Lab: timeouts are not deadlines¶
The scripts behind the numbers in the post. Every server runs in-process on loopback; nothing here needs a network.
uv venv --python 3.13 .venv
uv pip install --python .venv/bin/python "deadline-budget==0.1.3" "clientwright[httpx,deadline]==0.2.2" "grpc-client-kit[deadline]==0.1.0"
.venv/bin/python 01_read_timeout.py
.venv/bin/python 02_retry_multiplies.py
.venv/bin/python 03_grpc_chain.py
.venv/bin/python 04_deadline_header.py
| Script | What it measures |
|---|---|
01_read_timeout.py |
httpx timeout=1.0 against a body that drips for four seconds, against headers that stall, inside asyncio.timeout, and under clientwright's total |
02_retry_multiplies.py |
a hand-rolled three-attempt loop against a server that never answers, and the same attempts under one total |
03_grpc_chain.py |
gateway → orders → inventory, billing: the deadline each server saw, with a fresh timeout per call and with the gateway's deadline propagated; plus a 50 s budget against a 5 s configured call |
04_deadline_header.py |
the X-Deadline-Ms value an HTTP upstream receives on each attempt when the caller carries a request budget |