Improve startup error reporting - #20334
Conversation
Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>
|
🚀 Dogfood this PR with:
curl -fsSL https://raw.githubusercontent.com/microsoft/aspire/main/eng/scripts/get-aspire-cli-pr.sh | bash -s -- 20334Or
iex "& { $(irm https://raw.githubusercontent.com/microsoft/aspire/main/eng/scripts/get-aspire-cli-pr.ps1) } 20334" |
github-actions
Bot
added
the
area-app-model
This comment has been minimized.
This comment has been minimized.
There was a problem hiding this comment.
Copilot review overview
🟡 Changes recommended
Additional fields must be preserved for system logs with absent or malformed JSON metadata.
Get a fresh assessment by requesting another Copilot review.
Review effort: Balanced
Findings: 1
What changed in this PR
Improves startup diagnostics by surfacing container-runtime errors and executable working directories.
Changes:
- Includes DCP runtime-check details in warnings and exceptions.
- Adds working-directory context to executable system logs.
- Adds regression tests for both scenarios.
| File | Description |
|---|---|
tests/Aspire.Hosting.Tests/ResourceFailureLoggingTests.cs |
Tests invalid working-directory reporting. |
tests/Aspire.Hosting.Tests/Dcp/DcpLogParserTests.cs |
Tests contextual log-field formatting; missing no-JSON and malformed-JSON coverage. |
tests/Aspire.Hosting.Tests/Dcp/DcpDependencyCheckTests.cs |
Tests detailed runtime diagnostics. |
src/Aspire.Hosting/Dcp/ResourceLogSource.cs |
Supplies executable working-directory context. |
src/Aspire.Hosting/Dcp/DcpLogParser.cs |
Supports additional fields, but drops them when JSON metadata is absent or malformed. |
src/Aspire.Hosting/Dcp/DcpDependencyCheck.cs |
Includes DCP failure details in diagnostics. |
Karol Zadora-Przylecki (karolz-ms)
left a comment
Karol Zadora-Przylecki (karolz-ms)
left a comment
There was a problem hiding this comment.
Found 1 issue (correctness): in IDE mode, the new WorkingDirectory field on executable [sys] lines can report a directory DCP never used. Details are in the inline comment.
The targeted tests (DcpDependencyCheckTests, DcpLogParserTests, ExecutableResourceFailureLoggingTests) pass locally on macOS: 41/41.
Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>
Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>
Tests selector50 / 99 PR test projects · 4 PR jobs, from 7 changed files. Selected PR test projects (50 / 99)
Selected PR jobs (4)
How these were chosen — grouped by what changed
🔧 show 45
🧪 📦 affected project 🧪 🧪 🧪 Job reasons
Selection computed for commit |
PR Testing ReportPR Information
Artifact Version Verification
The PR head was rechecked after all scenarios completed and had not moved. Changes AnalyzedFiles Changed
Change Categories
Test Scenarios ExecutedScenario 1: Targeted PR-head automated testsObjective: Validate dependency-check diagnostics, system-log parsing, resource-log enrichment, and executable failure logging directly from the PR source. Coverage Type: Unit/integration and boundary validation Status: ✅ Passed Command: dotnet test --project tests/Aspire.Hosting.Tests/Aspire.Hosting.Tests.csproj --no-launch-profile -- \
--filter-class '*.DcpDependencyCheckTests' \
--filter-class '*.DcpLogParserTests' \
--filter-class '*.ResourceLogSourceTests' \
--filter-class '*.ExecutableResourceFailureLoggingTests' \
--filter-not-trait 'quarantined=true' \
--filter-not-trait 'outerloop=true'Result: 49 passed, 0 failed, 0 skipped. Boundary coverage included:
Environment note: The first invocation used macOS's logical Evidence:
Scenario 2: Missing container runtime surfaces the actionable DCP reasonObjective: Verify a real AppHost startup failure reports DCP's specific Docker check reason at both the warning and exception surfaces. Coverage Type: Unhappy path Status: ✅ Passed Steps:
Observed outcome:
Evidence:
Scenario 3: Invalid executable working directory is identifiedObjective: Verify a process-start failure caused by a nonexistent working directory reports the resolved working directory alongside the operating-system error. Coverage Type: Unhappy path Status: ✅ Passed Steps:
Observed outcome:
Expected Unhappy-Path Outcome: A safe failed resource state with enough context to distinguish an invalid working directory from an invalid executable path. Evidence:
Scenario 4: Valid executable working directory remains successfulObjective: Verify adding working-directory context does not regress successful executable startup or system-log formatting. Coverage Type: Happy-path regression Status: ✅ Passed Steps:
Observed outcome:
Evidence:
Summary
Overall Result✅ PR VERIFIED No blocking issues were found in the tested behavior. |
