What this tests

monitoringsemantics

Point your client at this specimen and find out — before prod does — whether it surfaces a real signal, or reports green while users suffer.

Anatomy

The process is dead. The database connection pool drained an hour ago. The queue consumer hasn't acked a message since Tuesday. But the /healthz endpoint — a different codepath, wired to a handler that only checks whether the binary still executes — returns a beautiful, gleaming 200 {"status":"ok"}. Every load balancer that ever kept routing to a corpse did so because of this specimen. Shallow health checks report liveness; your clients needed readiness. The difference is a page at 4am.

Replay it

curl -i https://faultydata.com/r/FD-2026-103

// response appears here

Signature

{Status:200 Headers:map[Content-Type:application/json] Body:{"status":"ok","checks":{"database":"ok","queue":"ok","cache":"ok"},"note":"all checks are stub implementations; nothing is ok"} Corrupt: Delay:<nil> Flaky:<nil> Sequence:<nil> Drip:<nil> StallAfter:0 Hang:false Reset:false Raw: Phases:[]}

Keep digging

FD-2026-023 — Global 502s from a bad WAF rule (Cloudflare, Jul 2019) FD-2026-048 — Origin unreachable (Cloudflare 523) FD-2026-091 — Control plane loses its buildings (Cloudflare, Nov 2023) FD-2026-096 — The endpoint that never came back (Code Spaces, Jun 2014) FD-2026-105 — HTTP 520 — the Cloudflare unknown-error junk drawer FD-2026-054 — Stale cache serving yesterday's answer