Add rebuild command and change detection for project resources

[mitchdenny]

Description

Adds a rebuild command to .NET project resources and a change detection service that monitors source files and prompts users when a rebuild is needed.

Fixes #14970

Rebuild Command

• Adds a "Rebuild" button to ProjectResource instances in the Aspire dashboard
• Creates a hidden ProjectRebuilderResource (an ExecutableResource subclass) that runs dotnet build on demand
• The rebuild flow: Stop resource → Run dotnet build → Restart resource
• Build output is streamed into the main resource's console logs with a [build] prefix so users can see progress and errors inline
• Build failures leave the resource stopped with an error state showing the exit code
• The rebuilder uses ExplicitStartupAnnotation (only starts on demand) and is hidden from the dashboard

Change Detection Service

• ProjectChangeDetectionService: a BackgroundService that monitors source files for running project resources
• Queries MSBuild for the project's file closure (Compile + ProjectReference items) via dotnet msbuild -getItem JSON output
• Periodically checks file timestamps (default 10s, configurable via ASPIRE_PROJECT_CHANGE_DETECTION_INTERVAL)
• Shows a dashboard notification banner via IInteractionService when changes are detected, with a Rebuild button that triggers the rebuild directly
• Also logs to the resource console when changes are detected
• Includes 2-second debounce and single-notification-per-cycle logic
• Runs automatically in run mode (no opt-in required)

Follow-up items

• --no-build optimization after rebuild to avoid double-build on restart
• Interaction with dotnet watch mode (potentially hide rebuild when watch is active)

Checklist

• Is this feature complete?
  • Yes. Ready to ship.
  • No. Follow-up changes expected.
• Are you including unit tests for the changes and scenario tests if relevant?
  • Yes
  • No
• Did you add public API?
  • Yes
  • No
• Does the change make any security assumptions or guarantees?
  • Yes
  • No
• Does the change require an update in our Aspire docs?
  • Yes
    • Link to aspire.dev issue:
      • New issue
  • No

[github-actions]

🚀 Dogfood this PR with:

⚠️ WARNING: Do not do this without first carefully reviewing the code of this PR to satisfy yourself it is safe.

curl -fsSL https://raw.githubusercontent.com/dotnet/aspire/main/eng/scripts/get-aspire-cli-pr.sh | bash -s -- 15133

Or

• Run remotely in PowerShell:

iex "& { $(irm https://raw.githubusercontent.com/dotnet/aspire/main/eng/scripts/get-aspire-cli-pr.ps1) } 15133"

[Copilot]

Pull request overview

This PR adds a rebuild workflow for .NET ProjectResource instances (via a hidden rebuilder executable + dashboard command) and introduces an opt-in background service that detects source changes and notifies users that a rebuild is needed.

Changes:

• Adds a rebuild lifecycle command for ProjectResource that stops the resource, runs dotnet build, forwards build logs, and restarts on success.
• Introduces ProjectChangeDetectionService + supporting build/closure utilities to periodically detect file changes and notify via resource logs + dashboard notification (opt-in).
• Adds localized command strings and test coverage for the new command and closure/build helper behavior.

Reviewed changes

Copilot reviewed 25 out of 26 changed files in this pull request and generated 7 comments.

Show a summary per file

File	Description
tests/Aspire.Hosting.Tests/ResourceCommandAnnotationTests.cs	Adds rebuild command state/description tests and verifies it’s only on project resources.
tests/Aspire.Hosting.Tests/Build/ProjectFileClosureTests.cs	Adds tests for detecting changed files via captured timestamps.
tests/Aspire.Hosting.Tests/Build/ProjectBuildHelperTests.cs	Adds basic tests around GetProjectFileClosureAsync error/cancellation behavior.
src/Aspire.Hosting/Resources/xlf/CommandStrings.zh-Hant.xlf	Adds Rebuild localized string entries (new/untranslated).
src/Aspire.Hosting/Resources/xlf/CommandStrings.zh-Hans.xlf	Adds Rebuild localized string entries (new/untranslated).
src/Aspire.Hosting/Resources/xlf/CommandStrings.tr.xlf	Adds Rebuild localized string entries (new/untranslated).
src/Aspire.Hosting/Resources/xlf/CommandStrings.ru.xlf	Adds Rebuild localized string entries (new/untranslated).
src/Aspire.Hosting/Resources/xlf/CommandStrings.pt-BR.xlf	Adds Rebuild localized string entries (new/untranslated).
src/Aspire.Hosting/Resources/xlf/CommandStrings.pl.xlf	Adds Rebuild localized string entries (new/untranslated).
src/Aspire.Hosting/Resources/xlf/CommandStrings.ko.xlf	Adds Rebuild localized string entries (new/untranslated).
src/Aspire.Hosting/Resources/xlf/CommandStrings.ja.xlf	Adds Rebuild localized string entries (new/untranslated).
src/Aspire.Hosting/Resources/xlf/CommandStrings.it.xlf	Adds Rebuild localized string entries (new/untranslated).
src/Aspire.Hosting/Resources/xlf/CommandStrings.fr.xlf	Adds Rebuild localized string entries (new/untranslated).
src/Aspire.Hosting/Resources/xlf/CommandStrings.es.xlf	Adds Rebuild localized string entries (new/untranslated).
src/Aspire.Hosting/Resources/xlf/CommandStrings.de.xlf	Adds Rebuild localized string entries (new/untranslated).
src/Aspire.Hosting/Resources/xlf/CommandStrings.cs.xlf	Adds Rebuild localized string entries (new/untranslated).
src/Aspire.Hosting/Resources/CommandStrings.resx	Adds RebuildName and RebuildDescription strings.
src/Aspire.Hosting/Resources/CommandStrings.Designer.cs	Updates the generated resource accessors for rebuild strings.
src/Aspire.Hosting/ProjectResourceBuilderExtensions.cs	Creates a hidden ProjectRebuilderResource for project resources in run mode.
src/Aspire.Hosting/DistributedApplicationBuilder.cs	Registers the new change detection hosted service in run mode.
src/Aspire.Hosting/Build/ProjectFileClosure.cs	Implements timestamp-based closure and change detection helper.
src/Aspire.Hosting/Build/ProjectChangeDetectionService.cs	Adds opt-in background monitoring of running projects and user notifications.
src/Aspire.Hosting/Build/ProjectBuildHelper.cs	Adds MSBuild -getItem JSON parsing + directory-scan fallback to build a closure.
src/Aspire.Hosting/ApplicationModel/ProjectRebuilderResource.cs	Adds the hidden executable resource type used for rebuilds.
src/Aspire.Hosting/ApplicationModel/KnownResourceCommands.cs	Introduces the rebuild command name constant.
src/Aspire.Hosting/ApplicationModel/CommandsConfigurationExtensions.cs	Wires the rebuild command into lifecycle commands and forwards build logs.

