Skip to main content
Enable -monitor, confirm the key health signals, and check that the node is earning. For full observability (Grafana dashboards, alerting, DCGM exporter), see the Metrics & Monitoring guide.

​
Enable Prometheus metrics

Add -monitor to your go-livepeer startup command: 
livepeer \
  -network arbitrum-one-mainnet \
  -ethUrl <YOUR_RPC_URL> \
  -orchestrator \
  -transcoder \
  -nvidia 0 \
  -monitor \
  ...
Metrics are exposed at: 
http://localhost:7935/metrics
Verify with: 
curl -s http://localhost:7935/metrics | head -20

​
Key metrics to watch

For AI inference metrics, also watch ai_request_count and ai_request_latency_seconds.

​
Check Explorer

After each round (~24 hours), verify on Livepeer Explorer:
  1. Status — should show Active
  2. Reward calls — should show a reward for each round your node was online
  3. Fees earned — will appear after you process transcoding or AI jobs
Missing reward calls usually point to one of these causes:
  • Your node was online and connected during the round
  • The startup command omits -reward=false
  • Your Arbitrum ETH balance is sufficient for gas

​
GPU monitoring

Check GPU health and utilisation: 
# Current status
nvidia-smi

# Continuous monitoring (updates every 2 seconds)
nvidia-smi -l 2

# CSV output for logging: utilisation, memory, temperature
nvidia-smi --query-gpu=utilization.gpu,utilization.memory,memory.used,memory.total,temperature.gpu --format=csv -l 5
Watch for:
  • Temperature above 85°C — indicates inadequate cooling
  • Memory utilisation near 100% — often causes out-of-memory errors for AI workloads
  • GPU utilisation at 0% when you should be processing jobs — check go-livepeer logs

​
Quick symptom guide

For detailed troubleshooting, see the FAQ.

​
Minimal Prometheus setup

To scrape and store metrics, add a prometheus.yml: 
global:
  scrape_interval: 15s

scrape_configs:
  - job_name: 'go-livepeer'
    static_configs:
      - targets: ['localhost:7935']
Run Prometheus with Docker: 
docker run -d \
  --name prometheus \
  --network host \
  -v $(pwd)/prometheus.yml:/etc/prometheus/prometheus.yml \
  prom/prometheus
Access the Prometheus UI at http://localhost:9090.

​
Next steps

Metrics & Monitoring

Full observability setup with Grafana dashboards and alerting.

Explorer Guide

Navigate the Livepeer Explorer to track your performance.

Troubleshooting

Common errors and fixes for orchestrator operations.

Community Tools

Third-party dashboards and monitoring tools.
Last modified on March 16, 2026