Quarantine flaky ExpectedNumberOfUrlsForReplicatedResource test

[sebastienros]

Summary

Aspire.Hosting.Tests.WithUrlsTests.ExpectedNumberOfUrlsForReplicatedResource is currently flaky in Windows Hosting-5 CI and should be quarantined until the underlying DCP teardown issue is understood and fixed.

Evidence

Two separate PR runs failed in the same job and same test:

• PR Add Docker Compose polyglot helpers #17128: https://github.com/microsoft/aspire/actions/runs/25929076765/job/76233717879?pr=17128
• PR Add TypeScript JavaScript hosting coverage #17057: https://github.com/microsoft/aspire/actions/runs/25893235489/job/76234665705?pr=17057

Both jobs are Tests / Hosting-5 / Hosting-5 (windows-latest) and both failed in Aspire.Hosting.Tests.WithUrlsTests.ExpectedNumberOfUrlsForReplicatedResource.

In both cases, the test timed out during await app.StopAsync().DefaultTimeout(), not during the URL-count assertion. The logs show DCP teardown/shutdown symptoms, including API server shutdown, failed watches/connection refused, process-monitor PID errors, and DCP logs still locked while diagnostics were collected.

Recommendation

Temporarily quarantine this test on Windows or in the affected CI configuration so unrelated PRs are not blocked while the DCP teardown race is investigated.

A separate DCP investigation session has been started to identify, reproduce, and fix the underlying shutdown flake.

[joperezr]

🤖 Aspire triage pass — 2026-05-15 19:56 UTC

Area: area-orchestrator
Kind: flaky-test
Repro: yes-large — two cited CI runs on Tests / Hosting-5 / Hosting-5 (windows-latest) (see linked PR runs in the issue body)
Affected version(s): main / CI (Aspire.Hosting.Tests.WithUrlsTests.ExpectedNumberOfUrlsForReplicatedResource)
Severity: S3 — flaky test blocking unrelated PRs intermittently; quarantine is the documented workaround

Urgency: ⚪ backlog — no milestone
Reason: Test-quarantine maintenance action, not a shipped-bits regression. Should be actioned soon to unblock CI, but does not target 13.3.x servicing.

Suggested owner(s): @karolz-ms (score 8), backup @mitchdenny (score 4)
Evidence:

• @karolz-ms is the most recently active author in area-orchestrator (merged Add session message notification type to IDE protocol #11746 on 2025-10-01, plus Retry both deletion and verification when restarting a resource #8238, Display more info when local orchestrator check fails with unexpected error #4789, IDE protocol spec: add DCP instance ID and clarify notification connection handling #3710, Add payload protection capability to IDE protocol #2991, Update IDE execution spec for Aspire v1 release #2631, Use streaming to get logs for Executables and Containers #2435, Update IDE run session request doc #2020 historically)
• CODEOWNERS lists @JamesNK @karolz-ms @danegsta @mitchdenny @davidfowl for /src/Aspire.Hosting (where DCP teardown lives)
• CODEOWNERS lists @mitchdenny for /tests/Aspire.Hosting.Tests (where the flaky test file lives)

Confidence: high
Needs human?: yes — a maintainer should run the quarantine tool (see tools/QuarantineTools / /quarantine-test command) and link this issue from the [QuarantinedTest] attribute. Bot does not perform code changes.

Dry-run mode — no labels or milestone applied. Reply /triage apply to promote, or edit and reapply.