{"id":43175,"date":"2022-11-21T20:39:11","date_gmt":"2023-09-21T03:16:11","guid":{"rendered":"https:\/\/www.silicloud.com\/zh\/blog\/%e4%bd%bf%e7%94%a8prometheus%e7%9b%91%e8%a7%86kubernetes%e7%9a%84gpu%e8%8a%82%e7%82%b9\/"},"modified":"2024-04-29T01:49:45","modified_gmt":"2024-04-28T17:49:45","slug":"%e4%bd%bf%e7%94%a8prometheus%e7%9b%91%e8%a7%86kubernetes%e7%9a%84gpu%e8%8a%82%e7%82%b9","status":"publish","type":"post","link":"https:\/\/www.silicloud.com\/zh\/blog\/%e4%bd%bf%e7%94%a8prometheus%e7%9b%91%e8%a7%86kubernetes%e7%9a%84gpu%e8%8a%82%e7%82%b9\/","title":{"rendered":"\u4f7f\u7528Prometheus\u76d1\u89c6Kubernetes\u7684GPU\u8282\u70b9"},"content":{"rendered":"<h1>\u9996\u5148<\/h1>\n<p>\u5728\u8fd1\u5e74\u7684\u6df1\u5ea6\u5b66\u4e60\u70ed\u6f6e\u4e2d\uff0c\u6211\u4eec\u7ecf\u5e38\u542c\u5230\u4f7f\u7528Kubernetes\u7ba1\u7406\u591a\u4e2aGPU\u8282\u70b9\u5e76\u8fdb\u884c\u591a\u4eba\u5171\u4eab\u7684\u64cd\u4f5c\u65b9\u5f0f\u3002\u6211\u4eec\u516c\u53f8\u4e5f\u5728\u672c\u5730\u73af\u5883\u642d\u5efa\u4e86GPU\u96c6\u7fa4\uff0c\u5e76\u5229\u7528\u5f00\u6e90\u8f6f\u4ef6KAMONOHASHI\u8fdb\u884c\u57fa\u4e8e\u6df1\u5ea6\u5b66\u4e60\u7684\u4eba\u5de5\u667a\u80fd\u5f00\u53d1\u3002<\/p>\n<p>\u5728\u5171\u4eab\u548c\u4f7f\u7528GPU\u96c6\u7fa4\u7684\u8fc7\u7a0b\u4e2d\uff0c\u6709\u4e00\u4e2a\u9700\u6c42\u662f\u8981\u6536\u96c6GPU\u8282\u70b9\u7684\u4f7f\u7528\u60c5\u51b5\u4f5c\u4e3a\u5ea6\u91cf\uff0c\u5e76\u53ef\u89c6\u5316\u8fc7\u53bb\u7684\u4f7f\u7528\u5386\u7a0b\u3002\u56e0\u6b64\u6211\u5c1d\u8bd5\u4f7f\u7528NVIDIA DCGM exporter for Prometheus\u548cPrometheus\u6765\u6536\u96c6GPU\u7684\u5ea6\u91cf\u3002<\/p>\n<p>\u53e6\u5916\uff0c\u672c\u6587\u4f7f\u7528\u7684\u6587\u4ef6\u5df2\u653e\u7f6e\u5728GitHub\u4e0a\u3002<\/p>\n<h1>\u8fdb\u7403 qi\u00fa)<\/h1>\n<p>\u5728\u5305\u542bGPU\u8282\u70b9\u7684Kubernetes\u96c6\u7fa4\u4e2d\uff0c\u5f15\u5165Prometheus\u548cDCGM exporter\uff0c\u4ee5\u4fbf\u901a\u8fc7Prometheus\u6536\u96c6GPU\u7684\u6307\u6807\u3002<\/p>\n<div><img decoding=\"async\" class=\"post-images\" title=\"\" src=\"https:\/\/cdn.silicloud.com\/blog-img\/blog\/img\/657d513737434c4406cc9247\/6-0.png\" alt=\"prometheus.png\" \/><\/div>\n<h1>\u7248\u672c\u4fe1\u606f x\u012b)<\/h1>\n<p>\u6bcf\u4e2a\u7ec4\u4ef6\u6240\u4f7f\u7528\u7684\u7248\u672c\u5982\u4e0b\uff1a<br \/>\n\u4f7f\u7528\u7684\u5404\u7ec4\u4ef6\u7248\u672c\u5982\u4e0b\u6240\u793a\u3002<\/p>\n<ul class=\"post-ul\">\n<li style=\"list-style-type: none;\">\n<ul class=\"post-ul\">Kubernetes: v1.14.1<\/ul>\n<\/li>\n<\/ul>\n<p>&nbsp;<\/p>\n<ul class=\"post-ul\">\n<li style=\"list-style-type: none;\">\n<ul class=\"post-ul\">helm: v2.13.0<\/ul>\n<\/li>\n<\/ul>\n<p>NVIDIA DCGM exporter for Prometheus: d39e0cc8675c0b10aab14d4f66191eedf5029f66<\/p>\n<p>Prometheus: 2.13.1(helm chart\u306fprometheus-9.3.1\u3092\u5229\u7528)<\/p>\n<p>nginx-ingress: v0.26.1(helm chart\u306fnginx-ingress-1.24.7\u3092\u5229\u7528)<\/p>\n<p>\u6211\u4f7f\u7528Helm\u6765\u5b89\u88c5\u4e86Prometheus\u548cnginx-ingress\u3002<\/p>\n<h1>\u96c6\u7fa4\u7684\u7ec4\u6210<\/h1>\n<p>\u5047\u8bbe\u96c6\u7fa4\u4e2d\u5305\u542b\u4ee5\u4e0b\u8282\u70b9\u3002<\/p>\n<ul class=\"post-ul\">\n<li style=\"list-style-type: none;\">\n<ul class=\"post-ul\">master: Kubernetes\u306emaster\u30ce\u30fc\u30c9<\/ul>\n<\/li>\n<\/ul>\n<p>&nbsp;<\/p>\n<ul class=\"post-ul\">\n<li style=\"list-style-type: none;\">\n<ul class=\"post-ul\">gpu1: NVIDIA Tesla K80\u642d\u8f09\u306eGPU\u30ce\u30fc\u30c9<\/ul>\n<\/li>\n<\/ul>\n<p>&nbsp;<\/p>\n<ul