Yocto Continuous Integration in a Kube
summary
Runs Yocto builds and on-target tests inside Kubernetes using Tekton pipelines, labgrid for board farm control, and centralized caches: an NFS-backed sstate cache and download cache with an HTTP frontend, plus a hash-equivalence server and PR server for package versioning. Uses local persistent volumes because NFS/rook were too slow, tunnels labgrid SSH over go-http-tunnel so cloud builds can test on-prem boards. Demonstration builds gzdoom images for multiple boards (including Rock Pi X), 24 builds finish in 1h8m31s (≈3 min avg) and tests pass.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 0:28Development Workflow
- 1:03Caching
- 8:26Testing Hardware
- 11:52Tekton
- 12:26Labgrid
- 13:30ancillary services
- 14:15workspace
- 15:44ssh connection
- 17:35gzdoom
- 18:08Build Management
- 18:48Demo
- 24:57Testing
related talks
48mLiving on Master: Using Yocto Project, Jenkins and LAVA for a Rolling Release
39mKubernetes-centric CI/CD
37mAutomated Testing and Continuous Integration/Delivery: Lessons Learned
52mCreating Continuous Delivery for Yocto Based IoT Distribution
27mKubernetes The Fun Way
29mProduction CI/CD w/CNBs: Tekton, Gitlab & CircleCI(plus), Oh My!
01h 00mJava on Kubernetes What I Wish I Knew First
32mHow we build OKD (Kubernetes distro) using Tekton