博客
关于我
强烈建议你试试无所不能的chatGPT,快点击我
如何利用ansible callback插件对执行结果进行解析
阅读量:5856 次
发布时间:2019-06-19

本文共 8695 字,大约阅读时间需要 28 分钟。

最近在写一个批量巡检工具,利用ansible将脚本推到各个机器上执行,然后将执行的结果以json格式返回来。

 

如下所示:

# ansible node2 -m script -a /root/python/health_check.py

node2 | SUCCESS => {    "changed": true,     "rc": 0,     "stderr": "Shared connection to 192.168.244.20 closed.\r\n",     "stdout": "{'cpu_iowait': '0.00', 'swap_out': 0, 'cpu_usr': '0.00', 'cpu_idle': '100.00', 'swap_total': '1999', 'swap_used': '78', 'load_average_5': '0.11', 'mem_util': '92.0', 'uptime': '5', 'load_average_1': '0.03', 'cpu_sys': '0.00', 'mem_total': '475', 'swap_in': 0, 'load_average_15': '0.06', 'disk': ['Filesystem      Size  Used Avail Use% Mounted on\\n', '/dev/sda3        18G  8.6G  8.1G  52% /\\n', 'tmpfs           238M     0  238M   0% /dev/shm\\n', '/dev/sda1       190M   27M  154M  15% /boot\\n'], 'numa': '1'}\r\n",         "stdout_lines": [        "{'cpu_iowait': '0.00', 'swap_out': 0, 'cpu_usr': '0.00', 'cpu_idle': '100.00', 'swap_total': '1999', 'swap_used': '78', 'load_average_5': '0.11', 'mem_util': '92.0', 'uptime': '5', 'load_average_1': '0.03', 'cpu_sys': '0.00', 'mem_total': '475', 'swap_in': 0, 'load_average_15': '0.06', 'disk': ['Filesystem      Size  Used Avail Use% Mounted on\\n', '/dev/sda3        18G  8.6G  8.1G  52% /\\n', 'tmpfs           238M     0  238M   0% /dev/shm\\n', '/dev/sda1       190M   27M  154M  15% /boot\\n'], 'numa': '1'}"    ]}

 

然后将结果重定向到一个文本文件中,再通过另外一个脚本,对该文本文件进行解析汇总,最后实现的结果如下:

ip              uptime          cpu_usr         cpu_sys         cpu_iowait      cpu_idle        load_average_1  load_average_5  ...           192.168.244.30  24              0               0               6               94              0.02            0.08            ...              192.168.244.20  24              0               0               0               100             0               0.01            ...

 

但总感觉这种方式有点low,对返回结果进行解析,这似乎是一个比较普遍的需求吧?

 

没道理,官方会对这种需求视而不见的,其实,官方提供了一个callback插件,来实现回调功能,里面定义了若干场景,譬如主机不可达,执行任务失败,执行任务成功等,分别对应不同的方法,这样就可以实现在不同的场景触发不同的操作,譬如,如果执行playbook失败了就发送邮件等,执行成功了将返回的结果保存到数据库中。

 

官方给了一个样例,具体可见:

 

基于上面这个样例,自己进行了定制性开发。本来想在callback插件中实现所有功能,但callback插件调试相当麻烦,不允许使用print函数,而且如果出现问题了,譬如列表下标越界,也只是在执行ansible时给出报错信息,并没有指出具体的报错行数。

 

最后,放弃了自己ALL IN ONE的想法,只是将返回的结果解析后保存到sqlite3数据库中,后续再基于数据库中的数据进行汇总。

 

代码如下:

from __future__ import (absolute_import, division, print_function)__metaclass__ = typeimport osimport timeimport jsonimport sqlite3from ansible.module_utils._text import to_bytesfrom ansible.plugins.callback import CallbackBaseclass CallbackModule(CallbackBase):    """    logs playbook results, per host, in /var/log/ansible/hosts    """    CALLBACK_VERSION = 2.0    CALLBACK_TYPE = 'notification'    CALLBACK_NAME = 'performance_check'    CALLBACK_NEEDS_WHITELIST = False    def __init__(self):        super(CallbackModule, self).__init__()    def runner_on_failed(self, host, res, ignore_errors=False):        pass    def runner_on_ok(self, host, res):        performance_data=PerformanceData()        create_table_sql = 'CREATE TABLE performance_data(ip varchar(20) primary key, uptime varchar(20),cpu_usr DECIMAL,cpu_sys DECIMAL, cpu_iowait DECIMAL,cpu_idle DECIMAL,load_average_1 DECIMAL,load_average_5 DECIMAL,load_average_15 DECIMAL, mem_total INTEGER,mem_util DECIMAL,swap_total INTEGER,swap_used INTEGER,swap_in INTEGER,swap_out INTEGER,numa TINYINT)'                insert_sql = 'insert into performance_data values (?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?)'        insert_value = str_to_json(host,res)        performance_data.create_table(create_table_sql)        performance_data.insert_command(insert_sql,insert_value)        performance_data.quit()    def runner_on_skipped(self, host, item=None):        #self.log(host, 'SKIPPED', '...')        pass    def runner_on_unreachable(self, host, res):        #self.log(host, 'UNREACHABLE', res)        pass    def runner_on_async_failed(self, host, res, jid):        #self.log(host, 'ASYNC_FAILED', res)        pass    def playbook_on_import_for_host(self, host, imported_file):        pass    def playbook_on_not_import_for_host(self, host, missing_file):        passclass PerformanceData():    def __init__(self):        self.conn = sqlite3.connect("/tmp/data.db")        self.cursor = self.conn.cursor()    def create_table(self,create_table_sql):        self.cursor.execute(create_table_sql)    def insert_command(self,insert_sql,insert_value):        self.cursor.execute(insert_sql,insert_value)    def query(self,query_sql):        self.cursor.execute(query_sql)        results=self.cursor.fetchall()        return results    def quit(self):        self.conn.commit()        self.conn.close()def str_to_json(host,res):            result= res["stdout"].strip(" ").replace("'",'"').strip('\n').strip('"')            results= '{"'+host+'":'+result+'}'            result_with_host = json.loads(results)            value=result_with_host[host]            return (host,value['uptime'],float(value['cpu_usr']),float(value['cpu_sys']),float(value['cpu_iowait']),               float(value['cpu_idle']), float(value['load_average_1']), float(value['load_average_5']), float(value['load_average_15']),               int(value['mem_total']), float(value['mem_util']),int(value['swap_total']),int(value['swap_used']),int(value['swap_in']),               int(value['swap_out']), int(value['numa'])               )

