prometheus+grafana+cadvisor监控docker容器

1、配置CAdvisor：

CAdvisor是谷歌开发的用于分析运行中容器的资源占用和性能指标的开源工具。CAdvisor是一个运行时的守护进程，负责收集、聚合、处理和输出运行中容器的信息,他将以容器的方式运行在所有的host上。

{
  "id": "/monitor/cadvisor106",
  "role": "monitor",
  "cmd": null,
  "cpus": 0.1,
  "mem": 256,
  "disk": 0,
  "gpus": 0,
  "instances": 1,
  "constraints": [
    [
      "hostname",
      "CLUSTER",
      "10.20.31.106"
    ]
  ],
  "acceptedResourceRoles": [
    "*"
  ],
  "container": {
    "type": "DOCKER",
    "docker": {
      "forcePullImage": true,
      "image": "10.20.31.104:5000/library/google/cadvisor:latest",
      "parameters": [],
      "privileged": false
    },
    "volumes": [
      {
        "containerPath": "/etc/localtime",
        "hostPath": "/etc/localtime",
        "mode": "RO"
      },
      {
        "containerPath": "/rootfs",
        "hostPath": "/",
        "mode": "RO"
      },
      {
        "containerPath": "/var/run",
        "hostPath": "/var/run",
        "mode": "RO"
      },
      {
        "containerPath": "/sys",
        "hostPath": "/sys",
        "mode": "RO"
      },
      {
        "containerPath": "/dev/disk/",
        "hostPath": "/dev/disk/",
        "mode": "RO"
      },
      {
        "containerPath": "/var/lib/docker",
        "hostPath": "/var/lib/docker",
        "mode": "RO"
      },
      {
        "containerPath": "/cgroup",
        "hostPath": "/cgroup",
        "mode": "RO"
      }
    ],
    "portMappings": [
      {
        "containerPort": 8080,
        "hostPort": 0,
        "labels": {},
        "protocol": "tcp",
        "servicePort": 10130
      }
    ]
  },
  "labels": {
    "HAPROXY_GROUP": "external"
  },
  "networks": [
    {
      "mode": "container/bridge"
    }
  ],
  "portDefinitions": [],
  "maxLaunchDelaySeconds": 300
}

2、配置prometheus：

修改prometheus.yml文件增加如下：

- job_name: 'docker161'
   scrape_interval: 5s
 
   file_sd_configs:
     - files: ['/etc/prometheus/fileconfig/docker-nodes161.json']

cat /etc/prometheus/fileconfig/docker-nodes161.json

[
    {
        "targets": ["10.20.31.164:10128"],
        "labels": {
            "instance": "10.20.31.161",
            "alias": "161-docker",
            "job": "docker161"
        }
    }
]

因为是修改了prometheus.yml文件，所以要重新加载一下prometheus服务

1	`curl -XPOST http://10.20.31.100:9090/-/reload`

访问prometheus页面——Status——Targets可以查看到刚添加的docker主机信息。

3、配置grafana：

{
  "id": "/grafana",
  "role": "slave_public",
  "cmd": null,
  "cpus": 0.3,
  "mem": 512,
  "disk": 0,
  "gpus": 0,
  "instances": 1,
  "constraints": [
    [
      "hostname",
      "CLUSTER",
      "10.20.31.161"
    ]
  ],
  "acceptedResourceRoles": [
    "*"
  ],
  "container": {
    "type": "DOCKER",
    "docker": {
      "forcePullImage": false,
      "image": "10.20.31.104:5000/library/grafana",
      "parameters": [],
      "privileged": false
    },
    "volumes": [
      {
        "containerPath": "/etc/localtime",
        "hostPath": "/etc/localtime",
        "mode": "RO"
      },
      {
        "containerPath": "/var/lib/grafana",
        "hostPath": "/opt/GrafanaData",
        "mode": "RW"
      }
    ],
    "portMappings": [
      {
        "containerPort": 3000,
        "hostPort": 0,
        "labels": {},
        "protocol": "tcp",
        "servicePort": 10135
      }
    ]
  },
  "labels": {
    "HAPROXY_GROUP": "external"
  },
  "networks": [
    {
      "mode": "container/bridge"
    }
  ],
  "portDefinitions": [],
  "maxLaunchDelaySeconds": 300
}