Nomad Best Practices for Reliable Deploys
summary
Defines concrete Nomad settings and tooling: force placement with constraint "distinct_host = true" and spread across datacenters, health checks on HTTP every 10s with restart limit 2 and start delay 15s, reschedule delay 30s, update parallelism 2 (example scaled to 10 instances). Demonstrates testing with Toxiproxy fault injection and dummy services, uses Consul service routers for blue/green and canary traffic control, and autoscaling via Sherpa job scaling plus Camtrail autoscale nodes — shown to reschedule and scale automatically.
AI-written summary · about
chapters (17)
- 0:00Introduction
- 4:51Local Job
- 5:57Multiple Instances
- 6:25Two Instances
- 7:19Spread
- 9:14Metadata
- 10:02Console
- 10:41Service of Name
- 11:35Check Limits
- 13:01Delay modes
- 15:40Deploying
- 17:53Canary Releases
- 18:26Tags
- 18:53Templates
- 19:30Console Routing
- 20:21Sherpa
- 21:26CamTrail
related talks
44mNomad, orchestration made in HashiCorp
35mLiving the Nomadic life
33mAbstracting Nomad: Patterns for Deploying to Nomad at Scale
25mAutoscaling Your Nomad Jobs Using SRE Insights
01h 06mSchedule AI workloads efficiently on Nomad with node pools
30mAutoscaling with HashiCorp Nomad
22mRainbow deployments with Nomad
01h 11mCrack the monolith with Consul and Nomad