Skip to content

Conversation

@arpadav
Copy link
Contributor

@arpadav arpadav commented Apr 2, 2025

tracing has an unresolved bug when attempting to fetch a span that should exist. Tracking: tokio-rs/tracing#3223 (comment)

#134 introduces this bug. hyper contributors aren't at fault — this is a tracing issue.

As a result, crates that depend on hyper's legacy client (e.g. reqwest) will immediately panic when a request is sent under specific circumstances configured by tracing spans. (The tracked comment shows a pseudocode example). This slightly more accurate tracing span is not worth the panics.

In theory, this PR can be reverted once tracing resolves this issue, but its up to the maintainers whether or not it should be prioritized.

@joshka
Copy link
Contributor

joshka commented Apr 2, 2025

The impact of reverting this is that DNS requests will no longer be under the same span that tracks the request. This may affect tracing and metrics on any places where this info is tracked. That said, this downside is less problematic than the alternative which causes panics, so I'm for this being merged

@joshka
Copy link
Contributor

joshka commented Apr 2, 2025

Captured the overall details of the span in DNS problem in hyperium/hyper#3870

@seanmonstar seanmonstar merged commit 912dc78 into hyperium:master Apr 3, 2025
16 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants