Abstracting Nomad: Patterns for Deploying to Nomad at Scale
summary
SeatGeek runs Nomad at scale and moves from ad hoc HCL edits to a layered platform: an admission-style Knack admission proxy enforces policies, a rollout YAML→HCL2 API converts tiny 5–6 line app specs into full jobs and tails logs to CI, and secrets live in Vault KVv2 versioned secrets so each deploy pins a secret version. Observability uses injected sidecars and Docker labels, Consul Connect handles networking, and the platform tradeoff is fewer knobs for engineers but far simpler, more reliable deployments.
AI-written summary · about
chapters (13)
- 0:00Intro
- 1:37Who do we work with?
- 6:13Just a CLI Away
- 10:47The Interface
- 15:21Complaints
- 15:53Refreshed Approach
- 19:05Secrets: Prototype
- 19:49Secrets: Deli
- 21:06Secrets: Rollout
- 24:00Observability: v2
- 29:38Networking: Origins
- 30:32Networking: Deli
- 31:50Networking: Rollout
related talks
35mLiving the Nomadic life
27mNomad Best Practices for Reliable Deploys
30mA beginner's journey: Scaling self-hosted platform infrastructure with Nomad
22mRainbow deployments with Nomad
12mNomad: Lightweight Orchestration That Complements Kubernetes
39mBuilding Efficient Continuous Integration Pipelines with Nomad, Vault and Jenkins
44mNomad, orchestration made in HashiCorp
24mHow we Orchestrate 7M+ Builds/Month with Kubernetes - Danielle Tomlinson