Ansible Playbooks基本使用

你将学到什么

如何使用playbook
如何编写playbook
如何使用roles

PlayBook使用

基础环境

### 64 位 Ubuntu 16.04 LTS，创建CentOS LXC容器web模拟托管节点
# ssh-keygen -t rsa
# apt-get install lxc
# apt-get install yum
# lxc-create -n centos -t centos -- -R 7
### 修改centos模板root密码
# chroot /var/lib/lxc/centos/rootfs passwd
# lxc-copy -n centos -N web -B aufs -s
# lxc-start -n web -d
### 进入容器
# lxc-console -n web
### 下面命令都在容器中执行，修改IP地址为10.0.3.200
# vi ifcfg-eth0
DEVICE=eth0
BOOTPROTO=static
ONBOOT=yes
HOSTNAME=centos
NM_CONTROLLED=no
TYPE=Ethernet
NAME=eth0
IPADDR=10.0.3.200
NETMASK=255.255.255.0
GATEWAY=10.0.3.1
DNS1=114.114.114.114

简单的playbook

# mkdir playbook
# cd playbook
# vim hosts
[web]
192.168.124.240

# vim site.yml
- name: Sample
  hosts: web
  # 收集host facts信息
  gather_facts: True
  tasks:
    # 在ansible托管节点上生成sample.txt文件
    - name: Web
      command: /bin/sh -c "echo 'web' > ~/sample.txt"

    # 在ansible控制主机上生成sample.txt文件
    - name: Local Web
      local_action: command /bin/sh -c "echo 'local web' > ~/sample.txt"

执行playbook

# ansible-playbook -i hosts site.yml

样例playbook

下载样例

### 在主机中下在ansible样例
$ git clone https://github.com/ansible/ansible-examples.git

修改样例配置文件

$ cd ansible-examples/tomcat-standalone
$ vim hosts
[tomcat-servers]
10.0.3.200
### 配置ssh登入密码
$ vim group_vars/tomcat-servers
# Here are variables related to the Tomcat installation

http_port: 8080
https_port: 8443

# This will configure a default manager-gui user:

admin_username: admin
admin_password: 123456

ansible_ssh_pass: 123456

执行playbook

### 出错就反复执行，不过要加上出错提示中的--limit @/home/ubuntu/ansible-examples/tomcat-standalone/site.retry参数
# ansible-playbook -i hosts site.yml

出错处理

问题1

TASK [selinux : Install libselinux-python] *************************************
fatal: [10.0.3.200]: FAILED! => {"changed": false, "failed": true, "msg": "Failure talking to yum: Cannot retrieve metalink for repository: epel/x86_64. Please verify its path and try again"}
    to retry, use: --limit @/home/ubuntu/ansible-examples/tomcat-standalone/site.retry

解决办法

### 在容器中执行一遍yum update更新下源，就是更新下缓存，不需要安装软件

问题2

TASK [tomcat : insert firewalld rule for tomcat http port] *********************
fatal: [10.0.3.200]: FAILED! => {"changed": false, "failed": true, "msg": "firewalld and its python 2 module are required for this module"}

RUNNING HANDLER [tomcat : restart tomcat] **************************************
    to retry, use: --limit @/home/ubuntu/ansible-examples/tomcat-standalone/site.retry

解决办法

### 为容器安装firewalld
# yum search firewalld |grep python
python-firewall.noarch : Python2 bindings for firewalld
# yum install python-firewall.noarch
# systemctl enable firewalld
# systemctl start firewalld

