Main CI has persistent Polyglot, Hosting-4, and VS Code E2E failures

[davidfowl]

Summary

main CI has been persistently red. The last green CI run I found was:

• Run: https://github.com/microsoft/aspire/actions/runs/26776338699
• Time: 2026-06-01 19:15 UTC
• Commit: e32b1add905370e15768ce24450f07b2805a503a

Latest investigated failing run:

• Run: https://github.com/microsoft/aspire/actions/runs/26992734870
• Time: 2026-06-05 03:02 UTC
• Commit: 18a8820ac366169d563865da67117610e0957204

Persistent failure buckets

Polyglot SDK validation

First observed in run https://github.com/microsoft/aspire/actions/runs/26799903389 after PR #17817 (d071d4954cab6f4b135bb4c0f25333665fe01b2d).

Representative failure:

AspireRestore depends on Aspire.Hosting (>= 13.5.0-ci) but Aspire.Hosting 13.5.0-ci was not found.
Unable to find a stable package Aspire.Hosting.Redis with version (>= 13.4.x)
- Versions from https://api.nuget.org/v3/index.json were not considered

This affects Rust, Java, Go, Python, and TypeScript SDK validation jobs.

Hosting-4 DistributedApplicationPipelineTests on Windows

First observed in run https://github.com/microsoft/aspire/actions/runs/26858200614 after PR #17698 (e20049fb4771c5ec838fc2f64461918140f94567). Latest run still shows 33 DistributedApplicationPipelineTests failures.

Representative failure:

Aspire.Hosting.DistributedApplicationException : Docker is not running. Start Docker and try again.
Docker daemon is not running. Exit code: 1.
Step 'check-container-runtime' failed.

VS Code extension zero-to-running E2E

First zero-to-running failure observed in run https://github.com/microsoft/aspire/actions/runs/26855094131 after PR #17851 (b178f82b5cf60b38995f65b8e6f711f238e11859), initially as a teardown-only failure.

The latest run shows a Linux editor-title timeout:

Timed out waiting for editor title containing 'localhost:17150'.
Wait timed out after 180011ms
Open editor titles: "apphost.cs", "ExtensionZeroToRunningApp reso…"

Logs show the dashboard started and the extension attempted to open it in the integrated browser before the assertion timed out.

Notes

The Polyglot failure is the earliest persistent current bucket and appears to be keeping main red continuously since the 13.4 merge. The Hosting-4 and VS Code E2E failures were layered on later and are still visible in the latest investigated CI run.

[joperezr]

🤖 Aspire triage pass — 2026-06-05 06:00 UTC

Area: area-engineering-systems, area-polyglot, area-extension
Kind: test-failure
Repro: yes-large — three persistent CI failure buckets enumerated in the issue body with run links and first-bad commits/PRs.
Affected version(s): main branch CI since 2026-06-01 19:15 UTC (last known green commit e32b1ad); not a shipped-customer-bits issue.
Severity: S2 — main CI is persistently red across Polyglot SDK validation, Hosting-4 DistributedApplicationPipelineTests on Windows, and the VS Code extension zero-to-running E2E. No workaround for contributors trying to validate PRs against green CI.

Urgency: 🟡 vNext — recommend milestone 13.5
Reason: blocks ongoing 13.5 development by keeping main red; not a shipped-product regression (no customer impact in 13.4.0 bits), so it does not qualify for 13.4.x servicing. Should be drained quickly in the 13.5 window.

Suggested owner(s): @joperezr (score 6 — authored the Polyglot-regressing PR #17817 in area-engineering-systems), backup @radical (score 6 — top recent author in area-engineering-systems, owns tests/helix, tests/Shared, tests/Aspire.Templates.Tests per CODEOWNERS)
Evidence:

• @joperezr merged Merge release/13.4 to main after v13.4.0 release #17817 in area-engineering-systems on 2026-06-02; reporter explicitly identifies that commit as the first bad Polyglot run.
• @radical merged Unblock WinGet Manifest job on locked-down 1ES agents; update manifest tags #17913 in area-engineering-systems on 2026-06-04 and is CODEOWNER for the test helix infrastructure that the Polyglot SDK validation jobs run on.
• The Hosting-4 bucket points at PR Display resource commands in VS Code extension tree view #17698 (area-engineering-systems, Docker daemon not running on Windows agents) — a separate human follow-up may be needed for that author. CODEOWNERS for /extension (VS Code E2E bucket) is @adamint, who merged the most recent area-extension work (Show extension compatibility error for empty running AppHost describe #17925 on 2026-06-05).

Confidence: high
Needs human?: yes — three independent failure buckets with distinct likely owners; the triage should be split into per-bucket follow-ups rather than landed on a single owner.

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