# cluster-backup playbook
# read the guide: 'op/cluster_restore.md'

- hosts:
  - localhost
  tasks:
  # step1: find a healthy member in the etcd cluster
  - name: set NODE_IPS of the etcd cluster
    set_fact: NODE_IPS="{% for host in groups['etcd'] %}{{ host }} {% endfor %}"

  - name: get etcd cluster status
    shell: 'for ip in {{ NODE_IPS }};do \
              ETCDCTL_API=3 {{ base_dir }}/bin/etcdctl \
              --endpoints=https://"$ip":2379 \
              --cacert={{ cluster_dir }}/ssl/ca.pem \
              --cert={{ cluster_dir }}/ssl/etcd.pem \
              --key={{ cluster_dir }}/ssl/etcd-key.pem \
              endpoint health; \
            done'
    register: ETCD_CLUSTER_STATUS
    ignore_errors: true

  - debug: var="ETCD_CLUSTER_STATUS"

  - name: get a running ectd node
    shell: 'echo -e "{{ ETCD_CLUSTER_STATUS.stdout }}" \
             "{{ ETCD_CLUSTER_STATUS.stderr }}" \
             |grep "is healthy"|sed -n "1p"|cut -d: -f2|cut -d/ -f3'
    register: RUNNING_NODE

  - debug: var="RUNNING_NODE.stdout"

  - name: get current time
    shell: "date +'%Y%m%d%H%M'"
    register: timestamp

  # step2: backup data on the healthy member
  - name: make a backup on the etcd node
    shell: "mkdir -p /etcd_backup && cd /etcd_backup && \
        ETCDCTL_API=3 {{ bin_dir }}/etcdctl snapshot save snapshot_{{ timestamp.stdout }}.db"
    args:
      warn: false
    delegate_to: "{{ RUNNING_NODE.stdout }}"

  - name: fetch the backup data
    fetch:
      src: /etcd_backup/snapshot_{{ timestamp.stdout }}.db
      dest: "{{ cluster_dir }}/backup/"
      flat: yes
    delegate_to: "{{ RUNNING_NODE.stdout }}"

  - name: update the latest backup
    shell: 'cd {{ cluster_dir }}/backup/ && /bin/cp -f snapshot_{{ timestamp.stdout }}.db snapshot.db'