class=\"post-ul\">\n<li style=\"list-style-type: none;\">\n<ul class=\"post-ul\">gpu2: NVIDIA Tesla K80\u642d\u8f09\u306eGPU\u30ce\u30fc\u30c9<\/ul>\n<\/li>\n<\/ul>\n<p>&nbsp;<\/p>\n<ul class=\"post-ul\">kamonohashi: KAMONOHASHI\u3067\u5229\u7528\u306e\u30ce\u30fc\u30c9\u3002\u4eca\u56de\u306f\u95a2\u4fc2\u306a\u3057<\/ul>\n<p>\u6211\u4eec\u6b63\u5728\u4f7f\u7528\u73b0\u6709\u7684KAMONOHASHI\u96c6\u7fa4\u3002\u5728\u5f15\u5165KAMONOHASHI\u65f6\uff0c\u6211\u4eec\u4f7f\u7528kubespray\u8fdb\u884c\u4e86Kubernetes\u7684\u6784\u5efa\u3002<\/p>\n<p>\u8fd9\u6b21\u6211\u4eec\u5c06\u5728\u4e3b\u8282\u70b9\u4e0a\u90e8\u7f72Prometheus\uff0c\u5e76\u901a\u8fc7Ingress\u8fdb\u884c\u8bbf\u95ee\u3002\u6211\u4eec\u5c06\u4f7f\u7528Nginx Ingress Controller\u4f5c\u4e3aIngress\u63a7\u5236\u5668\u3002<\/p>\n<p>\u7531\u4e8eDCGM\u5bfc\u51fa\u5668\u7684\u8fd0\u884c\u9700\u6c42\u5df2\u7ecf\u88ab\u63d0\u51fa\uff0c\u6211\u4eec\u5c06\u786e\u4fdd\u6bcf\u4e2aGPU\u8282\u70b9\u6ee1\u8db3\u8fd9\u4e9b\u8981\u6c42\u3002<\/p>\n<blockquote><p>\u5148\u51b3\u6761\u4ef6<\/p>\n<p>NVIDIA Tesla \u9a71\u52a8\u7a0b\u5e8f = R384+\uff08\u4ece NVIDIA \u9a71\u52a8\u7a0b\u5e8f\u4e0b\u8f7d\u9875\u9762\u4e0b\u8f7d\uff09<br \/>\nnvidia-docker \u7248\u672c &gt; 2.0\uff08\u53c2\u89c1\u5982\u4f55\u5b89\u88c5\u53ca\u5176\u5148\u51b3\u6761\u4ef6\uff09<br \/>\n\u53ef\u9009\u62e9\u914d\u7f6e docker \u5c06\u9ed8\u8ba4\u8fd0\u884c\u65f6\u8bbe\u7f6e\u4e3a nvidia<\/p><\/blockquote>\n<h1>NVIDIA\u63d0\u4f9b\u7684GPU\u76d1\u6d4b\u5de5\u5177<\/h1>\n<p>NVIDIA\u63d0\u4f9b\u7684GPU\u76d1\u63a7\u5de5\u5177\u53ef\u4ee5\u4eceGitHub\u7684NVIDIA\/gpu-monitoring-tools\u83b7\u53d6\u3002\u5176\u4e2d\uff0cNVIDIA DCGM exporter for Prometheus\u53ef\u4f5c\u4e3a\u5728Prometheus\u4e2d\u83b7\u53d6GPU\u6307\u6807\u7684\u5bfc\u51fa\u5de5\u5177\u4f7f\u7528\u3002DCGM\u662fNVIDIA\u6570\u636e\u4e2d\u5fc3GPU\u7ba1\u7406\u5668\u7684\u7f29\u5199\uff0c\u7528\u4e8e\u83b7\u53d6GPU\u96c6\u7fa4\u7684\u6307\u6807\u7684\u5de5\u5177\u96c6\u5408\u3002<\/p>\n<p>\u8fd9\u4e2a\u5b58\u50a8\u5e93\u6709\u70b9\u96be\u4ee5\u7406\u89e3\uff0c\u6240\u4ee5\u6211\u5728\u8fd9\u91cc\u6574\u7406\u4e00\u4e0b\u3002<br \/>\nk8s\/node-exporter\u5728NVIDIA DCGM exporter for Prometheus\u4e2d\uff0c\u662f\u5b58\u50a8\u4e86exporter\u7684DaemonSet\u6e05\u5355\u7684\u76ee\u5f55\uff08k8s\/pod-gpu-metrics-exporter\u5b58\u50a8\u4e86exporter\u672c\u8eab\u7684\u6e90\u4ee3\u7801\uff09\u3002<\/p>\n<p>k8s\/node-exporter \u4e2d\u5305\u542b\u4ee5\u4e0b\u4e09\u4e2a\u6e05\u5355\u6587\u4ef6\u3002<\/p>\n<p>gpu-node-exporter-daemonset.yaml<br \/>\n\u7528\u4e8e\u83b7\u53d6GPU\u8282\u70b9\u6307\u6807\u4ee5\u53ca\u5305\u542bCPU\u7b49\u4fe1\u606f\u7684\u5bfc\u51fa\u5668\u7684DaemonSet\u3002<\/p>\n<p>\u53ea\u9700\u8981\u4e00\u4e2a\u9009\u62e9\uff0c\u7528\u4e2d\u6587\u6539\u5199\u4ee5\u4e0b\u5185\u5bb9\uff1a<br \/>\ngpu-only-node-exporter-daemonset.yaml<br \/>\n\u83b7\u53d6\u4ec5 GPU \u4fe1\u606f\u7684\u8282\u70b9\u6307\u6807\u7684 exporter \u7684 