>_
.issue.db
/highway
Dashboard
Issues
Memory
Lessons
Audit Log
New Issue
Edit Issue #809
Update issue details
Title *
Description
Evidence: in the full harness, a run stayed 'sleeping' 301s after its approval was rejected by the sweeper at ~12s (emit_event fired). Patched with a wait-force (timeout_at=now()) but NOT root-caused. jumper.emit_event DOES durably re-queue (UPDATE r_<queue> SET state='pending' WHERE run_id IN affected AND state='sleeping'), so it's not a pure NOTIFY-miss. Hypotheses: (H1) tenant mismatch in emit_event scoping (w.tenant_id=$4 or $4='default'); (H2) lost-wakeup race - emit deletes wait + re-queues only if state='sleeping'; if the run re-parks (replay) around the emit, or wait_for_event doesn't peek an already-emitted event before parking, the wake is lost; NULL-timeout waits then have NO safety net (SleepingTaskWakeupService only sweeps timeout_at IS NOT NULL). Investigating emit_event tenant default, wait_for_event park/peek logic, and the wait-register SQL. Signal-wake + call_agent joins share this path.
Priority
Low
Medium
High
Critical
Status
Open
In Progress
Closed
Won't Do
Due Date (YYYY-MM-DD)
Tags (comma separated)
Related Issues (IDs)
Enter IDs of issues related to this one. They will be linked as 'related'.
Update Issue
Cancel