testing in production gathering actually useful data
summary
Describes a practical pre‑ and in‑production testing pipeline that verifies non‑breaking and state‑breaking changes to consensus, state machine and communication stacks by actually running networks. Teams build an initial Petri tool then replace it with Ironbird, which uses a GitHub app, Temporal workflows, DigitalOcean droplets, Tailscale and Grafana and ships prebuilt images so a devnet reaches usable telemetry in about 1 minute spin‑up. Features PR‑triggered networks, load tests, state replay from mainnet, and the hard lesson that tracking infra state without Terraform is painful.
AI-written summary · about
chapters (10)
- 0:00Testing challenges
- 1:55Pre-production testing needs
- 4:49Existing testing solutions
- 7:59First Petri iteration limits
- 9:24Ironbird and GitHub integration
- 11:15Speed and DX optimization
- 13:00User-friendly frontend
- 14:03Ironbird network config
- 15:08Ironbird future plans
- 16:33Project lessons learned