Adds troubleshooting guidance for the improved startup error messages introduced in microsoft/aspire#20334: container runtime dependency-check reasons now surface in the warning/exception, and executable process-start failures report the resolved working directory. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
|
Pull request created: #1735
|
|
📝 Documentation has been drafted in microsoft/aspire.dev#1735 targeting Step 5 branch taken: Triggered signals (2): Evidence:
Additional context used (PR comments): The author's own verification comment on the PR confirms the observed behavior with concrete examples:
Docs changes:
(summary truncated) Note This draft PR needs human review before merging. |
…ps (#1780) ## Summary <!-- Describe what this pull request changes and why. --> Reconcile the 13.6 wiki audit and **all 25 open `docs-from-code` proposals targeting `release/13.6`** against the actual release source. Add missing canonical guidance rather than putting all coverage in What's new. This is a new, isolated feature PR into `release/13.6`; it does not update the release rollup #1599, merge or close another proposal, or push directly to a release branch. **Draft with explicit remaining packaging/validation gates:** the six REPL walkthroughs are source-verified, but current publicly available 13.6 packages do not contain the late `WithRepl` exports. Generated API catalogs have deliberately not been fabricated or refreshed from 14.x. See the open checklist below. ### Evidence baseline - Documentation base: `717442f6666948bcf77f3d704dc2dadf7c080ec2`. - Product source of truth: [`microsoft/aspire@e8fd6fbb954f50ccd2e66479538392f65e13e71d`](https://github.com/microsoft/aspire/tree/e8fd6fbb954f50ccd2e66479538392f65e13e71d), current `release/13.6` at audit time. Source was read from that Git object, not the stale source working directory. - [13.6 wiki](https://github.com/microsoft/aspire/wiki/13.6-Change-log) snapshot `8e01a371d4f16a1306e48174d4cf1fdeca714348`, whose cutoff is product PR 20511. Later backports 20541/20546/20548 are included here. - Proposal base branches alone were **not** used as proof of release membership. Direct ancestry and known release backports were checked. Four fallback-targeted proposals are excluded below. - Wiki link corrections: its REPL link #1752 actually covers Sandboxes; the REPL proposal is #1740. Its AOT link #1714 covers PFX certificates, not AOT. ### Complete audit-gap checklist Checked items mean documentation coverage is implemented, not that cloud deployment or every product runtime scenario was executed. - [x] **1. Dotnet API graduation:** correct removal to **13.6**, not 14.0, in What's new, both Dotnet guides, and the diagnostic page; preserve the prerelease package caveat. This applies to core `AddDotnetProject`, `DotnetProjectResource`, and related `WithBuildEnvironment` overloads, not all uses of the diagnostic. Source: microsoft/aspire#20496. - [x] **2. Sandboxes:** remove obsolete API suppressions in the article and deployment guide while preserving Azure service preview/access and prerelease package limitations. Source: microsoft/aspire#20483. - [x] **3. Docked REPL documentation:** all six PostgreSQL/MySQL/MongoDB/SQL Server/Redis/Valkey guides plus the article now cover opt-in `WithRepl`/`withRepl`, run-only availability, actual client privileges, credential handling, and explicit exit versus closing a viewer. Source: microsoft/aspire#20419, backport of microsoft/aspire#20231. Package-backed checks remain open below. - [x] **4. Terminal CLI flag:** update current 13.6 article, `with-terminal`, and all three terminal command references. Preserve `terminals.v1` and experimental hosting API distinctions. Current configuration/schema data had no flag entry to remove; historical 13.5 notes remain historical. Source: microsoft/aspire#20548. - [x] **5. First-party Rust:** rewrite both canonical Rust guides around `Aspire.Hosting.Rust`; document Cargo versus application arguments, typed targets, debugging, generated Dockerfiles, workspace context, ABI constraints, and Toolkit migration. Bacon remains explicitly Toolkit-only. Add exact first-party package mapping. Source: microsoft/aspire#18906 and current Rust README. - [x] **6. Agent setup:** align command reference, skills guide, AI-agent guide, and article on MCP opt-in, `--mcp`, chained/non-interactive behavior, seven-skill catalog, Project v2 migration, and Copilot app detection. Also fix stale default-selection text: all applicable bundle skills are preselected; companion tools remain opt-in. Sources: microsoft/aspire#19893, microsoft/aspire#20405, microsoft/aspire#19820. - [x] **7. Deno