fix(tasks,runners): critical requeue and finalization bugs#4008
Draft
cursor[bot] wants to merge 2 commits into
Draft
fix(tasks,runners): critical requeue and finalization bugs#4008cursor[bot] wants to merge 2 commits into
cursor[bot] wants to merge 2 commits into
Conversation
When every runner is at capacity, ErrAllRunnersBusy requeued tasks by enqueueing them while they still sat in the running set, then sending EventTypeRequeued from a defer. A periodic queue tick could ClaimAndDequeue the task in that window and start a second dispatch on the same TaskRunner. Release running/active bookkeeping before enqueueing and notify the pool synchronously, matching the reconciler requeue paths. Co-authored-by: Denis Gukov <fiftin@outlook.com>
…finalization Runner client: when applyTerminatedJobs emergency-stops a job and Run() unwinds with an error, the error path overwrote stopped with failed. Add finalizeAfterRun that respects an already-finished status. HA reconciler: when failTaskRunnerLost wins the finalize lock but the DB already has a terminal status, complete finalization instead of bailing so pool/Redis state and autorun children are not leaked. Harden offline requeue with a pre-mutation DB re-check and rollback on persist failure. Co-authored-by: Denis Gukov <fiftin@outlook.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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Automated bug inspection of
developat5df51f03found three critical correctness bugs in the remote-runner task lifecycle that remain unfixed. This PR cherry-picks the validated fixes (e854df0e,7b9ffbea).Recent commits since yesterday (branch override gate, access-key validation, custom-role privilege escalation fix, playbook path validation) were reviewed and look correct — no new critical regressions.
Bug 1: Duplicate dispatch on all-runners-busy requeue
Impact: When every runner is at capacity,
ErrAllRunnersBusyenqueues the task while it still sits in the running set, then sendsEventTypeRequeuedfrom a defer. A concurrent queue tick canClaimAndDequeuethe task in that window and start a second dispatch on the sameTaskRunner— duplicate execution.Root cause: Running/active bookkeeping is released asynchronously via
EventTypeRequeuedinstead of before enqueue.Fix: Call
onTaskStopand notify the pool synchronously before enqueueing, matching reconciler requeue paths.Bug 2: Stopped → failed on runner after
terminated_jobsImpact: When the server returns
terminated_jobs(task reassigned or finished while runner was offline),applyTerminatedJobsemergency-stops the job and setsstopped. Whenjob.Run()unwinds with a kill error, the error path overwritesstoppedwithfailed, reporting the wrong terminal status to the server.Root cause:
job_pool.goerror handler unconditionally setsfailedwithout checking if status is already terminal.Fix: Add
finalizeAfterRunthat respects an already-finished status.Bug 3: HA pool state leak in
failTaskRunnerLostImpact: In HA mode, when the reconciler wins the finalize lock but the DB already has a terminal status (runner reported on another node),
failTaskRunnerLostreturns without completing finalization. Pool/Redis state (running set, claims,End, autorun children) leaks.Root cause: Early return on
IsFinished()without callingfinalizeRemoteTaskLocked. AlsofinalizeRemoteTaskLockedonly checkedEnd != nilunder HA guard.Fix: Complete finalization when DB is already terminal; harden offline requeue with pre-mutation DB re-check and rollback on persist failure.
Validation
go test ./services/tasks/... ./services/runners/...— all passfinalizeAfterRunterminal-status preservation, HA reconciler finalization races