DaemonSet\u3002<\/p>\n<p>pod-gpu-node-exporter-daemonset.yaml\u662f\u4e00\u4e2aDaemonSet\uff0c\u7528\u4e8e\u83b7\u53d6\u6bcf\u4e2aPod\u7684GPU\u5229\u7528\u60c5\u51b5\u7684\u6307\u6807\u3002<\/p>\n<p>\u8fd9\u4e9bDaemonSet\u7531\u4ee5\u4e0bexporter\u7684\u7ec4\u5408\u6784\u6210\u3002<\/p>\n<p>prometheus\/node-exporter\u662fPrometheus\u63d0\u4f9b\u7684\u4e00\u4e2a\u5e38\u89c1\u7684exporter\u3002<\/p>\n<p>nvidia\/dcgm-exporter\u662fNVIDIA\u63d0\u4f9b\u7684\u7528\u4e8e\u83b7\u53d6GPU\u8282\u70b9\u6307\u6807\u7684\u5bfc\u51fa\u7a0b\u5e8f\u3002\u4f7f\u7528\u6b64\u5bfc\u51fa\u7a0b\u5e8f\uff0c\u53ef\u4ee5\u83b7\u53d6\u6bcf\u4e2aGPU\u8282\u70b9\u7684GPU\u5229\u7528\u7387\u548cGPU\u5185\u5b58\u4f7f\u7528\u91cf\u3002gpu-node-exporter-daemonset\u548cgpu-only-node-exporter-daemonset\u662f\u5c06prometheus\/node-exporter\u4e0e\u6b64\u5bfc\u51fa\u7a0b\u5e8f\u96c6\u6210\u7684\u89e3\u51b3\u65b9\u6848\u3002<\/p>\n<p>nvidia\/pod-gpu-metrics-exporter\u662f\u4e00\u4e2a\u7528\u4e8e\u83b7\u53d6\u6bcf\u4e2aPod\u7684GPU\u4f7f\u7528\u60c5\u51b5\u6307\u6807\u7684\u5bfc\u51fa\u5668\u3002\u4f7f\u7528\u6b64\u5bfc\u51fa\u5668\u53ef\u4ee5\u83b7\u53d6\u6bcf\u4e2aGPU\u8282\u70b9\u4e0a\u8fd0\u884c\u7684Pod\u7684GPU\u5229\u7528\u7387\u548cGPU\u5185\u5b58\u4f7f\u7528\u91cf\u3002\u6b64\u5bfc\u51fa\u5668\u5df2\u4e0eprometheus\/node-exporter\u548cnvidia\/dcgm-exporter\u8fdb\u884c\u4e86\u6574\u5408\uff0c\u5e76\u5728pod-gpu-node-exporter-daemonset\u4e2d\u4f7f\u7528\u3002<br \/>\n\u8981\u4f7f\u7528kubelet\u83b7\u53d6Pod\u7684\u4fe1\u606f\uff0c\u9700\u8981\u542f\u7528FeatureGates\u7684KubeletPodResources\u529f\u80fd\uff08\u5728Kubernetes v1.15\u4e4b\u540e\u9ed8\u8ba4\u542f\u7528\uff09\u3002<\/p>\n<p>\u672c\u6587\u5c06\u4f7f\u7528pod-gpu-node-exporter-daemonset.yaml\u6765\u83b7\u53d6\u6bcf\u4e2aPod\u7684GPU\u4f7f\u7528\u60c5\u51b5\u7684\u6307\u6807\u3002<\/p>\n<h1>\u6b65\u9aa4<\/h1>\n<h3>1. kubelet\u7684\u914d\u7f6e\u66f4\u6539<\/h3>\n<p>\u7531\u4e8e\u672c\u6b21\u4f7f\u7528\u7684Kubernetes\u7248\u672c\u662fv1.14.1\uff0c\u6240\u4ee5\u4e3a\u4e86\u83b7\u53d6\u6bcf\u4e2aPod\u7684GPU\u4f7f\u7528\u60c5\u51b5\u6307\u6807\uff0c\u9700\u8981\u5728\u6bcf\u4e2aGPU\u8282\u70b9\u4e0a\u542f\u7528kubelet\u7684FeatureGates\u529f\u80fd\u3002<br \/>\n\u60a8\u9700\u8981\u7f16\u8f91\u6bcf\u4e2aGPU\u8282\u70b9\u4e0a\u7684\/etc\/systemd\/system\/kubelet.service\u6587\u4ef6\uff0c\u5e76\u4fee\u6539ExecStart\u90e8\u5206\u3002<\/p>\n<pre class=\"post-pre\"><code><span class=\"p\">ExecStart=\/usr\/local\/bin\/kubelet \\\r\n<\/span><span class=\"gi\">+                --feature-gates KubeletPodResources=true \\\r\n<\/span><\/code><\/pre>\n<p>\u7136\u540e\u4f7f\u7528\u4ee5\u4e0b\u547d\u4ee4\u91cd\u65b0\u542f\u52a8kubelet\u3002<\/p>\n<pre class=\"post-pre\"><code>systemctl daemon-reload\r\nsystemctl restart kubelet\r\n<\/code><\/pre>\n<p>\u901a\u8fc7\u8fd9\u4e9b\u64cd\u4f5c\uff0c\u4e0a\u8ff0\u7684 exporter \u5c06\u80fd\u591f\u83b7\u53d6\u5230 Pod \u7684\u4fe1\u606f\u3002<\/p>\n<h3>\u521b\u5efa\u547d\u540d\u7a7a\u95f4<\/h3>\n<p>\u6211\u5011\u5c07\u70ba\u9019\u6b21\u90e8\u7f72\u7684\u8cc7\u6e90\u5275\u5efa\u4e00\u500b\u547d\u540d\u7a7a\u9593\u3002\u540d\u7a31\u5c07\u70bamonitoring\u3002<\/p>\n<pre class=\"post-pre\"><code>kubectl create namespace monitoring\r\n<\/code><\/pre>\n<h3>3. \u4e3aGPU\u8282\u70b9\u6253\u6807\u7b7e<\/h3>\n<p>pod-gpu-node-exporter-daemonset\u8bbe\u7f6e\u4e86nodeSelector\uff0c\u5c06hardware-type\uff1aNVIDIAGPU\u8bbe\u7f6e\u4e3a\u6807\u7b7e\uff0c\u5e76\u5c06exporter\u90e8\u7f72\u5728\u5177\u6709\u6b64\u6807\u7b7e\u7684\u8282\u70b9\u4e0a\u3002<br \/>\n\u56e0\u6b64\uff0c\u53ef\u4ee5\u5bf9GPU\u8282\u70b9\u8fdb\u884c\u4ee5\u4e0b\u6807\u7b7e(\u53ef\u4ee5\u66f4\u6539\u4e3agpu\u8282\u70b9\u7684\u4e3b\u673a\u540d\uff0c\u5982gpu1)\u3002<\/p>\n<pre class=\"post-pre\"><code>kubectl label nodes &lt;gpu-node-name&gt; hardware-type<span