualbox - Vagrant
brew install --cask vagrant - (Optional) Vagrant Manager
brew install --cask vagrant-manager
cd ~
ssh-keygen -t rsa
Remember to enter ci as suffix in they key filename: id_rsa_ci
cd setup-ha-etcd-cluster
vagrant up
Check if we can ssh to the master VM, be aware that we use ~/.ssh/id_rsa_ci key to ssh
ssh -i ~/.ssh/id_rsa_ci [email protected]
This playbook can be run in parallel with load balancer playbook
ansible-playbook -i playbook/etcd_cluster_inventory.yml playbook/etcd_cluster_playbook.yml
Check etcd cluster health
ssh -i ~/.ssh/id_rsa_ci [email protected]
export ETCD_TAG=3.5.1-0
export HOST0=172.16.3.11
docker run --rm -it \
--net host \
-v /etc/kubernetes:/etc/kubernetes k8s.gcr.io/etcd:${ETCD_TAG} etcdctl \
--cert /etc/kubernetes/pki/etcd/peer.crt \
--key /etc/kubernetes/pki/etcd/peer.key \
--cacert /etc/kubernetes/pki/etcd/ca.crt \
--endpoints https://${HOST0}:2379 endpoint health --cluster
ansible-playbook -i playbook/k8s_cluster_inventory.yml playbook/k8s_cluster_playbook.yml
Download kubeconfig to your local, and verify the Kubernetes cluster
rm -rf ~/.kube-local
mkdir ~/.kube-local
scp -i ~/.ssh/id_rsa_ci [email protected]:/home/ci/.kube/config ~/.kube-local/config
export KUBECONFIG=$HOME/.kube-local/config && kubectl get nodes