pyth-lazer-agent fixes #2771
pyth-lazer-agent fixes #2771
Conversation
… drain pending updates
|
The latest updates on your projects. Learn more about Vercel for Git ↗︎
|
| let mut first_failure_time = Instant::now(); | ||
| let mut failure_count = 0; | ||
|
|
||
| loop { |
There was a problem hiding this comment.
Still doesn't use retry, I know. 😄
darunrs
left a comment
darunrs
left a comment
There was a problem hiding this comment.
LGTM. Can wait for one more approval to merge.
There was a problem hiding this comment.
Pull Request Overview
This PR refactors the relayer session and publisher to use a broadcast channel, adds a failure reset timer for backoff retries, fixes the default config file path, and ensures pending updates are drained.
- Switch mpsc channels to broadcast channels for fan-out to multiple relayer tasks
- Add a timer to reset backoff after 5 minutes without failures
- Update default config path and drain pending updates instead of clearing
Reviewed Changes
Copilot reviewed 3 out of 3 changed files in this pull request and generated 1 comment.
| File | Description |
|---|---|
| src/relayer_session.rs | Converted to broadcast::channel, made session struct public, added backoff reset logic, handled lagged receiver errors |
| src/main.rs | Changed default config file path to config/config.toml |
| src/lazer_publisher.rs | Replaced custom RelayerSender with broadcast sender, drained pending updates before publishing |
Comments suppressed due to low confidence (4)
pyth-lazer-agent/src/relayer_session.rs:61
- Exposing all fields of
RelayerSessionTaskpublicly can lead to misuse; consider adding a constructor or builder method and keeping fields private to maintain encapsulation.
pub struct RelayerSessionTask {
pyth-lazer-agent/src/relayer_session.rs:120
- [nitpick] The variable name
recv_resultis generic; consider renaming to something liketransaction_resultfor clearer intent.
recv_result = self.receiver.recv() => {
pyth-lazer-agent/src/relayer_session.rs:134
- Add unit tests to simulate
RecvError::LaggedandRecvError::Closedpaths to ensure logging and bail behavior works as expected.
broadcast::error::RecvError::Lagged(skipped_count) => {
pyth-lazer-agent/src/relayer_session.rs:78
- Consider adding tests to verify that the backoff strategy is reset after
FAILURE_RESET_TIMEhas elapsed.
const FAILURE_RESET_TIME: Duration = Duration::from_secs(300);
| return; | ||
| } | ||
| Err(e) => { | ||
| if first_failure_time.elapsed() > FAILURE_RESET_TIME { |
There was a problem hiding this comment.
Currently failure_count is only reset after the timer expires; on a successful WebSocket session you should also reset failure_count (and possibly first_failure_time) so that retries start fresh after each successful connection.
There was a problem hiding this comment.
A successful session will go indefinitely so I'll leave this as is.
Summary
Rationale
How has this been tested?