discussion 2026-11-19 1206 views 16 votes

Multi-region Redis for idempotency keys — worth it?

@kai_sre
SRE

Hit this after a routine bump.

Symptom: Multi-region Redis for idempotency keys — worth it?

TimeoutError / pool exhausted / upstream reset
(versions in next reply if needed)

Env: Python 3.12, FastAPI recent, uvicorn workers≥2.
What should I check first?

9 replies

@piotr
2026-11-19 09:45 UTC · 2 votes

marking solved after we pinned the dependency and re-smoked /docs.

@helen_go
2026-11-19 10:42 UTC · 10 votes

minimal repro with ASGITransport would help more than screenshots.

@yuna
2026-11-19 17:15 UTC · 9 votes

that traceback looks like Engine created per request — move it to lifespan.

@cleo_dev29
2026-11-20 10:13 UTC · 5 votes

dropping pool_size per worker fixed it for us on the same box size.

@nik
2026-11-20 20:40 UTC · 7 votes

minimal repro with ASGITransport would help more than screenshots.

@mira_py
2026-11-21 16:20 UTC · 1 votes

check proxy_read_timeout before blaming uvicorn resets.

@omar
2026-11-21 20:58 UTC · 5 votes

paste engine kwargs and max_connections. worker×pool is the usual culprit.

@raul
2026-11-22 06:12 UTC · 8 votes

I would put rate limits at the edge and keep app middleware as a second line.

@jon_async
2026-11-22 14:52 UTC · 3 votes

that traceback looks like Engine created per request — move it to lifespan.

Reply

Login or register to reply.