kubeasz/tools/clean_one_node.yml

# 警告：此脚本将清理单个node节点，使用请详细参阅 docs/op/del_one_node.md
# 如果该节点为kube-node节点，请先执行kubectl drain
# 请三思后运行此脚本，特别的：如果有pod使用了本地存储类型，请自行判断重要性
# 参考 docs/op/del_one_node.md说明
# 使用：
# 1. 假设待删节点为 192.168.1.1，执行 sed -i 's/NODE_TO_DEL/192.168.1.1/g' tools/clean_one_node.yml
# 2. 执行 ansible-playbook /etc/ansible/tools/clean_one_node.yml

# 执行kubectl drain(节点可能是kube-node节点)
- hosts: deploy
  tasks:
  - name: 执行kubectl drain(节点可能是kube-node节点)
    shell: "{{ bin_dir }}/kubectl drain NODE_TO_DEL --ignore-daemonsets --delete-local-data"
    ignore_errors: true

# 清理 kube-node 相关服务
- hosts: NODE_TO_DEL
  tasks:
  - name: stop kube-node service
    shell: "systemctl stop kubelet kube-proxy"
    ignore_errors: true

  - name: umount kubelet 挂载的目录
    shell: "mount | grep '/var/lib/kubelet'| awk '{print $3}'|xargs umount"
    ignore_errors: true

  - name: 清理目录和文件
    file: name={{ item }} state=absent
    with_items:
    - "/var/lib/kubelet/"
    - "/var/lib/kube-proxy/"
    - "/etc/systemd/system/kubelet.service"
    - "/etc/systemd/system/kube-proxy.service"
    - "/opt/kube/kube-system/"

# 清理 kube-master 相关
- hosts: NODE_TO_DEL
  tasks:
  - name: stop and disable kube-master service
    service: name={{ item }} state=stopped enabled=no
    with_items:
    - kube-apiserver
    - kube-controller-manager
    - kube-scheduler
    ignore_errors: true

  - name: 清理目录和文件
    file: name={{ item }} state=absent
    with_items:
    - "/var/run/kubernetes"
    - "/etc/systemd/system/kube-apiserver.service"
    - "/etc/systemd/system/kube-controller-manager.service"
    - "/etc/systemd/system/kube-scheduler.service"

# 清理集群docker服务、网络相关
- hosts: NODE_TO_DEL
  tasks:
  - name: 清理kube-router相关
    shell: "{{ bin_dir }}/docker run --privileged --net=host cloudnativelabs/kube-router --cleanup-config"
    ignore_errors: true
    when: "CLUSTER_NETWORK == 'kube-router'"

  - name: stop and disable docker service
    service:
      name: docker
      state: stopped
      enabled: no
    ignore_errors: true

  # 因为calico-kube-controller使用了host网络，相当于使用了docker -net=host，需要
  # 卸载 /var/run/docker/netns/default
  - name: 卸载docker 相关fs1
    mount: path=/var/run/docker/netns/default state=unmounted

  - name: 卸载docker 相关fs2
    mount: path=/var/lib/docker/overlay state=unmounted

  - name: 清理目录和文件
    file: name={{ item }} state=absent
    with_items:
    - "/etc/cni/"
    - "/root/.kube/"
    - "/run/flannel/"
    - "/etc/calico/"
    - "/var/run/calico/"
    - "/var/lib/calico/"
    - "/var/log/calico/"
    - "/etc/cilium/"
    - "/var/run/cilium/"
    - "/sys/fs/bpf/tc/"
    - "/var/lib/cni/"
    - "/var/lib/docker/"
    - "/var/lib/kube-router/"
    - "/var/run/docker/"
    - "/etc/systemd/system/calico-node.service"
    - "/etc/systemd/system/docker.service"
    - "/etc/systemd/system/docker.service.requires/"
    - "/opt/kube/kube-system/"
    - "/etc/bash_completion.d/docker"
    ignore_errors: true

  - name: 清理 iptables
    shell: "iptables -F && iptables -X \
	&& iptables -F -t nat && iptables -X -t nat \
	&& iptables -F -t raw && iptables -X -t raw \
	&& iptables -F -t mangle && iptables -X -t mangle"

  - name: 清理网络
    shell: "ip link del docker0; \
	ip link del tunl0; \
        ip link del flannel.1; \
        ip link del cni0; \
        ip link del mynet0; \
        ip link del kube-bridge; \
        ip link del dummy0; \
        ip link del kube-ipvs0; \
        ip link del cilium_net; \
        ip link del cilium_vxlan; \
        systemctl restart networking; \
	systemctl restart network"
    ignore_errors: true

  - name: 清理calico残留路由
    shell: "for rt in `ip route|grep bird|sed 's/blackhole//'|awk '{print $1}'`;do ip route del $rt;done;"
    when: "CLUSTER_NETWORK == 'calico'"
    ignore_errors: true

# 清理etcd 集群相关
- hosts: NODE_TO_DEL
  tasks:
  - name: stop and disable etcd service
    service:
      name: etcd
      state: stopped
      enabled: no
    ignore_errors: true

  - name: 清理目录和文件
    file: name={{ item }} state=absent
    with_items:
    - "/var/lib/etcd"
    - "/etc/etcd/"
    - "/backup/k8s"
    - "/etc/systemd/system/etcd.service"

# 清理负载均衡相关
- hosts: NODE_TO_DEL
  tasks:
  - name: stop keepalived service
    shell: systemctl disable keepalived && systemctl stop keepalived
    ignore_errors: true

  - name: stop haproxy service
    shell: systemctl disable haproxy && systemctl stop haproxy
    ignore_errors: true

  - name: 清理LB 配置文件目录
    file: name={{ item }} state=absent
    with_items:
    - "/etc/haproxy"
    - "/etc/keepalived"

# 清理其他
- hosts: NODE_TO_DEL
  tasks:
  - name: stop and disable chrony
    service: name={{ item }} state=stopped enabled=no
    with_items:
    - chrony
    - chronyd
    ignore_errors: true
    tags: rm_ntp

  - name: 清理证书目录和文件
    file: name={{ item }} state=absent
    with_items:
    - "/etc/kubernetes/"
    - "{{ ca_dir }}" 
    - "/root/.kube/"
    - "/etc/docker/"

  - name: 清理自动生成的PATH
    lineinfile:
      dest: ~/.bashrc
      state: absent
      regexp: '{{ item }}'
    with_items:
    - 'kubeasz'
    - 'helm'
    - 'kubectl completion'

# 执行kubectl delete(节点可能是kube-node节点)
- hosts: deploy
  tasks:
  - name: 执行kubectl delete(节点可能是kube-node节点)
    shell: "{{ bin_dir }}/kubectl delete node NODE_TO_DEL"
    ignore_errors: true