Skip to content

Latest commit

 

History

History
133 lines (101 loc) · 2.64 KB

README.md

File metadata and controls

133 lines (101 loc) · 2.64 KB

kubernetes 1.30 on nobel

prepare

# 必要なライブラリ
ansible
terraform
# Terraform 対象サーバーで実行
$ echo 'security_driver = "none"' | sudo tee /etc/libvirt/qemu.conf > /dev/null
$ sudo systemctl restart libvirtd

# Terraform 実行サーバーで実行
$ sudo apt update -y \
  sudo apt install -y mkisofs

# Terraform 実行サーバーで実行（定期的実行しておく）
$ wget -O terraform/modules/common/noble-server-cloudimg-amd64.img https://cloud-images.ubuntu.com/noble/current/noble-server-cloudimg-amd64.img

# Ansible 実行サーバーで実行（Mitogenインストール）
$ curl -Lo /tmp/mitogen-0.3.8.tar.gz https://files.pythonhosted.org/packages/source/m/mitogen/mitogen-0.3.8.tar.gz
  sudo tar zxvf /tmp/mitogen-0.3.8.tar.gz -C /opt/
  rm -f /tmp/mitogen-0.3.8.tar.gz

Usage

  • haproxy と keepalived の設定ファイルの Jinja テンプレートを生成する
$ cd ./ansible/files/lb/config_gen
$ rm -rf .venv && \
  python -m venv .venv && \
  source .venv/bin/activate && \
  pip install -r requirements.txt && \
  python haproxy.py && \
  python keepalived.py

ansible/vars_files/ 配下の変数を適宜修正。

terraform/env/main.tf の変数は適宜修正。

実行

$ ./create-k8s.sh

After running recreate-k8s.sh.

  • Login to argocd deployed as a sample
kubectl apply -f - <<EOF
apiVersion: v1
kind: Service
metadata:
  name: argocd-lb
  namespace: argocd
spec:
  type: NodePort
  ports:
    - port: 443
      targetPort: server
      nodePort: 30001
      protocol: TCP
  selector:
    app.kubernetes.io/instance: argocd
    app.kubernetes.io/name: argocd-server
EOF

  • Login

    https://192.168.11.161

    admin:admin

    Image

Env

  • Ubuntu 24.04
  • kubernetes 1.30
  • Containerd
  • Cilium 1.16.1
  • KVM
  • Haproxy + Keepalived

Host

hostname IP
k8s-api 192.168.11.130
k8s-lb-1 192.168.11.131
k8s-lb-2 192.168.11.132
k8s-cp-1 192.168.11.141
k8s-cp-2 192.168.11.142
k8s-cp-3 192.168.11.143
k8s-wk-1 192.168.11.151
k8s-wk-2 192.168.11.152
k8s-wk-3 192.168.11.153
argocd 192.168.11.161

Time of recreate_cluster

オンプレミス

Execution time: 257 seconds

GKE

Apply complete! Resources: 3 added, 0 changed, 0 destroyed.
Total Execution time: 664 seconds

$ terraform state list
google_compute_network.default
google_compute_subnetwork.default
google_container_cluster.default

Clean Up

$ terraform -chdir=./terraform/env destroy -auto-approve -input=false