Large Scale Deployments for Automated Kernel Testing
summary
Describes running LAVA-based automated kernel testing at rack scale with ~200 devices across 17 racks and multiple isolated instances including LKFT. Infrastructure fixes raised reliability from ~30% failure to under 1% by switching to FTDI serial cables, shielded USB, industrial Cambree Onyx hubs with per-port power/control, SNMP-controlled PDUs and serial concentrators. Covers failed SD‑MUX attempts, solder mods, staging instances for firmware testing, and a plan to pack 16 ninety‑six‑board form factor units into a single 1U enclosure.
AI-written summary · about
chapters (9)
- 0:00Intro
- 1:34Introduction to LAVA
- 3:06LAVA Device Challenges
- 7:04What if the board gets bricked?
- 10:49LAVA Devices - Connectivity
- 17:58LAVA - Physical Deployment
- 19:48LAVA- Administration
- 23:29LAVA - Staging instances
- 25:11LAVA-Large Scale Deployments
related talks
40mBoF: Automated Testing and Board Farming
58mBuilding a Bards Farm: Continuous Integration and Remote Control
37mAutomated Testing and Continuous Integration/Delivery: Lessons Learned
41mIntroducing the “Lab in a Box” Concept
44mAutomation beyond Testing and Embedded System Validation
36mFarming Together
45mUpdate on LAVA Testing for bare metal systems
24mContinuously Integrating the Upstream Linux Kernel on Hardware