ansible批量生产kerberos票据,并批量分发到所有其他主机脚本

  • name: Configure Kerberos for Hadoop Users

hosts: hadoop_servers

become: no

gather_facts: no

vars:

kerberos_server: hadoop1.xuexi.com

keytab_file_path: /home/hadoop/keys/hadoop.keytab

principals:

  • nn/

  • dn/

  • yarn/

  • starrock/

tasks:

  • name: Ensure key directory exists

ansible.builtin.file:

path: /home/hadoop/hxy

state: directory

mode: '0755'

  • name: Create Kerberos principals

ansible.builtin.command: >

kadmin.local -q 'addprinc -randkey { { item }}{ { inventory_hostname }}@XUEXI.COM'

register: addprinc_output

ignore_errors: yes

delegate_to: "{ { kerberos_server }}"

loop: "{ { principals }}"

loop_control:

extended: yes # Ensure extended loop variables are available

  • name: Check principal creation status

ansible.builtin.fail:

msg: "Failed to create principal for { { item.item }}: { { item.stderr }}"

when: "'Principal already exists' not in item.stderr and item.rc != 0"

loop: "{ { addprinc_output.results }}"

loop_control:

label: "{ { item.item }}{ { inventory_hostname }}@XUEXI.COM"

  • name: Generate keytab file for each principal

ansible.builtin.command: >

kadmin.local -q 'xst -k { { keytab_file_path }}.tmp -norandkey { { item }}{ { inventory_hostname }}@XUEXI.COM'

register: xst_output

delegate_to: "{ { kerberos_server }}"

loop: "{ { principals }}"

when: "'Principal already exists' in (addprinc_output.results | selectattr('item', 'equalto', item) | first).stderr or (addprinc_output.results | selectattr('item', 'equalto', item) | first).rc == 0"

Note: The above when condition is simplified and may need adjustment.

It assumes that if 'Principal already exists', it's okay to proceed.

However, a more robust solution would involve tracking success/failure per principal.

  • name: Move keytab file to final location (on Kerberos server)

ansible.builtin.command: mv { { keytab_file_path }}.tmp { { keytab_file_path }}

delegate_to: "{ { kerberos_server }}"

when: xst_output is changed # This might not be perfect, as 'changed' depends on file existence, not Kerberos operation.

  • name: Fetch the keytab file to the control machine

ansible.builtin.fetch:

src: "{ { keytab_file_path }}"

dest: "./hadoop.keytab"

flat: yes

delegate_to: "{ { kerberos_server }}"

run_once: yes # Ensure this task runs only once.

  • name: Distribute keytab files to each target host

ansible.builtin.copy:

src: ./hadoop.keytab

dest: /data1/tmp/hadoop.keytab

loop: "{ { groups'hadoop_servers' }}"

delegate_to: "{ { item }}"

  • name: Clean up local keytab file

ansible.builtin.file:

path: ./hadoop.keytab

state: absent

run_once: yes

相关推荐
屋外雨大,惊蛰出没6 小时前
深入浅出Spring Boot
java·spring boot·ioc·aop
Zella折耳根6 小时前
复习篇-继承和接口
java·开发语言·python
程序员二叉6 小时前
【JVM】OOM详解+JVM参数+FullGC排查+CPU飙高+死锁+内存泄漏+命令大全
java·开发语言·jvm·面试
云烟成雨TD6 小时前
Spring AI 1.x 系列【47】 MCP Annotations 模块
java·人工智能·spring
不知名的老吴6 小时前
线程的生命周期之线程同步
java·开发语言·jvm
协享科技6 小时前
Spring Boot 与 Go 双服务架构实践:从单体拆分到通信设计
java·人工智能·spring boot·后端·架构·golang·ai编程
码语智行7 小时前
地图上图、空间拓扑查询示例
java·arcgis
程序员黑豆7 小时前
AI全栈开发 - Java:变量
java·前端·ai编程
我是一颗柠檬7 小时前
【Java项目技术亮点】分库分表+数据路由策略:单表5000万后的架构升级方案
java·开发语言·分布式·架构