With its addition to the CERN OpenStack service a few years ago, Ironic set out to ease the provisioning of bare metal servers in CERN IT. With an initial focus to offer end-users the same interfaces they’re used to from creating virtual machines, Ironic has meanwhile started to take over server life-cycle workflows and replace some of the in-house developed tools CERN IT had used before. This presentation will follow the entire life-cycle of a server in CERN IT from physical rack installation, over burn-in, benchmarking, and customisation (e.g. with the recently added support for software RAID in Ironic), to allocation, re-allocation, and retirement and see where and how Ironic helps with the various steps … and where further work is needed!
Attendees will learn how Ironic helps with the life-cycle management of physical servers in CERN IT: inventory verification, initial burn-in (to detect hardware issues), benchmarking (to see if the servers comply with the specification), customisation (in particular software RAID setup with Ironic), instance creation and allocation to users, re-allocation to other users (with intermediate cleaning), and server retirement.
