Context
In crates/comenqd/src/worker.rs at lines 305-352, the new notify_one tests use tight timeouts:
Duration::from_millis(50) in notify_one_buffers_permit_when_no_waitersDuration::from_millis(100) in notify_one_wakes_exactly_one_waiterDuration::from_millis(10) sleep for waiter registration
Issue
These tight timeouts can reintroduce test flakiness under load or on slower CI runners, particularly when multiple tests run concurrently. Scheduler delays or system load could cause legitimate test failures.
Recommendation
Consider using the existing TimeoutConfig helper (from crates/comenqd/tests/util.rs) or slightly larger timeout bounds to maintain test intent whilst reducing scheduling sensitivity.
References
Context
In
crates/comenqd/src/worker.rsat lines 305-352, the newnotify_onetests use tight timeouts:Duration::from_millis(50)innotify_one_buffers_permit_when_no_waitersDuration::from_millis(100)innotify_one_wakes_exactly_one_waiterDuration::from_millis(10)sleep for waiter registrationIssue
These tight timeouts can reintroduce test flakiness under load or on slower CI runners, particularly when multiple tests run concurrently. Scheduler delays or system load could cause legitimate test failures.
Recommendation
Consider using the existing
TimeoutConfighelper (fromcrates/comenqd/tests/util.rs) or slightly larger timeout bounds to maintain test intent whilst reducing scheduling sensitivity.References