roles使用

roles标准结构

# tree ansible-sshd/
ansible-sshd/
├── CHANGELOG
├── defaults
│   └── main.yml
├── handlers
│   └── main.yml
├── LICENSE
├── meta
│   ├── 10_top.j2
│   ├── 20_middle.j2
│   ├── 30_bottom.j2
│   ├── main.yml
│   ├── make_option_list
│   ├── options_body
│   └── options_match
├── README.md
├── tasks
│   └── main.yml
├── templates
│   └── sshd_config.j2
├── tests
│   ├── inventory
│   ├── roles
│   │   └── ansible-sshd -> ../../.
│   └── test.yml
├── Vagrantfile
└── vars
    ├── Amazon.yml
    ├── Archlinux.yml
    ├── Debian_8.yml
    ├── Debian.yml
    ├── default.yml
    ├── Fedora.yml
    ├── FreeBSD.yml
    ├── OpenBSD.yml
    ├── RedHat_6.yml
    ├── RedHat_7.yml
    ├── Suse.yml
    ├── Ubuntu_12.yml
    ├── Ubuntu_14.yml
    └── Ubuntu_16.yml

目录名	说明
defaults	为当前角色设定默认变量时使用此目录，应当包含一个main.yml文件
handlers	此目录中应当包含一个main.yml文件，用于定义此角色用到的各handler，在handler中使用include包含的其它的handler文件也应该位于此目录中
meta	应当包含一个main.yml文件，用于定义此角色的特殊设定及其依赖关系
tasks	至少应该包含一个名为main.yml的文件，其定义了此角色的任务列表，此文件可以使用include包含其它的位于此目录中的task文件
templates	template模块会自动在此目录中寻找Jinja2模板文件
vars	定义当前角色使用的变量
files	存放由copy或script等模块调用的文件
tests	在playbook中角色的使用样例

roles使用

# cat ansible-sshd/tests/test.yml
---
- hosts: localhost
  become: true
  roles:
  - ansible-sshd
# cd ansible-sshd/tests/
# ansible-playbook test.yml

roles的任务执行顺序

### 首先执行meta下的main.yml文件内容
### 然后执行tasks下的main.yml文件内容

Ansible Playbooks高级使用

文件操作

文件创建

file

用于设置文件/链接/目录的属性，或者删除文件/链接/目录

### state如果是directory当目录不存在时会自动创建；如果是file当文件不存在时不会自动创建
- name: Create log dir
  file:
    path: "{{ item.src }}"
    state: directory
  with_items: "{{ log_dirs }}"
  when: is_metal | bool
  tags:
    - common-log

- name: Mask lxc-net systemd service
  file:
    src: /dev/null
    path: /etc/systemd/system/lxc-net.service
    state: link
  when:
    - ansible_service_mgr == 'systemd'
  tags:
    - lxc-files
    - lxc-net

修改文件

lineinfile

用于检测文件是否存在特殊行或者使用后端正则表达式来替换匹配到的特殊行

- name: Extra lxc config
  lineinfile:
    dest: "/var/lib/lxc/{{ inventory_hostname }}/config"
    line: "{{ item.split('=')[0] }} = {{ item.split('=', 1)[1] }}"
    insertafter: "^{{ item.split('=')[0] }}"
    backup: "true"
  with_items: "{{ extra_container_config | default([]) }}"
  delegate_to: "{{ physical_host }}"
  register: _ec
  when: not is_metal | bool
  tags:
    - common-lxc

replace

lineinfile的多行匹配版本，此模块会在文件中插入一段内容，并在内容开始和结束位置设置标签，后续可以使用标签可以对此块内容进行操作

### 在ml2_conf.ini文件的[ml2]和[ml2_type_vlan]字段之间插入一段内容
- name: Enable ovn in neutron-server
  replace:
    dest: "{{ node_config_directory }}/neutron-server/ml2_conf.ini"
    regexp: '\[ml2\][\S\s]*(?=\[ml2_type_vlan\])'
    replace: |+
      [ml2]
      type_drivers = local,flat,vlan,geneve
      tenant_network_types = geneve
      mechanism_drivers = ovn
      extension_drivers = port_security
      overlay_ip_version = 4