AppHost runtime:** document Deno 2+ detection, commands, permissions, native watch/type checking, doctor, and `DENO_CERT`, separately from Deno guest hosting. Source: microsoft/aspire#18627, distinct from microsoft/aspire#18628. - [x] **8. Native AOT / Fluent UI v5:** concise article, dashboard exploration, and standalone guidance; automatic packaged-dashboard selection, no invented performance figures. Source: microsoft/aspire#19565 and release packaging sources. - [x] **9. NuGet:** document bundled in-process operations, credential providers, non-interactive authentication, and realistic troubleshooting. Correct the proposal's `dotnet nuget locals` authentication advice: cache commands do not authenticate a feed. Source: microsoft/aspire#20391. - [x] **10. Multithreaded builds:** article and coordinated-build guide explain `-mt`, SDK detection, distinct project/file-based SDK floors, and fallback. Source: microsoft/aspire#20441. - [x] **11. Radius:** add a real deployment guide with C#/TypeScript setup, recipe-backed connections versus local endpoints, per-resource credential behavior, unauthenticated Redis limitation, secret exposure boundaries, and actionable runtime diagnostics 070–091. Wire navigation and exact package mapping. Source: microsoft/aspire#19555 and release README. - [x] **12. Connection aliases:** replace contradictory no-encoding guidance, retain composed logical-key-first lookup and portable-target behavior, explain collision detection and custom-publisher metadata. Source: microsoft/aspire#19729. - [x] **13. Connector Namespace / Toolbox / provisioning:** add Connector Namespace walkthrough, security/consent/revocation limits and mapping/sidebar; add Foundry Toolbox walkthrough, connection properties, roles, index prerequisites, approval enforcement boundaries, immutable versions, and existing-resource behavior. Extend existing Azure provisioning guide without a duplicate page. Sources: microsoft/aspire#19024, microsoft/aspire#17742, microsoft/aspire#20131. - [x] **14. Remaining high-impact items:** article covers opt-in manifest-aware DNX and new-template CLI bundling (existing SDK guides retained), migration skill and Copilot app detection; canonical inline `CsiVolumeSourceV1`/`VolumeV1.Csi` example, management links, Cosmos vNext telemetry, and AI Inference `GetModelInfoAsync`/`/info` health checks with `DisableHealthChecks`. No Azure OpenAI health-check claim. Sources: microsoft/aspire#19310, microsoft/aspire#19076, microsoft/aspire#19826, microsoft/aspire#20070, microsoft/aspire#15671, microsoft/aspire#15969. - [x] **15. All 25 proposal dispositions:** listed below, including newer dashboard backports and four exclusions. Existing Sandbox inference coverage is retained rather than copied from a stale draft. - [ ] **16. Refresh generated API/catalog/Twoslash data from an official post-backport 13.6 build.** Existing `26473.12`/`a11eca96` data remains untouched. The newest public `dotnet9` feed package checked, `13.6.0-preview.1.26474.10` at `43496a2a306c81c862c947b11b4f4e5494b6fe08`, still has no Redis `WithRepl` in its actual package XML. Do not use 14.x, hand-edit declarations, or attribute source changes to older binaries. - [ ] **Validate the six REPL examples against that actual post-backport SDK and running clients.** Their new TypeScript fences are plain TypeScript, not annotated with unsupported Twoslash data. No existing diagnostics are allowlisted or suppressed; no generated API exports are fabricated. Enable Twoslash when the genuine catalog catches up. ### All 25 open proposal dispositions and provenance Text is selectively adapted from these proposals, not merged wholesale. #1778 and #1748 are authored by @sebastienros; the other proposals are authored by the Aspire repo bot. The table credits the associated product-change authors where supplied by the proposals. Existing PRs remain open and unchanged. | Docs PR | Release source / credited product author | Disposition | | --- | --- | --- | | #1778 | microsoft/aspire#19729 — @sebastienros | **Adopted:** canonical connection-string alias correction, including logical-first resolution and migration. | | #1771 | microsoft/aspire#20481 — @sebastienros | **Excluded:** flat polyglot feature keys are not in the audited release tip; no verified backport. Preserve release key names. | | #1770 | microsoft/aspire#20525 → microsoft/aspire#20548 — @mitchdenny | **Corrected/adopted:** command guides plus the still-current 13.6 article, which the proposal incorrectly treats as historical. | | #1769 | microsoft/aspire#20416 — @JamesNK | **Excluded:** brand hover change has no verified 13.6 membership/backport. | | #1768 | microsoft/aspire#20523 → microsoft/aspire#20546 — @JamesNK | **Adopted:** run pin/unpin preserves selector and current selection. | | #1766 | microsoft/aspire#20537 → microsoft/aspire#20541 — @mitchdenny | **Adopted:** terminal dock empty state. | | #1761 | microsoft/aspire#20490 → microsoft/aspire#20496 — @eerhardt | **Corrected:** graduation is 13.6, package remains prerelease, Blazor-specific exception retained. | | #1760 | microsoft/aspire#20436 — @eerhardt | **Excluded:** CLI net11/tools-any retarget is not in the audited release; no fallback-base inference. | | #1748 | microsoft/aspire#20131 — @sebastienros | **Adopted:** extend existing provisioning guide with service-specific models/lookups and projection limits. | | #1744 | microsoft/aspire#20337 → microsoft/aspire#20441 — @karolz-ms | **Adopted:** precise SDK-conditional multithreaded build coverage. | | #1740 | microsoft/aspire#20231 → microsoft/aspire#20419 — @mitchdenny | **Adapted:** all six guides; TypeScript-first tabs, source-verified lifecycle/security. Actual post-backport SDK/runtime gate is open above. | | #1738 | microsoft/aspire#20158 → microsoft/aspire#20405 — @karolz-ms | **Partly already covered / completed:** existing seven-skill catalog retained; add project migration guidance and correct command catalog/defaults. Do not misclassify the bundled skill as a companion tool. | | #1735 | microsoft/aspire#20334 — @karolz-ms | **Excluded:** enhanced startup errors are not in the audited release; no verified backport. | | #1731 | microsoft/aspire#19847 → microsoft/aspire#20391 — @eerhardt | **Corrected/adopted:** in-process NuGet and real authenticated-restore troubleshooting, not cache-command authentication. | | #1719 | microsoft/aspire#20299 → microsoft/aspire#20407 — @JamesNK | **Corrected/adopted:** cookie naming/scoping; identical names can collide but do not guarantee cross-dashboard cookie decryptability or shared sign-in. | | #1664 | microsoft/aspire#20011 — @maddymontaquila | **Adopted:** concise Azure environment icon release note. | | #1628 | microsoft/aspire#17742 — @davidfowl | **Adapted/expanded:** canonical Toolbox examples, consumer contract, role/index prerequisites, approval/security and concurrency limits. | | #1623 | microsoft/aspire#19810 — @mitchdenny | **Already covered:** current Sandbox guide/article already describe compute inference, explicit selection and external endpoints. Preserve that guidance while removing obsolete suppressions. | | #1620 | microsoft/aspire#19243 — @sebastienros | **Adapted:** AKS credential-before-Helm cleanup and destructive-operation warning; omit misleading ambient-context workaround. | | #1614 | microsoft/aspire#19870 — @sebastienros | **Adopted:** typed callback handle behavior in extension authoring and article. | | #1574 | microsoft/aspire#19430 — @mitchdenny | **Adapted:** canonical hostname inheritance, explicit-host precedence, catch-all default backend. | | #1570 | microsoft/aspire#19590 — @karolz-ms | **Adopted:** Dev Tunnel URL regression troubleshooting. | | #1565 | microsoft/aspire#19429 — @mitchdenny | **Corrected/adopted:** Helm embedded parameters with real `refExpr` and `addParameter(name, { value })`, not stringifying a handle or using an invalid actual-SDK overload. | | #1564 | microsoft/aspire#19026 — @karolz-ms | **Corrected/adopted:** C#/TypeScript Dotnet gateway walkthrough. Retain both experimental diagnostics; remove obsolete run-only restriction after microsoft/aspire#19997 publishing support. Avoid imported ambiguous API reference. | | #1499 | microsoft/aspire#19248 — @IEvangelist | **Adopted:** describe exact secret-value redaction and embedded-secret limit; release article already covered the fix. | ### Important source-verified corrections to proposals / earlier audit assumptions - [`BlazorGatewayExtensions.cs`](https://github.com/microsoft/aspire/blob/e8fd6fbb954f50ccd2e66479538392f65e13e71d/src/Aspire.Hosting.Blazor/BlazorGatewayExtensions.cs): `AddDotnetProjectBlazorGateway` and the Dotnet `WithBlazorClientApp` overload still carry `ASPIREDOTNETPROJECT001`; the class carries `ASPIREBLAZOR001`. They share `WithBlazorClientAppCore`/`WithBlazorApp` and the publish-companion path. Thus neither blanket diagnostic retirement nor the proposal's old run-only claim is correct. - [`SkillDefinition.cs`](https://github.com/microsoft/aspire/blob/e8fd6fbb954f50ccd2e66479538392f65e13e71d/src/Aspire.Cli/Agents/SkillDefinition.cs) sets bundled skills' `IsDefault=true`; [`AgentInitCommand.cs`](https://github.com/microsoft/aspire/blob/e8fd6fbb954f50ccd2e66479538392f65e13e71d/src/Aspire.Cli/Commands/AgentInitCommand.cs) selects the applicable catalog defaults for both flows. MCP has its own standalone-only binding. - [`TypeScriptAppHostToolchainResolver.cs`](https://github.com/microsoft/aspire/blob/e8fd6fbb954f50ccd2e66479538392f65e13e71d/src/Aspire.Cli/Projects/TypeScriptAppHostToolchainResolver.cs) is the source for Deno flags and certificate variable; guest Deno hosting is separate. - [`Radius README`](https://github.com/microsoft/aspire/blob/e8fd6fbb954f50ccd2e66479538392f65e13e71d/src/Aspire.Hosting.Radius/README.md) supplies the resource-specific credential rules and publish diagnostics, not assumptions about local endpoints. ## Third-party links and affiliations <!