Overview

Full name
docker.io/nvidia/dcgm
Registry
docker.io
Namespace
nvidia
Repository
dcgm
Cached tags
59
Last synced
07/05/2026, 09:02 PM

Introduction

NVIDIA DCGM GPU 监控守护进程镜像,采集 GPU 利用率/温度/功耗指标,配合 dcgm-exporter 供 Prometheus 抓取。

Details

nvidia/dcgm 运行 NVIDIA Data Center GPU Manager 守护进程,采集 GPU 健康状态、利用率与 ECC 错误等指标。定位为 GPU 节点监控 DaemonSet,需 NVIDIA 驱动与 GPU 设备;K8s GPU NodePool DaemonSet + dcgm-exporter。

快速启动

docker run -d --name dcgm \
  --gpus all \
  docker.io/nvidia/dcgm:latest

推荐实践

docker run -d --name dcgm \
  --restart unless-stopped \
  --gpus all \
  -e DCGM_EXPORTER_LISTEN=:9400 \
  docker.io/nvidia/dcgm:3.3.5

核心参数说明

  • --gpus all — NVIDIA GPU 必需
  • 定位 — GPU 节点监控 agent
  • 集成 — dcgm-exporter → Prometheus
  • 部署 — GPU NodePool DaemonSet

Kubernetes

apiVersion: apps/v1
kind: DaemonSet
metadata:
  name: nvidia-dcgm
spec:
  selector:
    matchLabels:
      app: nvidia-dcgm
  template:
    metadata:
      labels:
        app: nvidia-dcgm
    spec:
      nodeSelector:
        nvidia.com/gpu: "true"
      containers:
        - name: dcgm
          image: docker.io/nvidia/dcgm:3.3.5
          securityContext:
            capabilities:
              add: [SYS_ADMIN]
---
apiVersion: v1
kind: Service
metadata:
  name: nvidia-dcgm
spec:
  selector:
    app: nvidia-dcgm
  ports:
    - port: 9400
      targetPort: 9400

配合 dcgm-exporter 暴露 Prometheus metrics;GPU NodePool 专用。

Latest tags (20)

View all 59 tags →
TagArchPushed AtSizeDigestSyncWishAction
docker.io/nvidia/dcgm
:4.5.2-1-ubi9
linux/amd64linux/arm64
2026/02/18 03:411.7 GBsha256:fb6391726c15
docker.io/nvidia/dcgm
:4.5.2-1-ubuntu22.04
linux/amd64linux/arm64
2026/02/18 02:581.7 GBsha256:32c33f3d65e9
docker.io/nvidia/dcgm
:4.4.2-1-ubuntu22.04
linux/amd64linux/arm64
2025/11/19 04:241.7 GBsha256:2408cab52703
docker.io/nvidia/dcgm
:4.4.1-2-ubi9
linux/amd64linux/arm64
2025/09/19 20:561.6 GBsha256:91e5e0825fc2
docker.io/nvidia/dcgm
:4.4.1-2-ubuntu22.04
linux/amd64linux/arm64
2025/09/19 03:591.6 GBsha256:b558c6bd6193
docker.io/nvidia/dcgm
:4.3.1-1-ubi9
linux/amd64linux/arm64
2025/08/08 01:40921.0 MBsha256:38d1f6ec94c6
docker.io/nvidia/dcgm
:4.3.1-1-ubuntu22.04
linux/amd64linux/arm64
2025/08/08 00:52933.9 MBsha256:c35799244ad6
docker.io/nvidia/dcgm
:4.1.1-1-ubuntu22.04
linux/amd64
2025/02/20 02:14654.9 MBsha256:14dab21e5800
docker.io/nvidia/dcgm
:4.1.1-1-ubi9
linux/amd64
2025/02/20 02:02902.8 MBsha256:18cc88ea3d9c
docker.io/nvidia/dcgm
:4.0.0-1-ubi9
linux/amd64linux/arm64
2025/01/16 00:50706.2 MBsha256:af5223fdc6ee
docker.io/nvidia/dcgm
:4.0.0-1-ubuntu22.04
linux/amd64linux/arm64
2025/01/16 00:43453.2 MBsha256:8d114f297bfe
docker.io/nvidia/dcgm
:3.3.9-1-ubi9
linux/amd64linux/arm64
2024/11/19 06:34987.5 MBsha256:0febd1c46bc2
docker.io/nvidia/dcgm
:3.3.9-1-ubuntu22.04
linux/amd64linux/arm64
2024/11/19 06:311001.8 MBsha256:b40226d395a3
docker.io/nvidia/dcgm
:3.3.8-1-ubi9
linux/amd64linux/arm64
2024/09/20 05:03982.7 MBsha256:fecd80d9bae8
docker.io/nvidia/dcgm
:3.3.8-1-ubuntu22.04
linux/amd64linux/arm64
2024/09/20 04:591001.1 MBsha256:a2e0e9660008
docker.io/nvidia/dcgm
:3.3.7-1-ubi9
linux/amd64linux/arm64
2024/07/24 22:35984.6 MBsha256:ce85f318b8c6
docker.io/nvidia/dcgm
:3.3.7-1-ubuntu22.04
linux/amd64linux/arm64
2024/07/24 22:311002.9 MBsha256:f38acedfeabc
docker.io/nvidia/dcgm
:3.3.6-1-ubi9
linux/amd64linux/arm64
2024/05/21 02:04988.6 MBsha256:75d9cdf9e744
docker.io/nvidia/dcgm
:3.3.6-1-ubuntu22.04
linux/arm64linux/amd64
2024/05/21 01:52603.0 MBsha256:295b1403463c
docker.io/nvidia/dcgm
:3.3.5-1-ubi9
linux/amd64linux/arm64
2024/04/04 01:20993.0 MBsha256:82a09bef3b2d