What this tests
semantics
Point your client at this specimen and find out — before prod does — whether it interprets status codes and content types correctly, or makes things up.
Anatomy
A maintenance script, run with the wrong arguments, deleted around four hundred customers' entire sites instead of marking them for later — and the restore meant rebuilding each tenant by hand, one at a time, over two weeks. Not hours. Weeks. For a fortnight, affected companies' Jira answered every request with a 404 while standups quietly relocated to a shared spreadsheet and a prayer. This is the outage genre's extended director's cut: not a bang but a long, low hum of absence. No 500 with a retry hint, no degraded mode, no read-only fallback — just the void where the project management used to live, persisted at HTTP's most honest status code for the duration of a small mortgage. The specimen: a flat 404 with a body that suggests the site "may have been deleted". May. There's always hope in the error message. There was less in the calendar.
Source: Atlassian — April 2022 outage retrospective · 2022-04-04
Replay it
curl -i https://faultydata.com/r/FD-2026-031
// response appears hereSignature
{Status:404 Headers:map[] Body:{"error":"site not found","message":"This site may have been deleted or is temporarily unavailable"} Corrupt: Delay:<nil> Flaky:<nil> Sequence:<nil> Drip:<nil> StallAfter:0 Hang:false Reset:false Raw: Phases:[]}
Keep digging
FD-2026-063 — Queue collapse under demand (Ticketmaster Eras presale, Nov 2022) FD-2026-098 — Power surge kills the data centre, recovery kills the airline (British Airways, May 2017) FD-2026-004 — GitHub secondary rate limit FD-2026-008 — WAF challenge page instead of JSON (Cloudflare pattern) FD-2026-010 — Gateway timeout from the edge (Cloudflare 522 pattern) FD-2026-023 — Global 502s from a bad WAF rule (Cloudflare, Jul 2019)