AI
AI-Related Exporters in a Kubernetes Cluster
· β˜• 17 min read
1. dcgm-exporter dcgm-exporter is the GPU monitoring tool provided by the official NVIDIA community. Project address https://github.com/NVIDIA/dcgm-exporter 1.1 Installation Add the Helm chart repository 1 helm repo add gpu-helm-charts https://nvidia.github.io/dcgm-exporter/helm-charts 1 helm repo update Install 1 2 3 4 5 helm install dcgm-exporter gpu-helm-charts/dcgm-exporter --namespace monitor --create-namespace \ --set serviceMonitor.enabled=false \ --set image.

Large Model Inference with Triton Server and TensorRT-LLM in a Container
· β˜• 5 min read
1. Compiling Models with TensorRT-LLM 1.1 Introduction to TensorRT-LLM When using TensorRT, you usually need to convert the model to ONNX format, then convert the ONNX to TensorRT format, and finally run inference in TensorRT or Triton Server. But this conversion process is not simple, and you often run into

Basic Usage of nvidia-smi
· β˜• 9 min read
1. What Is nvidia-smi nvidia-smi stands for NVIDIA System Management Interface, the interface NVIDIA provides for managing and monitoring GPUs. nvidia-smi calls NVML. NVML stands for NVIDIA Management Library, a library that provides a set of C APIs for NVIDIA GPU monitoring and management. 1.1 Queryable State ECC error counts GPU utilization Active compute processes Clocks and PState Temperature and fan speed Power management Hardware identification 1.

Managing Data in Kubernetes with Fluid and JuiceFS
· β˜• 7 min read
1. Introduction to Fluid Below is the Fluid architecture diagram from https://github.com/fluid-cloudnative/fluid: Fluid abstracts two concepts: Dataset, a collection of data, the abstraction from the user’s point of view Runtime, the abstraction of the real services such as data storage and acceleration Fluid mainly solves the problems of using traditional caching systems on Kubernetes:

Distributing Cluster Images with Dragonfly V2
· β˜• 10 min read
1. Introduction to Dragonfly The Dragonfly documentation at https://d7y.io/zh/docs/ already covers things in detail. Here is just a brief introduction to the main components of V2: Manager, which provides the UI, user management, cluster monitoring, task management, and other features Scheduler, which schedules traffic between Peers and provides preheating and other features Seed Peer, the back-to-source node used to download data from origin sites (Harbor, Docker.