wptserve H2 tests flakily failing

[gsnedders]

I've seen this a few times recently; see e.g. this Azure Pipelines step. (Logs attached here for archival reasons: 209.txt.)

That job had:

FAILED wptserve/tests/functional/test_response.py::TestH2Response::test_raw_data_frame
FAILED wptserve/tests/functional/test_server.py::TestFileHandlerH2::test_not_handled
FAILED wptserve/tests/functional/test_server.py::TestRewriterH2::test_rewrite
FAILED wptserve/tests/functional/test_server.py::TestRequestHandlerH2::test_exception

They all failed with:

httpx.ReadTimeout: The read operation timed out

And had captured logs like:

----------------------------- Captured log setup ------------------------------
INFO     root:server.py:867 Starting http2 server on https://localhost:51070/
---------------------------- Captured log teardown ----------------------------
ERROR    root:server.py:426 (923a45a7) Closing Connection - 
[WinError 10054] An existing connection was forcibly closed by the remote host
INFO     root:server.py:885 Stopped http server on localhost:51070

Pretty sure I've also seen this on not-Windows too.

[Denis793]

Hello!
I think this timeout issue could be caused by the server not being fully ready or network delays during the test.
Possible solutions:

Increase the timeout settings in the httpx client during tests.

Add retries for requests that fail with ReadTimeout.

Insert a short sleep/delay after starting the server and before running tests.

I can try to help if you'd like me to suggest a concrete fix!