      [ml2_type_geneve]
      vni_ranges = 1:65536
      max_header_size = 38

      [ovn]
      ovn_nb_connection = tcp:{{ api_interface_address }}:{{ ovn_northdb_port }}
      ovn_sb_connection = tcp:{{ api_interface_address }}:{{ ovn_sourthdb_port }}
      ovn_l3_mode = False
      ovn_l3_scheduler = chance
      ovn_native_dhcp = True
      neutron_sync_mode = repair
    backup: yes
  when:
    - action == "deploy"
    - inventory_hostname in groups['network']
  notify:
    - Restart neutron-server container

ini_file

ini后缀格式文件修改

### 设置l3_agent.ini文件[DEFAULT]字段的external_network_bridge选项值为br-ex
- name: Set the external network bridge
  vars:
      agent: "{{ 'neutron-vpnaas-agent' if enable_neutron_vpnaas | bool else 'neutron-l3-agent' }}"
  ini_file:
      dest: "{{ node_config_directory }}/{{ agent }}/l3_agent.ini"
      section: "DEFAULT"
      option: "external_network_bridge"
      value: "{{ neutron_bridge_name | default('br-ex') }}"
      backup: yes
  when:
      - action == "deploy"
      - inventory_hostname in ovn_central_address
  delegate_to: "{{ item }}"
  with_items: "{{ groups['neutron-server'] }}"
  notify:
      - Restart {{ agent }} container

assemble

将多个文件聚合成一个文件

### 将/etc/haproxy/conf.d目录下的文件内容聚合成/etc/haproxy/haproxy.cfg文件
- name: Regenerate haproxy configuration
  assemble:
    src: "/etc/haproxy/conf.d"
    dest: "/etc/haproxy/haproxy.cfg"
  notify: Restart haproxy
  tags:
    - haproxy-general-config

循环控制

with_items

标准循环，用于执行重复任务，{{ item }}类似宏展开

- name: add several users
  user:
    name: "{{ item.name }}"
    state: present
    groups: "{{ item.groups }}"
  with_items:
    - { name: 'testuser1', groups: 'wheel' }
    - { name: 'testuser2', groups: 'root' }

with_nested

嵌套循环

### 修改neutron-server组所有主机的ml2_conf.ini文件的对应字段值
- name: Enable ovn in neutron-server
  vars:
    params:
      - { section: 'ml2', option: 'type_drivers', value: 'local,flat,vlan,geneve' }
      - { section: 'ml2', option: 'tenant_network_types', value: 'geneve' }
      - { section: 'ml2', option: 'mechanism_drivers', value: 'ovn' }
      - { section: 'ml2', option: 'extension_drivers', value: 'port_security' }
      - { section: 'ml2', option: 'overlay_ip_version', value: '4' }
      - { section: 'securitygroup', option: 'enable_security_group', value: 'True' }
  ini_file:
    dest: "{{ node_config_directory }}/neutron-server/ml2_conf.ini"
    section: "{{ item[0].section }}"
    option: "{{ item[0].option }}"
    value: "{{ item[0].value }}"
    backup: yes
  when:
    - action == "deploy"
    - inventory_hostname in ovn_central_address
  delegate_to: "{{ item[1] }}"
  with_nested:
    - "{{ params }}"
    - "{{ groups['neutron-server'] }}"
  notify:
    - Restart neutron-server container

流程控制

tags

设置任务标签

tasks:
    - yum: name={{ item }} state=installed
      with_items:
         - httpd
         - memcached
      tags:
         - packages

    - template: src=templates/src.j2 dest=/etc/foo.conf
      tags:
         - configuration

### 执行playbook可以指定只执行标签对应任务或跳过标签对应任务
# ansible-playbook example.yml --tags "configuration,packages"
# ansible-playbook example.yml --skip-tags "notification"

fail_when

用来控制playbook退出

- name: Check if firewalld is installed
  command: rpm -q firewalld
  register: firewalld_check
  failed_when: firewalld_check.rc > 1
  when: ansible_os_family == 'RedHat'