class=\"o\">=<\/span>NVIDIAGPU\r\n<\/code><\/pre>\n<p>\u60a8\u53ef\u4ee5\u4f7f\u7528\u4ee5\u4e0b\u547d\u4ee4\u68c0\u67e5\u6807\u7b7e\u7684\u6dfb\u52a0\u60c5\u51b5\u3002<\/p>\n<pre class=\"post-pre\"><code>kubectl get nodes --show-labels\r\n<\/code><\/pre>\n<h3>\u521b\u5efaPrometheus\u7684\u6570\u636e\u5b58\u50a8\u4f4d\u7f6e\u3002<\/h3>\n<p>\u5728\u90e8\u7f72 Prometheus \u65f6\uff0c\u9700\u8981\u6307\u5b9a\u6536\u96c6\u5230\u7684\u6570\u636e\u7684\u5b58\u50a8\u4f4d\u7f6e\u3002\u672c\u6b21\u6211\u4eec\u5c06\u6570\u636e\u4fdd\u5b58\u5728 master \u8282\u70b9\u7684 \/var\/lib\/prometheus \u76ee\u5f55\u4e0b\u3002\u6211\u4eec\u53ef\u4ee5\u4f7f\u7528\u4ee5\u4e0b\u547d\u4ee4\u521b\u5efa\u8be5\u76ee\u5f55\uff0c\u5e76\u5c06\u5176\u6240\u6709\u8005\u8bbe\u7f6e\u4e3a nobody\u3002\u7531\u4e8e Prometheus \u670d\u52a1\u5668\u5c06\u4ee5 nobody \u7528\u6237\u8eab\u4efd\u8fd0\u884c\uff0c\u56e0\u6b64\u9700\u8981\u4fee\u6539\u6240\u6709\u8005\u3002<\/p>\n<pre class=\"post-pre\"><code>mkdir \/var\/lib\/prometheus\r\nchown nobody \/var\/lib\/prometheus\r\n<\/code><\/pre>\n<h3>5. \u90e8\u7f72\u5404\u79cd\u8d44\u6e90<\/h3>\n<p>\u5c06\u4ee5\u4e0b\u8d44\u6e90\u90e8\u7f72\u5230GitHub\u3002\u4f7f\u7528\u7684\u6587\u4ef6\u5df2\u7ecf\u653e\u5728GitHub\u4e0a\u4e86\u3002<\/p>\n<ul class=\"post-ul\">\n<li style=\"list-style-type: none;\">\n<ul class=\"post-ul\">GPU\u30ce\u30fc\u30c9\u7528\u306eexporter\u3092\u914d\u7f6e\u3059\u308b\u305f\u3081\u306eDaemonSet(pod-gpu-node-exporter-daemonset.yaml)<\/ul>\n<\/li>\n<\/ul>\n<p>&nbsp;<\/p>\n<ul class=\"post-ul\">\n<li style=\"list-style-type: none;\">\n<ul class=\"post-ul\">exporter\u306ePod\u306b\u30a2\u30af\u30bb\u30b9\u3059\u308b\u305f\u3081\u306eservice(exporter-service.yaml)<\/ul>\n<\/li>\n<\/ul>\n<p>&nbsp;<\/p>\n<ul class=\"post-ul\">Prometheus\u3067\u5229\u7528\u3059\u308bPersistent Volume, Persistent Volume Claim(prometheus-pv-pvc.yaml)<\/ul>\n<p>\u5173\u4e8eGPU\u8282\u70b9\u7684exporter\uff0c\u6211\u4eec\u5c06\u4f7f\u7528\u4fee\u6539\u4e86\u539f\u59cb\u5e93\u4e2d\u7684pod-gpu-node-exporter-daemonset.yaml\u6587\u4ef6\uff0c\u5c06\u5176namespace\u4ecekube-system\u66f4\u6539\u4e3amonitoring\u3002\u6b64\u5916\uff0c\u8fd8\u4f1a\u521b\u5efa\u4e00\u4e2aservice\uff0c\u4ee5\u4f7fPrometheus\u53ef\u4ee5\u4eceexporter\u7684Pod\u4e2d\u6536\u96c6\u4fe1\u606f\u3002\u53e6\u5916\uff0c\u6211\u4eec\u4f1a\u4ee58GB\u7684\u5bb9\u91cf\u521b\u5efa\u7528\u4e8e\u5b58\u50a8Prometheus\u6570\u636e\u7684PV\u548cPVC\uff08\u6839\u636ePrometheus\u56fe\u8868\u7684\u9ed8\u8ba4\u503c\uff09\u3002<\/p>\n<p>\u8bf7\u4f7f\u7528\u4ee5\u4e0b\u547d\u4ee4\u5e94\u7528\u8fd9\u4e09\u4e2a\u6e05\u5355\u6587\u4ef6\u3002<\/p>\n<pre class=\"post-pre\"><code>kubectl apply -f pod-gpu-node-exporter-daemonset.yaml -f exporter-service.yaml -f prometheus-pv-pvc.yaml\r\n<\/code><\/pre>\n<p>\u60a8\u53ef\u4ee5\u4f7f\u7528\u4ee5\u4e0b\u547d\u4ee4\u6765\u786e\u8ba4\u5bfc\u51fa\u5668\u90e8\u7f72\u540e\uff0cGPU\u8282\u70b9\u7684\u6307\u6807\u662f\u5426\u6b63\u5e38\u8f93\u51fa\u3002<\/p>\n<pre class=\"post-pre\"><code>curl &lt;gpu-node-name&gt;:9100\/metrics | grep dcgm\r\n<\/code><\/pre>\n<p>\u5982\u679c\u8f93\u51fa\u4e86 dcgm_gpu_utilization \u7b49\u4fe1\u606f\uff0c\u5219\u8868\u793a exporter \u6b63\u5e38\u542f\u52a8\u3002\u5efa\u8bae\u5728\u6b64\u5904\u8fdb\u884c\u786e\u8ba4\u4e00\u6b21\u3002<\/p>\n<h3>6. Prometheus\u7684\u5b89\u88c5<\/h3>\n<p>\u7528Helm\u5de5\u5177\u5b89\u88c5Prometheus\u3002<br \/>\n\u901a\u8fc7\u4ee5\u4e0b\u547d\u4ee4\u4e0b\u8f7d\u5e76\u7f16\u8f91Prometheus\u914d\u7f6e\u6587\u4ef6\u3002<\/p>\n<pre