This talk describes the architecture of a manufacturing test framework that Facebook is developing to validate the quality of OCP compute, storage, and rack products. Every rack product, from prototype to mass production, that enters Facebook's data center fleet is rigorously tested to identify and fix manufacturing defects. The talk will describe the main components of the framework including asset management, provisioning, job engine, scheduling, test cases, test suites, analytics, and monitoring. Testing focuses on compute and storage stress testing as well as configuration checks agains “golden” configs from both a Bill Of Material (BOM) and expected software version standpoints. We look forward to contribute Facebook's approach to manufacturing validation to the OCP community.
Link to Slides: http://files.opencompute.org/oc/public.php?service=files&t=da7f257111bd46567117e32155072892