fix: revive the shared game ticker if its worker dies mid-game - #12
Open
Rl0007 wants to merge 1 commit into
Open
fix: revive the shared game ticker if its worker dies mid-game#12Rl0007 wants to merge 1 commit into
Rl0007 wants to merge 1 commit into
Conversation
Every live game is driven by a single shared, deduplicated ticker job. If that worker is lost mid-game — an OOM under a large answer flood on a small instance, a deploy or worker restart — nothing advances the games and they freeze exactly where they stood, with no exception to log (an OOM leaves no traceback). The host then only sees the game stall, or the abandoned-guard ends it early. Add ensure_ticker_running(), a deduplicated re-enqueue that is a no-op while the loop is alive and otherwise starts a fresh one that resumes every game from the Redis state it left behind. Drive it from the scheduler (every minute) and the host poll (recovery within a poll), so a killed ticker comes back without a human. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Problem
At ~1000 concurrent players a game freezes: the question shows, the timer hits 0, and the quiz stops advancing. Reproduced deterministically with a 1000-player load harness against a small (Frappe Cloud-sized) instance. The session sits
Activeand frozen (or gets force-ended by the abandoned-guard withcurrent_question = -1), and no error is logged.Cause
Every live game is driven by a single shared, deduplicated ticker job (
run_ticker,job_id="qz_ticker"). On a small instance, the 1000-answer flood at a question close pushes the worker over its memory limit and the kernel OOM-kills the process — which is why there's no Python traceback. Nothing re-enqueues the loop, so every game it was driving freezes exactly where it stood. (A deploy/worker restart has the same effect.)Fix
ensure_ticker_running()— a deduplicated re-enqueue that is a no-op while the loop is alive, and otherwise starts a fresh ticker that resumes every game from the Redis state it left behind. Driven from:get_host_state) — recovery within a single poll, since that's the most frequent caller.No state model changes; a revived ticker picks up each session's saved phase/
next_tsand continues.Tests
Added
TestTickerRecovery: revives (deduplicatedqz_tickerenqueue) when a game is live, no-op when none are.Not covered here
This restores the driver after a kill; it does not prevent the OOM. Reducing the per-close memory/DB footprint so the worker survives 1000 answers in the first place is a good follow-up.