class=\"post-pre\"><code>helm inspect values stable\/prometheus <span class=\"o\">&gt;<\/span> prometheus-values.yaml\r\n<\/code><\/pre>\n<p>\u7531\u4e8e\u672c\u6b21\u7684\u76ee\u7684\u662f\u5148\u4f7f\u5176\u80fd\u6b63\u5e38\u8fd0\u884c\uff0c\u56e0\u6b64\u6211\u4eec\u5c06\u4ee5\u51e0\u4e4e\u6240\u6709\u529f\u80fd\u88ab\u7981\u7528\u7684\u72b6\u6001\u8fdb\u884c\u90e8\u7f72\u3002<\/p>\n<ul class=\"post-ul\">\n<li style=\"list-style-type: none;\">\n<ul class=\"post-ul\">Prometheus server\u306e\u307f\u6709\u52b9\u5316<\/ul>\n<\/li>\n<\/ul>\n<p>http:\/\/master-hostname\/prometheus\u306e\u3088\u3046\u306a\u5f62\u3067\u30a2\u30af\u30bb\u30b9\u3059\u308b\u305f\u3081\u306bprefixURL\u3068baseURL\u3092\u5909\u66f4<br \/>\nprometheus server\u306eingress\u3092\u6709\u52b9\u5316\u3057\u3001ingress.class\u3068\u3057\u3066prometheus-nginx\u3092\u6307\u5b9a<br \/>\nmaster\u30ce\u30fc\u30c9\u306b\u30c7\u30d7\u30ed\u30a4\u3059\u308b\u3088\u3046nodeSelector\u3092\u8a2d\u5b9a<br \/>\npersistentVolume\u3068\u3057\u3066\u4e0a\u8a18\u3067\u4f5c\u6210\u3057\u305f\u3082\u306e\u3092\u6307\u5b9a<\/p>\n<p>\u7de8\u8f2f\u5b8c\u6210\u5f8c\uff0c\u4f7f\u7528\u4ee5\u4e0b\u547d\u4ee4\u9032\u884cPrometheus\u7684\u5b89\u88dd\u3002<\/p>\n<pre class=\"post-pre\"><code>helm install --name prometheus --namespace monitoring --version 9.3.1 -f prometheus-values.yaml stable\/prometheus\r\n<\/code><\/pre>\n<h3>7. \u5b89\u88c5 Ingress Controller<\/h3>\n<p>\u4e3a\u4e86Prometheus Server\u7684\u5165\u53e3\uff0c\u5b89\u88c5Nginx Ingress Controller\u3002<br \/>\n\u4f7f\u7528\u4ee5\u4e0b\u547d\u4ee4\u4e0b\u8f7d\u5e76\u7f16\u8f91\u914d\u7f6e\u6587\u4ef6\u3002<\/p>\n<pre class=\"post-pre\"><code>helm inspect values stable\/nginx-ingress <span class=\"o\">&gt;<\/span> nginx-ingress-values.yaml\r\n<\/code><\/pre>\n<p>\u4ee5\u4e0b\u662f\u7f16\u8f91\u5185\u5bb9\u7684\u6982\u8ff0\u3002<\/p>\n<ul class=\"post-ul\">\n<li style=\"list-style-type: none;\">\n<ul class=\"post-ul\">ingressClass\u3092prometheus-nginx\u306b\u5909\u66f4<\/ul>\n<\/li>\n<\/ul>\n<p>&nbsp;<\/p>\n<ul class=\"post-ul\">\n<li style=\"list-style-type: none;\">\n<ul class=\"post-ul\">scope\u3092\u6709\u52b9\u306b\u3057\u3001namespace\u3068\u3057\u3066monitoring\u3092\u6307\u5b9a<\/ul>\n<\/li>\n<\/ul>\n<p>&nbsp;<\/p>\n<ul class=\"post-ul\">\n<li style=\"list-style-type: none;\">\n<ul class=\"post-ul\">defaultBackendService\u3068\u3057\u3066monitoring\/prometheus-server\u3092\u6307\u5b9a<\/ul>\n<\/li>\n<\/ul>\n<p>&nbsp;<\/p>\n<ul class=\"post-ul\">\n<li style=\"list-style-type: none;\">\n<ul class=\"post-ul\">master\u30ce\u30fc\u30c9\u306b\u30c7\u30d7\u30ed\u30a4\u3059\u308b\u3088\u3046nodeSelector\u3092\u8a2d\u5b9a<\/ul>\n<\/li>\n<\/ul>\n<p>&nbsp;<\/p>\n<ul class=\"post-ul\">externalIPs\u3068\u3057\u3066master\u30ce\u30fc\u30c9\u306eIP\u3092\u6307\u5b9a<\/ul>\n<p>\u7f16\u8f91\u5b8c\u6210\u540e\uff0c\u4f7f\u7528\u4ee5\u4e0b\u547d\u4ee4\u5b89\u88c5Ingress Controller\u3002<\/p>\n<pre class=\"post-pre\"><code>helm install stable\/nginx-ingress --name prometheus-nginx --namespace monitoring --version 1.24.7 -f nginx-ingress-values.yaml\r\n<\/code><\/pre>\n<h3>8. \u4f7f\u7528Prometheus<\/h3>\n<div><img decoding=\"async\" class=\"post-images\" title=\"\" src=\"https:\/\/cdn.silicloud.com\/blog-img\/blog\/img\/657d513737434c4406cc9247\/72-0.png\" alt=\"prometheus.png\" \/><\/div>\n<h2>\u603b\u7ed3<\/h2>\n<p>\u901a\u8fc7\u5c06NVIDIA