编写一个探测端口异常的Python脚本

在工作中需要时刻关注服务进程的健康状态或者metrics进程的健康状态，可以监控其服务的端口，下面编写一个监控kafka服务进程和其metrics进程的端口的Python脚本，直接上代码：

import socket
from prometheus_client import start_http_server,Summary
from prometheus_client import start_http_server, Gauge
from prometheus_client.core import CollectorRegistry
from prometheus_client.core import Gauge
import random
import time

# IP 地址到主机名的映射
ip_to_hostname = {
    "10.110.1.110":"hostname1",
    "10.110.1.111":"hostname2",
    "10.110.1.112":"hostname3",
}

def check_port(server, port):
    try:
        s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
        s.settimeout(1)  # 设置超时时间为1秒
        s.connect((server, port))
        s.close()
        return True
    except Exception as e:
        return False

def check_servers_ports(servers, ports):
    status = {}
    for server in servers:
        hostname = ip_to_hostname.get(server,"Unknown") #获取主机名,如果位置则默认为"Unknown"
        status[server] = {'hostname':hostname}

        for port in ports:
            status[server][port] = check_port(server, port)
    return status

gauge_metric = Gauge('kafka_server_and_monitor_port', 'Current status of kafka server and monitor', ['server', 'port', 'hostname'])

def get_kafka_server_and_monitor_status():
    servers = list(ip_to_hostname.keys())
    ports = [9092, 9100, 9990, 9308]

    port_status = check_servers_ports(servers, ports)

    print("# HELP kafka_server_and_monitor_port Current status of kafka server and monitor.")
    print("# TYPE kafka_server_and_monitor_port Gauge")

    for server in port_status:
        for port, status in port_status[server].items():
            if port == 'hostname': #跳过主机名
                continue
            gauge_metric.labels(server=server, port=str(port),hostname=port_status[server]['hostname']).set(1 if status else 0)

if __name__ == "__main__":
    registry = CollectorRegistry(auto_describe=False)
    registry.register(gauge_metric) 
    start_http_server(1333,registry=registry)
    while True:
        get_kafka_server_and_monitor_status()
        time.sleep(60)

使用nohup在后台启动进程：

nohup python3.6 kafka_port_monitor.py >/dev/null 2>&1 &

最后查看监控指标是否正常输出

curl  http://localhost:1334/metrics