How eFishery Uses Nomad, Vault, and Consul for Rural Edge Deployments
summary
eFishery runs feeders and water sensors on ESP32 devices and forwards logs to a Raspberry Pi 2 ad server that runs Raspberry Pi 2 ad server running Nomad and Consul agents and an MQTT broker, using an offline-first pufferC sync to store 100-record rotations locally and push when online. Deployment uses Nomad jobs with the exec driver (no containers) to reduce RAM, claiming 180 MB Nomad agent footprint, Vault holds CI/CD keys, env and metadata secrets, and the fleet reports ~200 deployments/day, 500 jobs, 900 containers and 80 TB/month transfer.
AI-written summary · about
chapters (13)
- 0:00Introduction to eFishery
- 1:51Feeder machine problem
- 3:59Feeder machine features
- 6:09Control box and edge computing
- 8:38Architecture deep dive
- 10:45Nomad and Consul roles
- 13:13Memory usage optimization
- 13:57Why choose Nomad over Kubernetes
- 15:19HashiCorp stack selection
- 17:07Vault for secrets management
- 18:55Current deployment metrics
- 19:15Benefits of HashiCorp stack
- 20:59Conclusion and Q&A
related talks
40mNomad: Scheduling Applications at Scale
46mProgressive Deployments with Consul Service Mesh
25mAutoscaling Your Nomad Jobs Using SRE Insights
32mFuzzing Nomad with Nomad
32mLow Latency Serving of Offline Data: Efficient, Safe, and Reliable Data...
36mKnee Deep Into P2P
32mGetting distributed with Firenest
17mFun, interesting and common RabbitMQ use cases