-- List third-party links and disclose material affiliations. --> Links point to official Microsoft Learn, VS Code Marketplace debugger extensions, Rust/Cargo/Bacon documentation, Radius documentation, and source repositories. No sponsorship, commercial endorsement, or affiliation claim is introduced. Maintainers should supply any personal affiliation disclosure required by policy; automation has not inferred one. ## Validation <!-- List the checks you ran or explain why validation isn't needed. --> - **97 passing focused unit checks** across API-reference authoring/rendering, Twoslash blocks, file-tree formatting, CLI configuration schema, SEO lengths, and resource catalog. - **82 passing structured-data checks**, including exact integration mapping uniqueness and page resolution. - **11 C# samples compile**, zero warnings/errors, using genuine `13.6.0-preview.1.26473.12` packages. Scope: Rust, Connector Namespace, Radius, Toolbox, inline CSI, Helm, Blazor gateway, and provisioning. `Projects.Api/Worker/Client` use compile-only `IProjectMetadata` stand-ins; no claim of running those apps or provisioning cloud resources. - **10 TypeScript samples pass `tsc`** under `strict`, `NodeNext`, and `ES2022` against three **unmodified actual SDK files**, not just the site's declaration bundle. The fixture uses the exact `e8fd6fbb` release `AtsCapabilityScanner` and genuine `26473.12` TypeSystem/code-generator/integration binaries, whose informational source is `a11eca96`. This is an isolated local generation fixture, **not** a claim that official CLI generation or a new packaged release was tested. An attempted restore with the older handed-off local CLI could not discover an AppHost server; the bounded direct generator fixture was used instead. - The SDK scan is **not globally warning-free**: it reports a Radius `withContainerImage` collision on `CSharpAppResource` and an App Configuration `createRoleAssignment` overload collision. None of the compiled examples calls those colliding methods; the warnings are retained in evidence, not suppressed, and no generated declarations were edited. - Browser: Connector Namespace, Radius, both Rust pages, Foundry hosting, and What's new return **HTTP 200**, correct headings, and no rendered Twoslash errors. New guide/article page-local anchors and the cross-page Blazor anchor resolve. Connector/Radius mobile layouts have no horizontal overflow; Connector language-tab interaction works. Standalone Astro preview emits expected `/api/live` 404s because StaticHost is not running. - `git diff --check` passes. No production `pnpm build`, cloud deployment, REPL runtime session, full product suite, or blanket validation of every pre-existing example was performed. - Generated C#/TypeScript API data, declaration bundles, integration catalogs, image catalogs, and contributor data are unchanged. Only the authored package-to-guide mapping is updated. **Before merging:** complete the two packaging/REPL checkboxes above, inspect CI, and obtain human review. This PR intentionally does not close or merge the source documentation proposals. --------- Co-authored-by: David Pine <7679720+IEvangelist@users.noreply.github.com> Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>
Description
Startup failures currently hide useful context in two common scenarios:
This change surfaces the DCP dependency-check reason in the warning and thrown exception when one is available. It also allows system-log callers to append contextual fields and uses that support to include a non-empty executable working directory in process logs.
No DCP changes or new dependencies are required.
User-facing behavior
Container runtime startup failures now include the specific DCP-provided reason at the default warning and exception surfaces. Executable system logs now include
WorkingDirectory = <resolved path>when the executable has a working directory.Validation:
All 41 targeted tests passed.
Fixes #19593
Fixes #18652
Checklist
<remarks />and<code />elements on your triple slash comments?