Skip to main content
Run every check that applies to the configured node mode. A node passes verification when all applicable checks produce expected output. Proceed to Guides only after this page is complete.
Finish setup by confirming the node is on-chain, processing jobs, and exposing the minimum monitoring signals. Full monitoring setup (Grafana, alerting, DCGM exporter) is in .

​
Quick-reference checklist

​
Common issues

Check the GPU was detected at startup (Transcoding on Nvidia GPU in logs). Confirm the -nvidia flag matches the GPU device ID. Confirm port 8935 is open from external machines. Check the orchestrator logs for error messages on job receipt. See .
Check the AI runner container is running (docker ps). Check its logs for model loading errors (docker logs <container>). Confirm aiModels.json is valid JSON with correct field names. Ensure the model_id matches exactly (case-sensitive, including the org prefix). Confirm /var/run/docker.sock is mounted in the orchestrator container.
Confirm the activation transaction succeeded on Arbiscan. Check the service URI is set and reachable. Verify the stake bonding transactions (both approve and bond) confirmed. Recent activations appear on Explorer after the next round, because the active set updates once per round (~22 hours).
Confirm the startup command omits -reward=false. Check the Arbitrum ETH balance, because reward calls require gas. Confirm the node was online during round initialisation. Check logs for reward call errors. See for low-stake profitability analysis.
The #1 cause is port 8935 not reachable from the internet. Re-test from an external machine. The second cause is price above gateway caps - check current rates on tools.livepeer.cloud and compare. Third: not in the active set (for video jobs) - check rank on Explorer. AI jobs route by capability, not stake, so ensure capabilities appear on tools.livepeer.cloud.

​
Setup complete

The node is now on-chain, processing jobs, and verified. The setup section is complete. Pricing optimisation, delegation strategy, workload selection, and long-term monitoring are covered in the Guides section.

Staking and Earning

How orchestrators earn from service fees and LPT inflation rewards.

Pricing Strategy

Set competitive video and AI prices to attract work from gateways.

Monitoring and Tools

Full monitoring: Grafana dashboards, alerting, Explorer operations, troubleshooting.

Workload Options

Expand or change workloads based on hardware and network demand.
Last modified on March 16, 2026