The Automotive Initiative Discussion
summary
Argues Kubernetes is unsuitable for cars because eventual consistency, heavy runtime activity and non-determinism break functional safety guarantees; instead a lightweight orchestrator (herte), Podman optimizations and quadlets run containers from systemd. Podman start is measured from ~2.0s down to ~0.3s on a low-end board. Work includes composeFS with dm-verity/fs-verity for corruption detection, ISO 26262 certification targeting ASIL-B, RHEL with RT kernel and 100% requirements-linked test coverage using code-coverage tooling.
AI-written summary · about
chapters (17)
- 0:00Introduction
- 1:05Containers on Wheels team
- 2:08Automotive testing challenges
- 3:31Kubernetes not suitable for cars
- 6:38Functional safety core challenge
- 9:04Safety certification evidence
- 12:12Linux functional safety beyond cars
- 17:06Boot time and UX requirements
- 20:17Vehicle security threats
- 23:05Functional safety certification layers
- 26:02ISO 26262 and SPICE standards
- 28:54Red Hat Enterprise Linux for automotive
- 30:27ASIL levels and stacks
- 32:13Security updates OTA
- 34:51Automotive use cases future
- 37:30Functional safety impact
- 38:57Q&A
related talks
28mLinux in Cars! CentOS AutoSD, SDVs, and more
01h 00mAutoSD: A Linux Development and Prototyping Framework for the Automotive Community
48mPanel Discussion: Do One Thing, and Do It Well: Special Purpose OSes Apply the Unix App Philosoph...
53mAutomotive Collaboration: What's Really Going On? Has Something Improved During the Last Year?
31mUnsung Hero Of the Cloud Native Revolution: Container Linux Then And Now
50mLinux You Can Drive My Car
35mIntroduction to AGL Architecture and Roadmap
34mOf Quadlets and Other Daemonless