[timers] Fix missing wakeups in timer service - #5093
Conversation
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: fde725cd21
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| /// | ||
| /// Since v1.7.3 | ||
| #[cfg_attr(feature = "schemars", schemars(skip))] | ||
| pub self_proposal_queue_memory_limit: NonZeroByteCount, |
There was a problem hiding this comment.
Add release notes for the new queue limit
This adds a user-facing worker.self-proposal-queue-memory-limit option and switches self-proposal buffering/backpressure to a per-partition memory budget, but the commit does not add a release-notes/unreleased entry. Operators upgrading will otherwise miss the new knob and its memory/latency trade-off, so please add an unreleased note for this behavior/configuration change.
AGENTS.md reference: AGENTS.md:L51-L51
Useful? React with 👍 / 👎.
Exactly as (#5085) but for the timer service. Changes to the scheduling were not correctly waking up the wakers for `poll_next_timer`. This was ok in the existing code because we were reconstructing the timer stream in every `LeaderState::run` call. Now that this is changing in the next PR, this has to be fixed. > For full disclaimer, this was caught and fixed by codex.
Exactly as (#5085) but for the timer service. Changes to the scheduling were not correctly waking up
the wakers for
poll_next_timer. This was ok in the existing code because we were reconstructing the timer stream in everyLeaderState::runcall.Now that this is changing in the next PR, this has to be fixed.