Skip to content

fix(performance): update both feedback buffers - #1221

Merged
tmigone merged 1 commit into
graphprotocol:mainfrom
graphops:chris/qos-feedback-buffers
Nov 10, 2026
Merged

tmigone merged 1 commit into
graphprotocol:mainfrom
graphops:chris/qos-feedback-buffers

Conversation

@chriswessels

Copy link
Copy Markdown
Contributor

Actor::handle_msgs drains its batch while updating the first snapshot, so the second snapshot never receives query feedback. Readers falling back to it while the first buffer is write-locked can see stale performance and indexing progress.

Apply the complete batch to both buffers before clearing it. The regression forces reads through each buffer after successful and failed feedback and checks latency, success rate, and monotonic observed progress.

Validation (on this independent branch):

  • cargo +nightly-2026-06-11 fmt --all -- --check
  • cargo +1.95.0 clippy --all-targets --all-features -- -Dwarnings --force-warn deprecated --force-warn dead-code
  • cargo +1.95.0 test

@chriswessels
chriswessels marked this pull request as ready for review October 29, 2026 13:48
@tmigone
tmigone merged commit 5fcf376 into graphprotocol:main Nov 10, 2026
2 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.

2 participants