从KubeSphere v3.1.0边缘节点的监控问题排查，简要解析边缘监控原理

zhu733756
我复现了一下，并没有发现大佬描述的问题：

1 docker 安装

sudo apt-get install docker-ce=5:19.03.9~3-0~ubuntu-focal docker-ce-cli=5:19.03.9~3-0~ubuntu-focal containerd.io

查看docker 版本

2 边缘节点 node metrics


3 边缘节点pod metrics

PS： 我的边缘节点均通过外网ip与cloudcore通信

zhu733756 上面那个docker版本高一些，我重新测了

zhu733756

zhu733756 版本号看错了，我重新测一下，我看截图去了
———-
更新新进展： 确实跟大佬发现的问题一样，跟内核版本和操作系统没关系，我测试centos和ubuntu两个操作系统下的docker版本都是 边缘节点所在的pod metrics为空：

curl -ki localhost:10350/stats/summary

{
 "node": {
  "nodeName": "edgenode-wnlb",
  "systemContainers": [
   {
    "name": "pods",
    "startTime": "2021-05-19T02:51:55Z",
    "cpu": {
     "time": "2021-05-19T02:58:41Z",
     "usageNanoCores": 0,
     "usageCoreNanoSeconds": 81040512
    },
    "memory": {
     "time": "2021-05-19T02:58:41Z",
     "availableBytes": 1033695232,
     "usageBytes": 6631424,
     "workingSetBytes": 6430720,
     "rssBytes": 1507328,
     "pageFaults": 0,
     "majorPageFaults": 0
    }
   },
   {
    "name": "kubelet",
    "startTime": "2021-05-19T02:51:55Z",
    "cpu": {
     "time": "2021-05-19T02:58:39Z",
     "usageNanoCores": 7826966,
     "usageCoreNanoSeconds": 4676221589
    },
    "memory": {
     "time": "2021-05-19T02:58:39Z",
     "usageBytes": 40861696,
     "workingSetBytes": 39350272,
     "rssBytes": 37896192,
     "pageFaults": 33726,
     "majorPageFaults": 17
    }
   },
   {
    "name": "runtime",
    "startTime": "2021-05-19T02:51:55Z",
    "cpu": {
     "time": "2021-05-19T02:58:36Z",
     "usageNanoCores": 2928595,
     "usageCoreNanoSeconds": 7161199061
    },
    "memory": {
     "time": "2021-05-19T02:58:36Z",
     "usageBytes": 212058112,
     "workingSetBytes": 98967552,
     "rssBytes": 59850752,
     "pageFaults": 94660,
     "majorPageFaults": 55
    }
   }
  ],
  "startTime": "2021-05-19T02:21:35Z",
  "cpu": {
   "time": "2021-05-19T02:58:39Z",
   "usageNanoCores": 17408633,
   "usageCoreNanoSeconds": 65648880947
  },
  "memory": {
   "time": "2021-05-19T02:58:39Z",
   "availableBytes": 557957120,
   "usageBytes": 847716352,
   "workingSetBytes": 482168832,
   "rssBytes": 149295104,
   "pageFaults": 42402,
   "majorPageFaults": 85
  },
  "network": {
   "time": "2021-05-19T02:58:39Z",
   "name": "eth0",
   "rxBytes": 222483820,
   "rxErrors": 0,
   "txBytes": 8539498,
   "txErrors": 0,
   "interfaces": [
    {
     "name": "eth0",
     "rxBytes": 222483820,
     "rxErrors": 0,
     "txBytes": 8539498,
     "txErrors": 0
    }
   ]
  },
  "fs": {
   "time": "2021-05-19T02:58:39Z",
   "availableBytes": 18767896576,
   "capacityBytes": 20749852672,
   "usedBytes": 1965178880,
   "inodesFree": 2492827,
   "inodes": 2560000,
   "inodesUsed": 67173
  },
  "runtime": {
   "imageFs": {
    "time": "2021-05-19T02:58:39Z",
    "availableBytes": 18767896576,
    "capacityBytes": 20749852672,
    "usedBytes": 133865025,
    "inodesFree": 2492827,
    "inodes": 2560000,
    "inodesUsed": 67173
   }
  },
  "rlimit": {
   "time": "2021-05-19T02:58:42Z",
   "maxpid": 32768,
   "curproc": 177
  }
 },
 #正常来说这里应该是有数据的
 "pods": []

zhu733756 边缘节点监控数据不显示还可以使用如下方法，感谢你的指导支持。

kubectl -n kubeedge edit cm edge-watcher

修改完毕后重启edge-watcher-controller-manager服务

yinzhipeng123 改成0.0.0.0试试？

边缘测得通过云端得ip去通信。所以云端暴露得ip，边缘需要能访问到即可。

metrics是通过metric-server来发起的，cloudcore：tunelport中转来访问边缘测，只要边缘测和云端正常通信，应该是没有问题的。

如果你需要再其他主机访问边缘metrics，哪得用0.,0.0.0，以及主机可通信。

zhu733756 现在就是边缘端的监控没有，根据楼主的思路去排查，现在发现边缘端的监控信息现在暴露在边缘端的127.0.0.1上，这个如何改成0.0.0.0，改什么配置呢？

yinzhipeng123 云端（部署cloudcore的node节点）的iptables转发规则，grep 30003或者10350，图片中的都是边缘端。
之前写得排查指南，https://kubesphere.com.cn/forum/d/4362-kubeedge，
基本上涵盖了所有类型的问题，遇到问题可以先搜搜。