Bare Metal DevOps
summary
Describes a working bare‑metal pipeline that boots racks via pixie > ipixie chainload, looks up machines by serial in a web service, and loads a squashfs root over HTTP then runs a Chef client configures RAID and disks step. Covers Ubuntu 12.04/14.04 testing 16.04 and systemd issues, persistent firmware and NIC bugs, 10Gb/40Gb cabling, 40Gb SFP optics ~$15,000, feedstock of a petabyte or two, ECC error seasonality (six‑week spring spike), Seagate supply cuts, no SANs (direct‑attached + NetApp homes), and a cost claim ~one‑third cheaper than AWS.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 2:21CORE PRINCIPLES
- 3:06AGILE INFRASTRUCTURE
- 4:37MACHINES AS CATTLE
- 6:10BOOT PROCESS
- 11:15AUTOMATION
- 13:19COMMUNICATION
- 15:24ADDITIONAL CONSIDERATIONS
- 15:48MECHANICAL SYMPATHY
- 20:09NETWORKING
- 23:46MONITORING
- 29:43FORECASTING
- 36:51GOTCHAS
- 39:34DEVOPS IS NOT
related talks
48mA road trip into AWS and back to bare metal
24mImage-Based Bare-Metal Provisioning With Metal3.io, Cluster API and…
28mHow to Deploy CF and Other Bosh Releases on Bare-Metal Environment
40mOperating a Self-Healing Bare-Metal Kubernetes Platform at Global Scale
29mLarge Scale Deployments for Automated Kernel Testing
22mFrom Bare Metal to Abstraction: A Tale to the Heart of Cloud-Native Computing
35mBare-Metal Chronicles: Intertwinement Of Tinkerbell, Cluster API And GitOps
51mUsing Chef to manage hardware-based infrastructure