只是记录一下:
#! /usr/bin/env python3 import os import sys import sched import time import subprocess import configparser import json import requests import logging import socket # 初始化sched模块的scheduler类 # 第一个参数是一个可以返回时间戳的函数,第二个参数可以在定时未到达之前阻塞。 schedule = sched.scheduler(time.time, time.sleep) device_ip = '' remote_url = '' period = 0 program = '' #初始化日志处理部分 logger = logging.getLogger(__name__) logger.setLevel(level = logging.INFO) handler = logging.FileHandler("/var/log/monitor_push.log") handler.setLevel(logging.INFO) formatter = logging.Formatter('%(asctime)s - %(name)s - %(levelname)s - %(message)s') handler.setFormatter(formatter) logger.addHandler(handler) # 获取进程 def get_process_info(): process_datas = os.popen('ps aux').readlines() process_datas = process_datas[1:] process_info = '' body_cpu = '' body_mem = '' for process_data in process_datas: process_data = process_data.split() #process_msg={'USER':ans[0],'PID':int(ans[1]),'%CPU':float(ans[2]),'%MEM':float(ans[3]),'COMMAND':ans[-1]} #tags={'process_id':int(ans[1]), 'process_name':ans[-1], 'user':ans[0]} process_name = process_data[-1] if len(process_name) > 60: process_name = process_name[ : 31] + '....' + process_name[len(process_name) - 30 : ] body_cpu += 'process_cpu{user="' + process_data[0] + '",process_id="' + process_data[1] + '",metric="' + str(float(process_data[2])) + '",process_name="' + process_name + '"\n' body_mem += 'process_mem{user="' + process_data[0] + '",process_id="' + process_data[1] + '",metric="' + str(float(process_data[3])) + '",process_name="' + process_name + '"\n' head_cpu = '# HELP process_cpu This is process_cpu' + '\n' + '# TYPE process_cpu gauge' + '\n' head_mem = '# HELP process_mem This is process_mem' + '\n' + '# TYPE process_mem gauge' + '\n' process_info = head_cpu + body_cpu + head_mem + body_mem return process_info # 被周期性调度触发的函数 def execute_command_node(cmd, inc): try: url = "http://localhost:9100/metrics" # 打开请求,获取对象 response = requests.get(url) # 打印Http状态码 # 读取服务器返回的数据,对HTTPResponse类型数据进行读取操作 the_page = str(response.text) # 获取进程信息,并追加到the_page后面 process_info = get_process_info() the_page += process_info # json.dumps()用于将字典形式的数据转化为字符串,json.loads()用于将字符串形式的数据转化为字典 json_str = json.dumps({'target_ip': device_ip,'program': program,'metrics_str': the_page,'period': int(period)}) headers = {"Content-type": "application/json","Accept": "*/*"} response = requests.post(remote_url_prometheus,data=json_str,headers=headers) logger.info('send a metric batch success.') except Exception as e: logger.info('send message error ',e) # 中文编码格式打印数据 schedule.enter(inc, 0, execute_command_node, (cmd, inc)) def main(cmd, inc=60): # enter四个参数分别为:间隔事件、优先级(用于同时间到达的两个事件同时执行时定序)、被调用触发的函数, # 给该触发函数的参数(tuple形式) schedule.enter(0, 0, execute_command_node, (cmd,inc)) schedule.run() def get_ip(): p = subprocess.Popen("hostname -I", shell=True, stdout=subprocess.PIPE) data = p.stdout.read() # 获取命令输出内容 data = str(data,encoding = 'UTF-8') # 将输出内容编码成字符串 ip_list = data.split(' ') # 用空格分隔输出内容得到包含所有IP的列表 if "\n" in ip_list: # 发现有的系统版本输出结果最后会带一个换行符 ip_list.remove("\n") return ip_list if __name__ == '__main__': file = os.path.abspath(os.path.join(os.getcwd(),'config.ini')) cf = configparser.ConfigParser() cf.read(file,encoding='utf-8') remote_url_prometheus = f.get('remote','url_prometheus_data') #remote_url_process = cf.get('remote','url_structure_data') prefix = cf.get('local','ip_prefix') period = cf.get('local','period') program = cf.get('local','program') device_ips = get_ip() for ip_addr in device_ips: if ip_addr.startswith(prefix): device_ip = ip_addr break; if device_ip is '': print('no ip found') sys.exit(1) pid=os.fork() if pid != 0: os._exit(0) else: main("",int(period))