部署大模型的granfa监控

一、部署文件内容

1、docker-compose-prometheus.yml文件

version: '2'

networks:
    monitor:
        driver: bridge

services:
    prometheus:
        image: docker-0.unsee.tech/prom/prometheus
        container_name: prometheus
        hostname: prometheus
        restart: always
        volumes:
            - /home/kekaoxing/Monitoring/prometheus/prometheus.yml:/etc/prometheus/prometheus.yml
            - /home/kekaoxing/Monitoring/prometheus/node_down.yml:/etc/prometheus/node_down.yml
        ports:
            - "9090:9090"
        networks:
            - monitor


    grafana:
        image: docker-0.unsee.tech/grafana/grafana
        container_name: grafana
        hostname: grafana
        restart: always
        ports:
            - "3000:3000"
        networks:
            - monitor

    node-exporter:
        image: quay.io/prometheus/node-exporter
        container_name: node-exporter
        hostname: node-exporter
        restart: always
        ports:
            - "9100:9100"
        networks:
            - monitor
    cadvisor:
        image: docker-0.unsee.tech/lagoudocker/cadvisor:v0.37.0
       #image:swr.cn-north-4.myhuaweicloud.com/ddn-k8s/gcr.io/cadvisor/cadvisor:v0.51.0-linuxarm64         ##arm架构的可以用下面这个镜像  
        container_name: cadvisor
        hostname: cadvisor
        restart: always
        volumes:
            - /:/rootfs:ro
            - /var/run:/var/run:rw
            - /sys:/sys:ro
            - /var/lib/docker/:/var/lib/docker:ro
            - /sys/fs/cgroup:/sys/fs/cgroup:ro
        ports:
            - "8088:8088"
        networks:
            - monitor

2、node_down.yml

groups:
- name: node_down
  rules:
  - alert: InstanceDown
    expr: up == 0
    for: 1m
    labels:
      user: test
    annotations:
      summary: "Instance {{ $labels.instance }} down"
      description: "{{ $labels.instance }} of job {{ $labels.job }} has been down for more than 1 minutes."

3、prometheus.yml

global:
  scrape_interval: 5s
  evaluation_interval: 30s

alerting:
  alertmanagers:
  - static_configs:
    - targets: ['10.6.81.21:9093']

rule_files:
  - "node_down.yml"

scrape_configs:

  - job_name: vllm
    static_configs:
      - targets:
          - '10.6.81.21:7814'

二、启动docker容器

1、使用docker-compose启动

docker-compose -f docker-compose-prometheus.yml up -d

三、访问

1、prometheus界面

浏览器访问http://10.6.81.21:9090/targets

四、配置grafana

1、浏览器访问http://10.6.81.21:3000/

账号密码为admin/admin 登录修改密码为123456

2、配置prometheus连接源

3、保存测试

其他选项保持默认，保存测试，显示successfully即为成果

4、导入dashboard

5、dashboard部署成功