Downstream reports of increased go aways with "too_many_internal_resets" #856
Description
Sorry for the vague issue, but we often are the first to encounter regressions and I figure saying something is better than nothing :)
After upgrading from h2 0.4.7 -> 0.4.11 in astral-sh/uv#14899, we're getting reports of dramatically increased HTTP/2 errors with a trace like
Caused by:
0: client error (SendRequest)
1: http2 error
2: connection error detected: detected excessive load generating behavior (b"too_many_internal_resets")))) }
We happen to debug print error types for debugging our retry with nested error types:
Reqwest(
reqwest::Error {
kind: Request,
url: "https://<redacted>/repository/python/simple/markdown-it-py/",
source: hyper_util::client::legacy::Error(
SendRequest,
hyper::Error(
Http2,
Error {
kind: GoAway(
b"too_many_internal_resets",
ENHANCE_YOUR_CALM,
Library
)
}
)
)
}
)