pre_tasks/post_tasks

用来设置在执行roles模块之前和之后需要执行的任务

- name: Install the aodh components
  hosts: aodh_all
  gather_facts: "{{ gather_facts | default(True) }}"
  max_fail_percentage: 20
  user: root
  pre_tasks:
    - include: common-tasks/os-lxc-container-setup.yml
    - include: common-tasks/rabbitmq-vhost-user.yml
      static: no
      vars:
        user: "{{ aodh_rabbitmq_userid }}"
        password: "{{ aodh_rabbitmq_password }}"
        vhost: "{{ aodh_rabbitmq_vhost }}"
        _rabbitmq_host_group: "{{ aodh_rabbitmq_host_group }}"
      when:
        - inventory_hostname == groups['aodh_api'][0]
        - groups[aodh_rabbitmq_host_group] | length > 0
    - include: common-tasks/os-log-dir-setup.yml
      vars:
        log_dirs:
          - src: "/openstack/log/{{ inventory_hostname }}-aodh"
            dest: "/var/log/aodh"
    - include: common-tasks/mysql-db-user.yml
      static: no
      vars:
        user_name: "{{ aodh_galera_user }}"
        password: "{{ aodh_container_db_password }}"
        login_host: "{{ aodh_galera_address }}"
        db_name: "{{ aodh_galera_database }}"
      when: inventory_hostname == groups['aodh_all'][0]
    - include: common-tasks/package-cache-proxy.yml
  roles:
    - role: "os_aodh"
      aodh_venv_tag: "{{ openstack_release }}"
      aodh_venv_download_url: "{{ openstack_repo_url }}/venvs/{{ openstack_release }}/{{ ansible_distribution | lower }}/aodh-{{ openstack_release }}-{{ ansible_architecture | lower }}.tgz"
    - role: "openstack_openrc"
      tags:
        - openrc
    - role: "rsyslog_client"
      rsyslog_client_log_rotate_file: aodh_log_rotate
      rsyslog_client_log_dir: "/var/log/aodh"
      rsyslog_client_config_name: "99-aodh-rsyslog-client.conf"
      tags:
        - rsyslog
  vars:
    is_metal: "{{ properties.is_metal|default(false) }}"
    aodh_rabbitmq_userid: aodh
    aodh_rabbitmq_vhost: /aodh
    aodh_rabbitmq_servers: "{{ rabbitmq_servers }}"
    aodh_rabbitmq_port: "{{ rabbitmq_port }}"
    aodh_rabbitmq_use_ssl: "{{ rabbitmq_use_ssl }}"
  tags:
    - aodh

主机路由

delegate_to

可以将当前任务放到其他hosts上执行

### 这是一段在容器中执行的playbook的一部分，这时候需要检测容器所在的宿主机上的对应目录是否存在，这时候就需要用到委托来跳出当前容器到宿主机上执行当前任务
- name: Ensure mount directories exists
  file:
    path: "{{ item['mount_path'] }}"
    state: "directory"
  with_items:
    - "{{ lxc_default_bind_mounts | default([]) }}"
    - "{{ list_of_bind_mounts | default([]) }}"
  delegate_to: "{{ physical_host }}"
  when:
    - not is_metal | bool
  tags:
    - common-lxc

local_action

将任务放在ansible控制主机（运行ansible-playbook的主机）上执行

- name: Check if the git cache exists on deployment host
  local_action:
    module: stat
    path: "{{ repo_build_git_cache }}"
  register: _local_git_cache
  when: repo_build_git_cache is defined

用户和用户组控制

group

创建用户组

### 创建系统管理员组haproxy，present表示不存在创建，absent表示存在删除
- name: Create the haproxy system group
  group:
    name: "haproxy"
    state: "present"
    system: "yes"
  tags:
    - haproxy-group

user

创建用户

### 创建haproxy：haproxy用户，并创建home目录
- name: Create the haproxy system user
  user:
    name: "haproxy"
    group: "haproxy"
    comment: "haproxy user"
    shell: "/bin/false"
    system: "yes"
    createhome: "yes"
    home: "/var/lib/haproxy"
  tags:
    - haproxy-user

