NFD and My Adventure in the Cloud Native Project Jungle
summary
Node Feature Discovery (NFD) extends Kubernetes by detecting node hardware and software and advertising them as labels so the scheduler can place workloads; it runs daemonset workers per node and a gRPC master that updates Node labels/annotations, rediscovers every 60s, and enforces an allowlist. Feature sources include CPUID, power/turbo, kernel config, NUMA, PCI GPUs and a local hooks and feature files plugin mechanism; labels use the feature.node.kubernetes.io/ namespace. Deployments include master+workers, master with batch workers, or combined pods; mTLS is supported but certificate provisioning is manual.
AI-written summary · about
chapters (26)
- 0:00Intro
- 0:22Agenda
- 1:29Introduction
- 1:43How I Got Involved
- 4:41Resource Management in K8s
- 7:11What About Node Features?
- 9:01Node Feature Discovery
- 9:14NFD - Architecture
- 10:46NFD-Worker
- 12:56Feature Sources - CPU
- 14:58Feature Sources - Kernel
- 15:47Feature Sources - System
- 16:03Feature Sources - Memory
- 16:40Feature Sources - PCI
- 17:07Feature Sources - Network
- 17:48Feature Sources - IOMMU
- 18:07Feature Sources - Local
- 18:37Hooks
- 20:19Feature Sources - Feature Files
- 20:40Configuration Options
- 21:30NFD Deployment
- 23:15Workload Deployment - node Selector
- 23:41Workload Deployment - node Affinity
- 25:15Demo Time
- 29:07The (Near) Future
- 30:07Want To Help?
related talks
33mNFD and My Adventure in the Cloud Native Project Jungle
34mStreamline Your Cluster with Automated Node Labels,Taints, and Annotations
30mTaming Unbounded Resources with Node Feature Discovery
25mK8s Labels Everywhere! Decluttering With Node Profile Discovery.
26mMake Kubernetes Greater for NFV@Arm
50mFlying through Cloud Native (CNCF)
27mCloud Native Landscape for Node.js Developers
33mCNCF TAG-Runtime: Cloud Native Open...