Elevated API latency
Sep 12, 2026 · 38 minBetween 14:02 and 14:40 UTC roughly 12% of API requests took longer than 5 seconds. Cause: a slow upstream in the image validation pool held worker slots. Fix: the image fetch now runs in a bounded queue with its own timeout and the pool was widened.
- 14:40 Resolved — latency back to baseline, backlog drained.
- 14:21 Identified — image validation queue saturated; workers widened.
- 14:07 Investigating — 5xx rate and p95 latency alerts fired.