DCGM\u5bfc\u51fa\u7a0b\u5e8f\u4e0ePrometheus\u7ed3\u5408\u4f7f\u7528\uff0c\u53ef\u4ee5\u76d1\u89c6GPU\u8d44\u6e90\u7684\u5229\u7528\u60c5\u51b5\u3002\u540c\u65f6\uff0c\u7ed3\u5408Grafana\uff0c\u8fd8\u53ef\u4ee5\u5c06Prometheus\u6536\u96c6\u5230\u7684\u7ed3\u679c\u8fdb\u884c\u7f8e\u89c2\u7684\u53ef\u89c6\u5316\u3002<\/p>\n<p>\u5e0c\u671b\u8fd9\u53ef\u4ee5\u5e2e\u52a9\u5230\u5e0c\u671b\u5b9e\u73b0\u7c7b\u4f3c\u76ee\u6807\u7684\u4eba\uff0c\u4f46\u5728\u5b9e\u9645\u8fd0\u4f5c\u4e2d\uff0c\u9700\u8981\u8be6\u7ec6\u8003\u8651\u6307\u6807\u5b58\u50a8\u7a7a\u95f4\u7684\u5927\u5c0f\u7b49\u95ee\u9898\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u9996\u5148 \u5728\u8fd1\u5e74\u7684\u6df1\u5ea6\u5b66\u4e60\u70ed\u6f6e\u4e2d\uff0c\u6211\u4eec\u7ecf\u5e38\u542c\u5230\u4f7f\u7528Kubernetes\u7ba1\u7406\u591a\u4e2aGPU\u8282\u70b9\u5e76\u8fdb\u884c\u591a\u4eba\u5171\u4eab\u7684\u64cd\u4f5c\u65b9\u5f0f\u3002 [&hellip;]<\/p>\n","protected":false},"author":5,"featured_media":0,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[],"class_list":["post-43175","post","type-post","status-publish","format-standard","hentry","category-uncategorized"],"yoast_head":"<!-- This site is optimized with the Yoast SEO Premium plugin v21.5 (Yoast SEO v21.5) - https:\/\/yoast.com\/wordpress\/plugins\/seo\/ -->\n<title>\u4f7f\u7528Prometheus\u76d1\u89c6Kubernetes\u7684GPU\u8282\u70b9 - Blog - Silicon Cloud<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/www.silicloud.com\/zh\/blog\/\u4f7f\u7528prometheus\u76d1\u89c6kubernetes\u7684gpu\u8282\u70b9\/\" \/>\n<meta property=\"og:locale\" content=\"zh_CN\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"\u4f7f\u7528Prometheus\u76d1\u89c6Kubernetes\u7684GPU\u8282\u70b9\" \/>\n<meta property=\"og:description\" content=\"\u9996\u5148 \u5728\u8fd1\u5e74\u7684\u6df1\u5ea6\u5b66\u4e60\u70ed\u6f6e\u4e2d\uff0c\u6211\u4eec\u7ecf\u5e38\u542c\u5230\u4f7f\u7528Kubernetes\u7ba1\u7406\u591a\u4e2aGPU\u8282\u70b9\u5e76\u8fdb\u884c\u591a\u4eba\u5171\u4eab\u7684\u64cd\u4f5c\u65b9\u5f0f\u3002 [&hellip;]\" \/>\n<meta property=\"og:url\" content=\"https:\/\/www.silicloud.com\/zh\/blog\/\u4f7f\u7528prometheus\u76d1\u89c6kubernetes\u7684gpu\u8282\u70b9\/\" \/>\n<meta property=\"og:site_name\" content=\"Blog - Silicon Cloud\" \/>\n<meta property=\"article:published_time\" content=\"2023-09-21T03:16:11+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2024-04-28T17:49:45+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/cdn.silicloud.com\/blog-img\/blog\/img\/657d513737434c4406cc9247\/6-0.png\" \/>\n<meta name=\"author\" content=\"\u6e05, \u5b87\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"\u4f5c\u8005\" \/>\n\t<meta name=\"twitter:data1\" content=\"\u6e05, \u5b87\" \/>\n\t<meta name=\"twitter:label2\" content=\"\u9884\u8ba1\u9605\u8bfb\u65f6\u95f4\" \/>\n\t<meta name=\"twitter:data2\" content=\"2 \u5206\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\/\/www.silicloud.com\/zh\/blog\/%e4%bd%bf%e7%94%a8prometheus%e7%9b%91%e8%a7%86kubernetes%e7%9a%84gpu%e8%8a%82%e7%82%b9\/\",\"url\":\"https:\/\/www.silicloud.com\/zh\/blog\/%e4%bd%bf%e7%94%a8prometheus%e7%9b%91%e8%a7%86kubernetes%e7%9a%84gpu%e8%8a%82%e7%82%b9\/\",\"name\":\"\u4f7f\u7528Prometheus\u76d1\u89c6Kubernetes\u7684GPU\u8282\u70b9 - Blog - Silicon