其他

authorized_key

添加用户的SSH认证key

- name: Create authorized keys file from host vars
  authorized_key:
    user: "{{ repo_service_user_name }}"
    key: "{{ hostvars[item]['repo_pubkey'] | b64decode }}"
  with_items: "{{ groups['repo_all'] }}"
  when: hostvars[item]['repo_pubkey'] is defined
  tags:
    - repo-key
    - repo-key-store

slurp

用来读取远程主机上文件内容是base64加密的文件

### 读取id_rsa.pub文件的内容，并设置到变量repo_pub中
- name: Get public key contents and store as var
  slurp:
    src: "{{ repo_service_home_folder }}/.ssh/id_rsa.pub"
  register: repo_pub
  changed_when: false
  tags:
    - repo-key
    - repo-key-create

web访问，类似执行curl命令

- name: test proxy URL for connectivity
  uri:
    url: "{{ repo_pkg_cache_url }}/acng-report.html"
    method: "HEAD"
  register: proxy_check
  failed_when: false
  tags:
    - common-proxy

wait_for

等待一个端口变得可用或者等待一个文件变得可用

- name: Wait for container ssh
  wait_for:
    port: "22"
    delay: "{{ ssh_delay }}"
    search_regex: "OpenSSH"
    host: "{{ ansible_host }}"
  delegate_to: "{{ physical_host }}"
  register: ssh_wait_check
  until: ssh_wait_check | success
  retries: 3
  when:
    - (_mc is defined and _mc | changed) or (_ec is defined and _ec | changed)
    - not is_metal | bool
  tags:
    - common-lxc

command

执行shell命令

### ignore_errors为true表示命令执行出错也不会退出playbook
- name: Check if clean is needed
  command: docker exec openvswitch_vswitchd ovs-vsctl br-exists br-tun
  register: result
  ignore_errors: True

切换用户

### 使用become会先切换成apache用户，再执行command命令，默认become_user用户为root（如果你ansible配置的就是root用户的免密码登入那就不需要become了）
- name: Run a command as the apache user
  command: somecommand
  become: true
  become_user: apache

检测链表是否为空

### pip_wheel_install为链表变量
- name: Install wheel packages
  shell: cd /tmp/wheels && pip install {{ item }}*
  with_items:
    - "{{ pip_wheel_install | default([]) }}"
  when: pip_wheel_install > 0

Ansible Jinja2使用

常用方法

ternary

根据结果的真假来决定返回值

- name: Set container backend to "dir" or "lvm" based on whether the lxc VG was found
  set_fact:
     lxc_container_backing_store: "{{ (vg_result.rc != 0) | ternary('dir', 'lvm') }}"
  when: vg_result.rc is defined
  tags:
    - lxc-container-vg-detect

vg_result.rc不为0返回dir，否则返回lvm

if语法

根据结果的真假来决定返回值

- name: Set the external network bridge
  vars:
    agent: "{{ 'neutron-vpnaas-agent' if enable_neutron_vpnaas | bool else 'neutron-l3-agent' }}"
  ini_file:
    dest: "{{ node_config_directory }}/{{ agent }}/l3_agent.ini"
    section: "DEFAULT"
    option: "external_network_bridge"
    value: "{{ neutron_bridge_name | default('br-ex') }}"
    backup: yes
  when:
    - action == "deploy"
    - inventory_hostname in ovn_central_address
  delegate_to: "{{ item }}"
  with_items: "{{ groups['neutron-server'] }}"
  notify:
  - Restart {{ agent }} container

when中使用jinja2

when表达式中不建议直接使用{{}}的方式来获取变量值，如果变量是字符串可以使用管道操作| string来获取变量值

- name: Checking free port for OVN
  vars:
    service: "{{ neutron_services[item.name] }}"
  wait_for:
    host: "{{ hostvars[inventory_hostname]['ansible_' + api_interface]['ipv4']['address'] }}"
    port: "{{ item.port }}"
    connect_timeout: 1
    state: stopped
  when:
    - container_facts[ item.facts | string ] is not defined
    - service.enabled | bool
    - service.host_in_groups | bool
  with_items:
    - { name: "ovn-nb-db-server", port: "{{ ovn_northdb_port }}", facts: "ovn_nb_db" }
    - { name: "ovn-sb-db-server", port: "{{ ovn_sourthdb_port }}", facts: "ovn_sb_db" }

