45559c0e-d9d4-4741-94b7-ca25b55059d3
Forgeon Docs

Documentation, guides, and patterns to help you build on Forgeon — from your first deploy to running serious infrastructure.

Runtime metrics

Metrics for running services, grouped by environment and instance.

Last updated: 2026-03-17

Runtime metrics show live health of each environment and instance.

What you can filter by

  • Environment (prod / staging / preview)
  • Instance (latest or older)
  • Status (running vs stopped)

What you can see

  • Requests, latency, errors
  • CPU and memory
  • Egress and bandwidth

Reading runtime metrics effectively

  1. Start with error rate and latency to detect user impact first.
  2. Correlate with CPU/memory to decide scale vs code fix.
  3. Compare across environments (staging vs prod) after each deployment.
  4. Check instance-level divergence to detect one unhealthy replica.

Suggested SLO starter targets

| Metric | Starter target | | --- | --- | | Availability | >= 99.9% monthly | | p95 latency | <= 500 ms for critical APIs | | 5xx error rate | < 1% sustained | | Restart loops | 0 recurring loops |

Troubleshooting

No data

  • Confirm the instance is receiving traffic
  • Wait for metrics sampling interval

Latency up but CPU normal

  • Inspect downstream dependencies and network path (DB/API/DNS).

Errors up right after deploy

  • Compare with previous deployment and rollback if needed.

See also