Cloud\",\"isPartOf\":{\"@id\":\"https:\/\/www.silicloud.com\/zh\/blog\/#website\"},\"datePublished\":\"2023-09-21T03:16:11+00:00\",\"dateModified\":\"2024-04-28T17:49:45+00:00\",\"author\":{\"@id\":\"https:\/\/www.silicloud.com\/zh\/blog\/#\/schema\/person\/1a6ecd3d914d22a5ac32791ffc1fbd8e\"},\"breadcrumb\":{\"@id\":\"https:\/\/www.silicloud.com\/zh\/blog\/%e4%bd%bf%e7%94%a8prometheus%e7%9b%91%e8%a7%86kubernetes%e7%9a%84gpu%e8%8a%82%e7%82%b9\/#breadcrumb\"},\"inLanguage\":\"zh-Hans\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/www.silicloud.com\/zh\/blog\/%e4%bd%bf%e7%94%a8prometheus%e7%9b%91%e8%a7%86kubernetes%e7%9a%84gpu%e8%8a%82%e7%82%b9\/\"]}]},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/www.silicloud.com\/zh\/blog\/%e4%bd%bf%e7%94%a8prometheus%e7%9b%91%e8%a7%86kubernetes%e7%9a%84gpu%e8%8a%82%e7%82%b9\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"\u9996\u9875\",\"item\":\"https:\/\/www.silicloud.com\/zh\/blog\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"\u4f7f\u7528Prometheus\u76d1\u89c6Kubernetes\u7684GPU\u8282\u70b9\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/www.silicloud.com\/zh\/blog\/#website\",\"url\":\"https:\/\/www.silicloud.com\/zh\/blog\/\",\"name\":\"Blog - Silicon Cloud\",\"description\":\"\",\"inLanguage\":\"zh-Hans\"},{\"@type\":\"Person\",\"@id\":\"https:\/\/www.silicloud.com\/zh\/blog\/#\/schema\/person\/1a6ecd3d914d22a5ac32791ffc1fbd8e\",\"name\":\"\u6e05, \u5b87\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-Hans\",\"@id\":\"https:\/\/www.silicloud.com\/zh\/blog\/#\/schema\/person\/image\/\",\"url\":\"https:\/\/secure.gravatar.com\/avatar\/4b2016c18459a605fc469c7566608f5686491baa112d0871ee613f61b7210565?s=96&d=mm&r=g\",\"contentUrl\":\"https:\/\/secure.gravatar.com\/avatar\/4b2016c18459a605fc469c7566608f5686491baa112d0871ee613f61b7210565?s=96&d=mm&r=g\",\"caption\":\"\u6e05, \u5b87\"},\"url\":\"https:\/\/www.silicloud.com\/zh\/blog\/author\/qingyu\/\"},{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-Hans\",\"@id\":\"https:\/\/www.silicloud.com\/zh\/blog\/%e4%bd%bf%e7%94%a8prometheus%e7%9b%91%e8%a7%86kubernetes%e7%9a%84gpu%e8%8a%82%e7%82%b9\/#local-main-organization-logo\",\"url\":\"\",\"contentUrl\":\"\",\"caption\":\"Blog - Silicon Cloud\"}]}<\/script>\n<!-- \/ Yoast SEO Premium plugin. -->","yoast_head_json":{"title":"\u4f7f\u7528Prometheus\u76d1\u89c6Kubernetes\u7684GPU\u8282\u70b9 - Blog - Silicon Cloud","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/www.silicloud.com\/zh\/blog\/\u4f7f\u7528prometheus\u76d1\u89c6kubernetes\u7684gpu\u8282\u70b9\/","og_locale":"zh_CN","og_type":"article","og_title":"\u4f7f\u7528Prometheus\u76d1\u89c6Kubernetes\u7684GPU\u8282\u70b9","og_description":"\u9996\u5148 \u5728\u8fd1\u5e74\u7684\u6df1\u5ea6\u5b66\u4e60\u70ed\u6f6e\u4e2d\uff0c\u6211\u4eec\u7ecf\u5e38\u542c\u5230\u4f7f\u7528Kubernetes\u7ba1\u7406\u591a\u4e2aGPU\u8282\u70b9\u5e76\u8fdb\u884c\u591a\u4eba\u5171\u4eab\u7684\u64cd\u4f5c\u65b9\u5f0f\u3002 [&hellip;]","og_url":"https:\/\/www.silicloud.com\/zh\/blog\/\u4f7f\u7528prometheus\u76d1\u89c6kubernetes\u7684gpu\u8282\u70b9\/","og_site_name":"Blog - Silicon Cloud","article_published_time":"2023-09-21T03:16:11+00:00","article_modified_time":"2024-04-28T17:49:45+00:00","og_image":[{"url":"https:\/\/cdn.silicloud.com\/blog-img\/blog\/img\/657d513737434c4406cc9247\/6-0.png"}],"author":"\u6e05, \u5b87","twitter_card":"summary_large_image","twitter_misc":{"\u4f5c\u8005":"\u6e05, \u5b87","\u9884\u8ba1\u9605\u8bfb\u65f6\u95f4":"2 \u5206"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/www.silicloud.com\/zh\/blog\/%e4%bd%bf%e7%94%a8prometheus%e7%9b%91%e8%a7%86kubernetes%e7%9a%84gpu%e8%8a%82%e7%82%b9\/","url":"https:\/\/www.silicloud.com\/zh\/blog\/%e4%bd%bf%e7%94%a8prometheus%e7%9b%91%e8%a7%86kubernetes%e7%9a%84gpu%e8%8a%82%e7%82%b9\/","name":"\u4f7f\u7528Prometheus\u76d1\u89c6Kubernetes\u7684GPU\u8282\u70b9 - Blog - Silicon Cloud","isPartOf":{"@id":"https:\/\/www.silicloud.com\/zh\/blog\/#website"},"datePublished":"2023-09-21T03:16:11+00:00","dateModified":"2024-04-28T17:49:45+00:00","author":{"@id":"https:\/\/www.silicloud.com\/zh\/blog\/#\/schema\/person\/1a6ecd3d914d22a5ac32791ffc1fbd8e"},"breadcrumb":{"@id":"https:\/\/www.silicloud.com\/zh\/blog\/%e4%bd%bf%e7%94%a8prometheus%e7%9b%91%e8%a7%86kubernetes%e7%9a%84gpu%e8%8a%82%e7%82%b9\/#breadcrumb"},"inLanguage":"zh-Hans","potentialAction":[{"@type":"ReadAction","target":["https:\/\/www.silicloud.com\/zh\/blog\/%e4%bd%bf%e7%94%a8prometheus%e7%9b%91%e8%a7%86kubernetes%e7%9a%84gpu%e8%8a%82%e7%82%b9\/"]}]},{"@type":"BreadcrumbList","@id":"https:\/\/www.silicloud.com\/zh\/blog\/%e4%bd%bf%e7%94%a8prometheus%e7%9b%91%e8%a7%86kubernetes%e7%9a%84gpu%e8%8a%82%e7%82%b9\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"\u9996\u9875","item":"https:\/\/www.silicloud.com\/zh\/blog\/"},{"@type":"ListItem","position":2,"name":"\u4f7f\u7528Prometheus\u76d1\u89c6Kubernetes\u7684GPU\u8282\u70b9"}]},{"@type":"WebSite","@id":"https:\/\/www.silicloud.com\/zh\/blog\/#website","url":"https:\/\/www.silicloud.com\/zh\/blog\/","name":"Blog - Silicon Cloud","description":"","inLanguage":"zh-Hans"},{"@type":"Person","@id":"https:\/\/www.silicloud.com\/zh\/blog\/#\/schema\/person\/1a6ecd3d914d22a5ac32791ffc1fbd8e","name":"\u6e05, \u5b87","image":{"@type":"ImageObject","inLanguage":"zh-Hans","@id":"https:\/\/www.silicloud.com\/zh\/blog\/#\/schema\/person\/image\/","url":"https:\/\/secure.gravatar.com\/avatar\/4b2016c18459a605fc469c7566608f5686491baa112d0871ee613f61b7210565?s=96&d=mm&r=g","contentUrl":"https:\/\/secure.gravatar.com\/avatar\/4b2016c18459a605fc469c7566608f5686491baa112d0871ee613f61b7210565?s=96&d=mm&r=g","caption":"\u6e05, \u5b87"},"url":"https:\/\/www.silicloud.com\/zh\/blog\/author\/qingyu\/"},{"@type":"ImageObject","inLanguage":"zh-Hans","@id":"https:\/\/www.silicloud.com\/zh\/blog\/%e4%bd%bf%e7%94%a8prometheus%e7%9b%91%e8%a7%86kubernetes%e7%9a%84gpu%e8%8a%82%e7%82%b9\/#local-main-organization-logo","url":"","contentUrl":"","caption":"Blog - Silicon Cloud"}]}},"_links":{"self":[{"href":"https:\/\/www.silicloud.com\/zh\/blog\/wp-json\/wp\/v2\/posts\/43175","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.silicloud.com\/zh\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.silicloud.com\/zh\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.silicloud.com\/zh\/blog\/wp-json\/wp\/v2\/users\/5"}],"replies":[{"embeddable":true,"href":"https:\/\/www.silicloud.com\/zh\/blog\/wp-json\/wp\/v2\/comments?post=43175"}],"version-history":[{"count":2,"href":"https:\/\/www.silicloud.com\/zh\/blog\/wp-json\/wp\/v2\/posts\/43175\/revisions"}],"predecessor-version":[{"id":72881,"href":"https:\/\/www.silicloud.com\/zh\/blog\/wp-json\/wp\/v2\/posts\/43175\/revisions\/72881"}],"wp:attachment":[{"href":"https:\/\/www.silicloud.com\/zh\/blog\/wp-json\/wp\/v2\/media?parent=43175"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.silicloud.com\/zh\/blog\/wp-json\/wp\/v2\/categories?post=43175"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.silicloud.com\/zh\/blog\/wp-json\/wp\/v2\/tags?post=43175"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}