How Nomad 1.3 Enables Running Work at the Edge
summary
Nomad 1.3 adds concrete edge changes: a task-group parameter max_client_disconnect lets Nomad wait (default up to one hour) before marking nodes lost so an allocation becomes unknown instead of being restarted, letting clients reconnect without churn; a native service-discovery provider block (nomad or consul) plus an optional address override exposes service registration for edge clients; demo shows doom game servers staying live while Prometheus/Grafana scrape remote clients. CSI graduates to GA and a new evaluations UI panel improves debugging.
AI-written summary · about
chapters (14)
- 0:00Introduction to Nomad
- 2:29What is Nomad
- 4:58Edge Computing Explained
- 8:05Edge Markets Overview
- 10:11Cluster Topologies
- 11:37Nomad Networking
- 15:30Nomad 1.3 Edge Features
- 18:18Live Demo Setup
- 20:46Service Discovery Demo
- 23:55Monitoring with Grafana
- 25:38Game Server Deployment
- 28:06Handling Network Disconnects
- 33:35Other Nomad 1.3 Features
- 35:18Resources and Q&A