Ansible基本使用

你将学到什么

如何配置ansible运行环境
如何执行ansible命令
如何配置Inventory

环境

角色	操作系统	网络地址
管理主机	ubuntu 14.04 TLS	192.168.200.250
托管节点	ubuntu 16.04 TLS	192.168.200.11 192.168.200.12

管理主机配置

安装ansible

$ sudo apt-get install software-properties-common
$ sudo apt-add-repository ppa:ansible/ansible
$ sudo apt-get update
$ sudo apt-get install ansible

设置托管节点

### 编辑配置文件添加托管节点网络地址
# vim /etc/ansible/hosts
...
[webservers]
192.168.200.11
[compute]
192.168.200.12
...

生成ssh公钥

# ssh-keygen -t rsa
# ssh-agent bash
# ssh-add ~/.ssh/id_rsa

托管节点配置

添加管理主机公钥

# ssh-keygen -t rsa
# scp root@192.168.200.250:~/.ssh/id_rsa.pub ./
# cat id_rsa.pub >> ~/.ssh/authorized_keys
# chmod 600 ~/.ssh/authorized_keys

安装python

# apt-get install python

测试ansible命令

# ansible all -m ping
192.168.200.11 | success >> {
    "changed": false,
    "ping": "pong"
}

192.168.200.12 | success >> {
    "changed": false,
    "ping": "pong"
}
# ansible compute -m ping
192.168.200.12 | success >> {
    "changed": false,
    "ping": "pong"
}
# ansible webservers -m ping
192.168.200.11 | success >> {
    "changed": false,
    "ping": "pong"
}

# ansible all -a "/bin/echo hello"
192.168.200.12 | success | rc=0 >>
hello

192.168.200.11 | success | rc=0 >>
hello

配置Inventory

静态方式

就是前面在文件/etc/ansible/hosts中指定的主机和组的方式

动态方式

通过外部脚本获取主机列表，并按照ansible所要求的格式返回给ansilbe命令的方式。需要注意的是，用于生成JSON代码的脚本必须支持两个选项：

--list ：返回一个JSON散列/字典，它包含所管理的所有组.每个组的value应该是一个关于其包含的主机/IP哈希/字典，它可能是一个子组或者组的变量或者仅仅是一个主机/IP的列表，例如:

{
"databases"   : {
    "hosts"   : [ "host1.example.com", "host2.example.com" ],
    "vars"    : {
        "a"   : true
    }
},
"webservers"  : [ "host2.example.com", "host3.example.com" ],
"atlanta"     : {
    "hosts"   : [ "host1.example.com", "host4.example.com", "host5.example.com" ],
    "vars"    : {
        "b"   : false
    },
    "children": [ "marietta", "5points" ]
},
"marietta"    : [ "host6.example.com" ],
"5points"     : [ "host7.example.com" ]
}

--host <hostname> ：返回一条空的JSON哈希/字典，或者关于变量的JSON哈希/字典，这些变量将被用来模板或者playbooks。返回变量是可选的，如果脚本不希望这样做，返回一条空的哈希/字典即可：
```
{
"favcolor"   : "red",
"ntpserver"  : "wolf.example.com",
"monitoring" : "pack.example.com"
}
```

编写样例脚本inventory-script：

[web]
192.168.200.100

调用方式如下：

### Ansible默认通过调用脚本的--list选项来获取主机列表
# ansible web -i inventory-script -m ping

`done`

http://www.cnblogs.com/silvermagic/p/7666269.html

posted on 2018-05-14 15:55 flyoss 阅读(5990) 评论(0) 编辑收藏举报

会员力量，点亮园子希望

刷新页面返回顶部

flyoss