Edge Gateways & Compute Nodes
Edge gateways aggregate sensor data, run local ML inference, and host flows in tools like Node-RED. They need routine maintenance: editing a flow, restarting a model server, or checking a pipeline, often on constrained hardware in a hard-to-reach spot. ThinRemote makes each node reachable on demand without leaving anything listening on the public internet.
The problem
These nodes usually sit on private or cellular networks with no inbound access, so the local dashboards are invisible from your desk. Running a permanent VPN or an exposed port on a low-resource device wastes scarce CPU and widens the attack surface. Field visits for a config tweak are not worth the trip.
How ThinRemote fits
- Open the Node-RED editor (or Grafana, or a Jupyter server) in your browser through an on-demand HTTP tunnel that only exists while you need it:
thinr device http edge-node-17 1880- Restart a local inference service or data collector without a full reboot:
thinr device exec edge-node-17 "systemctl restart inference-server"- Tail the aggregation pipeline live to confirm data is flowing:
thinr device logs edge-node-17 -f- Drop in a shell for deeper work when a flow misbehaves:
thinr device console edge-node-17At fleet scale
Treat a gateway family as a product and roll a new model file or flow to every node, then restart the service across the group in batches:
thinr product push edge-fleet ./model-v3.onnx /opt/models/
thinr product exec edge-fleet "systemctl restart inference-server" -g region-eu