ansible的脚本中我们可能会要启动某项服务并等待，直到服务启动起来，然后再进行下一步
这个非常重要，我们举例来实验一下：
1- hosts: localhost
2 vars:
3 local__service: ssh
4 tasks:
5 - block:
6 - name: "Stop {{ local__service }} service"
7 systemd:
8 service: "{{ local__service }}"
9 state: stopped
10
11 - name: "Populate ansible_facts.services variable"
12 ansible.builtin.service_facts:
13
14 - name: "{{ local__service }} state will be stopped as expected"
15 assert:
16 that:
17 ansible_facts.services[local__service].state == 'stopped'
18
19 - name: "Start {{ local__service }} service"
20 systemd:
21 service: "{{ local__service }}"
22 state: started
23
24 - name: "Registered {{ local__service }} state will still be stopped as it was not refreshed"
25 assert:
26 that:
27 ansible_facts.services[local__service].state == 'stopped'
28
29 - name: "Refresh ansible_facts.services variable"
30 ansible.builtin.service_facts:
31
32 - name: "{{ local__service }} state will be running as expected"
33 assert:
34 that:
35 ansible_facts.services[local__service].state == 'running'
上面的 playbook 一共有7步：
1、停止 ssh 服务
2、获取 ansible_fact 变量
3、断言 ansible_facts 中的 ssh 服务是 stopped 状态
4、重新启动 ssh 服务
5、断言 ansible_facts 中的 ssh 服务仍然是 stopped 状态
6、重新获取 ansible_fact 变量
7、这时 ansible_facts 中的 ssh 服务状态刷新了，变成了 running
就是服务启动后，必须去不断刷新 ansible_facts，才能获得刷新后的状态，所以我们利用这一点，脚本如下：
1- hosts: localhost
2 vars:
3 local__service: ssh
4 tasks:
5 - block:
6 - name: "Stop {{ local__service }} service"
7 systemd:
8 service: "{{ local__service }}"
9 state: stopped
10
11 - name: "Wait until {{ local__service }} service is stopped"
12 ansible.builtin.service_facts:
13 register: temp__service_facts
14 until: temp__service_facts.ansible_facts.services[local__service].state == 'stopped'
15 retries: 20
16 delay: 2
17
18 - name: "Start {{ local__service }} service"
19 systemd:
20 service: "{{ local__service }}"
21 state: started
22
23 - name: "Wait until {{ local__service }} service is running"
24 ansible.builtin.service_facts:
25 register: temp__service_facts
26 until: temp__service_facts.ansible_facts.services[local__service].state == 'running'
27 retries: 20
28 delay: 2
上面wait until的步骤就是，注册一个临时的变量，然后不断去测试，重试20次，每次的延时是2秒。这样就可以解决了。
