Famous outages, told hop by hop. Every outage took a route.
中文：一座著名故障博物馆，每次宕机都按它走过的路线讲。中文版：failrouter.com/zh
Each exhibit is one outage drawn as the route it took: a small change, a few hops, then a very bad day. All 12 are condensed from a public postmortem or investigation report, and each card names its source.
| Exhibit | The route in one line |
|---|---|
| CrowdStrike outage, July 2024 | One content update, about 8.5 million Windows machines boot-looping on a blue screen |
| Facebook outage, October 2021 | A maintenance command cut every backbone link; DNS withdrew its routes for about 6 hours |
| AWS S3 outage, February 2017 | One mistyped playbook input, S3 in us-east-1 down for about 4 hours |
Then walk the 3D hall, where every exhibit hangs in a gallery next to a sculpture of its impact score, or read how impact is scored on five dimensions.
| Project | What it does | Stack | ⭐ |
|---|---|---|---|
| failrouter · live ↗ | Museum of Failed Routes: famous outages, told hop by hop. | HTML | |
- ✅ Merged #11 docs: one activity chart instead of three; CI and last-commit badges in failrouter · 2026-10-10
- ⬆️ Pushed to
mainin failrouter · 2026-10-10 - ⬆️ Pushed to
mainin failrouter · 2026-10-09 - ✅ Merged #10 README: project activity charts (snake, heatmap, 3D) in failrouter · 2026-10-09
- ⬆️ Pushed to
mainin failrouter · 2026-09-28 - ✅ Merged #9 site: impact sculpture fills red from the bottom, brass top ring, room… in failrouter · 2026-09-28
Know a public postmortem that belongs here? Open an issue with the link, or follow Add an exhibit and send a PR. Two rules: only facts from the report, and describe systems and decisions, not people.
If an exhibit ever helped you explain a risk to your team, a ⭐ on FailRouter/failrouter helps the next SRE find the museum.
Charts, the repository table and the activity list are regenerated by GitHub Actions with github-action-shares.