Consider a rate-limiting strategy: stress_small_msg is flaky
Open
- Dominant language
- Rust
- Stars
- 74
- Forks
- 11
- PR merge metrics
- No merged PRs in 30d
Description
Consider:
- sending back a choke message if we're running out of send channel buffer on receiving messages, or
- expanding the permit system to include single block messages as well, or
- some other possibly just straight up sender-side rate-limit solution?
So that we can slow down / provide back-pressure to the sender and not have to close connections because we are receiving too fast.
Contributor guide
Assessment
This issue has not been assessed yet.