Learn by example
These pages are examples you can deploy on Datasance PoT. Each example stands on its own.
Two examples are here today. Later examples, including control plane deploys, Edgelet node deploys, and smaller microservices, will each get their own folder.
These examples pull Datasance-published images on both sites. apiVersion still follows your site. Image names do not.
| Artifact | Reference |
|---|---|
| Services | ghcr.io/datasance/pot-edge-patterns/<service>:1.2.0 |
| Model repo | datasance/pot-edge-patterns/models/wafer-defect |
| Model tag | 1.2.0 |
Services: sensor-simulator, smart-anomaly-detector, diagnostic-service, alert-dashboard, frame-generator, object-detector, snapshot-service.
Default Edgelet node names are site-a, ops-b, and edge-c. You can change agent.name. Deploy with potctl deploy -f.
Manifests live in pot-edge-patterns under deploy/steps/. Run deploy commands from that repo root.
Flow
Read wafer first if you want inference and a gallery. Read the plant first if you want factory telemetry, a second NATS account, and a Router diagnose call. The last plant page is optional and needs both.
| Page | Reader does | Already in place |
|---|---|---|
| Edge AI wafer defect | Inference and a gallery | An Edgelet node you will call edge-c |
| Acme Smart Plant | Factory telemetry, a second NATS account, and a Router diagnose call | Edgelet nodes site-a and ops-b |
| Fab gallery link | Optional. Set SNAPSHOT_URL on alert-dashboard | Both examples deployed, and the snapshot service bridge port |
Wafer does not require the plant. The plant does not require wafer.
If you do not have a control plane yet, start with Quick start or platform deployment.
Edge AI wafer defect Acme Smart Plant