 

这里一并附上,上述解析文本的脚本,似乎更能实现我ALL IN ONE的想法,哈哈~

#coding: utf8import re,json,sqlite3def get_ip_success():    with open(r'C:\Users\Administrator\Desktop\2.txt') as f:        ip_unreachable = []        ip_failed = []        ip_success=[]        line_num=0        for line in f.readlines():            if re.search('UNREACHABLE', line):                ip=line.split()[0]                ip_unreachable.append(ip)                flag=0            elif re.search('FAILED',line):                ip = line.split()[0]                ip_failed.append(ip)                flag=0            elif re.search('SUCCESS',line):                ip = line.split()[0]                flag=1                line_num=1            elif flag == 1 and line_num == 7:                line= line.strip(" ").replace("'",'"').strip('\n').strip('"')                stdout_lines= '{"'+ip+'":'+line+'}'                stdout_lines_with_ip = json.loads(stdout_lines)                ip_success.append(stdout_lines_with_ip)            line_num =line_num + 1    return ip_successdef os_status_generator(ip_success):    for os_status in ip_success:        for key,value in os_status.iteritems():            yield (key,value['uptime'],float(value['cpu_usr']),float(value['cpu_sys']),float(value['cpu_iowait']),               float(value['cpu_idle']), float(value['load_average_1']), float(value['load_average_5']), float(value['load_average_15']),               int(value['mem_total']), float(value['mem_util']),int(value['swap_total']),int(value['swap_used']),int(value['swap_in']),               int(value['swap_out']), int(value['numa'])               )class OsStatus():    def __init__(self,ip_success):        try:            self.conn = sqlite3.connect(":memory:")            self.cursor = self.conn.cursor()            self.cursor.execute('''CREATE TABLE os_status                     (ip varchar(20) primary key, uptime varchar(20),cpu_usr DECIMAL,cpu_sys DECIMAL,cpu_iowait DECIMAL,cpu_idle DECIMAL,                      load_average_1 DECIMAL,load_average_5 DECIMAL,load_average_15 DECIMAL,mem_total INTEGER,mem_util DECIMAL,                      swap_total INTEGER,swap_used INTEGER,swap_in INTEGER,swap_out INTEGER,numa TINYINT)''')            self.cursor.executemany("insert into os_status values (?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?)",os_status_generator(ip_success) )        except Exception as e:            print e;    def query(self,sql):        self.cursor.execute(sql)        results=self.cursor.fetchall()        column_size=len(results[0])        column_name= [column[0] for column in self.cursor.description]        for i in range(column_size):            print column_name[i].ljust(15),        print        for each_result in results:            for i in range(column_size):                print str(each_result[i]).ljust(15),            print    def quit(self):        try:            self.cursor.close()            self.conn.close()        except Exception as e:            print e;ip_success=get_ip_success()os_status=OsStatus(ip_success)sql = "select * from os_status"os_status.query(sql)

 

最后,再提一下ansible中如何开启callback插件功能,默认是关闭的。

开启两个选项:

callback_plugins   = /usr/share/ansible/plugins/callbackbin_ansible_callbacks = True

这两个是必需的,另外一个选项是

callback_whitelist = performance_check

其中,performance_check对应的是上面callback插件中定义的“CALLBACK_NAME”,

另一个相关参数是“CALLBACK_NEEDS_WHITELIST”,如果设置为False,则无需设置callback_whitelist选项,反之,则必须在callback_whitelist选项中指定“CALLBACK_NAME”。

 

转载地址:http://ojojx.baihongyu.com/

你可能感兴趣的文章
前缀(字典树)
查看>>
C语言编程好习惯(持续更新)
查看>>
HDU 4628 Pieces(状压DP)题解
查看>>
[hadoop新实战2]hadoop伪分布式安装序列(支持ubuntu和redhat)
查看>>
MFC通过sql访问excel的方法
查看>>
自定义mvc
查看>>
软工网络15团队作业4——Alpha阶段敏捷冲刺-1
查看>>
常用的js函数
查看>>
SQL优化
查看>>
android - GridView的使用
查看>>
Android 退出整个应用程序解决方案
查看>>
SQL优化:读懂explain
查看>>
Java中static关键字的几种用法
查看>>
内存四区中全局区见解
查看>>
UVA 1614 - Hell on the Markets 奇怪的股市(贪心,结论)
查看>>
angular2 学习笔记 (Pipes)
查看>>
二叉树非递归于递归遍历
查看>>
shiro标签
查看>>
Linux Nginx
查看>>
zabbix3.4.7表结构
查看>>