[glen-84]

Summary

• SubscriptionEventTimeoutTests.Timer_Should_Be_Reset_Between_Events guarded its whole run with a 10-second CTS and hit it in CI inside a Task.Delay(600), after every event had already completed without errors. The outer guard is now 30 seconds, matching the other tests in the file.
• The lost time is thread-pool queueing, not the per-event timer. Every await in the test resumes on the thread pool (in-memory subscription channel hops and Task.Delay), and when Issue9500Tests runs concurrently its 100k-node DataLoader query fills the pool for the 13 to 18 seconds it takes under coverage. A probe measured a 600 ms delay resuming after 3 seconds with 4335 pending work items. The per-event 1-second budget is armed only after an event is dequeued, so the wider outer guard does not weaken what the test asserts.

Test plan

• Ran SubscriptionEventTimeoutTests alone and alongside Issue9500Tests and StarWarsCodeFirstTests on 4 pinned cores with --coverage: 1.8 s alone, 4.6 to 5.4 s under contention, all green.
• In the failing CI run the test took 10.08 s while overlapping Issue9500Tests (18 s); the re-run had no overlap and the test took 2.18 s.

[Copilot]

🟢 Approval recommended

The change is a minimal, test-only timeout adjustment that improves reliability without altering the asserted per-event timeout behavior.

Pull request overview

This PR adjusts an overly tight outer test deadline in SubscriptionEventTimeoutTests.Timer_Should_Be_Reset_Between_Events to reduce CI flakiness caused by thread-pool scheduling delays under concurrent test load.

Changes:

• Increased the test-wide CancellationTokenSource timeout from 10 seconds to 30 seconds for Timer_Should_Be_Reset_Between_Events.

File summaries

File	Description
src/HotChocolate/Core/test/Execution.Tests/Pipeline/SubscriptionEventTimeoutTests.cs	Widens the test’s outer cancellation deadline to better tolerate thread-pool contention while keeping the per-event timeout assertion intact.

Review details

• Files reviewed: 1/1 changed files
• Comments generated: 0
• Review effort level: Lite

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.