Files not reviewed (1)

• src/Aspire.Hosting/Resources/CommandStrings.Designer.cs: Language not supported

[github-actions]

🎬 CLI E2E Test Recordings

The following terminal recordings are available for commit 7d51c9c:

Test	Recording
AddPackageInteractiveWhileAppHostRunningDetached	▶️ View Recording
AddPackageWhileAppHostRunningDetached	▶️ View Recording
AgentCommands_AllHelpOutputs_AreCorrect	▶️ View Recording
AgentInitCommand_DefaultSelection_InstallsSkillOnly	▶️ View Recording
AgentInitCommand_MigratesDeprecatedConfig	▶️ View Recording
AspireAddPackageVersionToDirectoryPackagesProps	▶️ View Recording
AspireUpdateRemovesAppHostPackageVersionFromDirectoryPackagesProps	▶️ View Recording
Banner_DisplayedOnFirstRun	▶️ View Recording
Banner_DisplayedWithExplicitFlag	▶️ View Recording
CertificatesClean_RemovesCertificates	▶️ View Recording
CertificatesTrust_WithNoCert_CreatesAndTrustsCertificate	▶️ View Recording
CertificatesTrust_WithUntrustedCert_TrustsCertificate	▶️ View Recording
CreateAndDeployToDockerCompose	▶️ View Recording
CreateAndDeployToDockerComposeInteractive	▶️ View Recording
CreateAndPublishToKubernetes	▶️ View Recording
CreateAndRunAspireStarterProject	▶️ View Recording
CreateAndRunAspireStarterProjectWithBundle	▶️ View Recording
CreateAndRunJsReactProject	▶️ View Recording
CreateAndRunPythonReactProject	▶️ View Recording
CreateAndRunTypeScriptStarterProject	▶️ View Recording
CreateEmptyAppHostProject	▶️ View Recording
CreateStartAndStopAspireProject	▶️ View Recording
CreateTypeScriptAppHostWithViteApp	▶️ View Recording
DescribeCommandResolvesReplicaNames	▶️ View Recording
DescribeCommandShowsRunningResources	▶️ View Recording
DetachFormatJsonProducesValidJson	▶️ View Recording
DoctorCommand_DetectsDeprecatedAgentConfig	▶️ View Recording
DoctorCommand_WithSslCertDir_ShowsTrusted	▶️ View Recording
DoctorCommand_WithoutSslCertDir_ShowsPartiallyTrusted	▶️ View Recording
LogsCommandShowsResourceLogs	❌ Upload failed
PsCommandListsRunningAppHost	❌ Upload failed
PsFormatJsonOutputsOnlyJsonToStdout	▶️ View Recording
RestoreGeneratesSdkFiles	▶️ View Recording
RunWithMissingAwaitShowsHelpfulError	▶️ View Recording
SecretCrudOnDotNetAppHost	▶️ View Recording
SecretCrudOnTypeScriptAppHost	▶️ View Recording
StagingChannel_ConfigureAndVerifySettings_ThenSwitchChannels	▶️ View Recording
StopAllAppHostsFromAppHostDirectory	▶️ View Recording
StopAllAppHostsFromUnrelatedDirectory	▶️ View Recording
StopNonInteractiveMultipleAppHostsShowsError	▶️ View Recording
StopNonInteractiveSingleAppHost	▶️ View Recording
StopWithNoRunningAppHostExitsSuccessfully	▶️ View Recording
TypeScriptAppHostWithProjectReferenceIntegration	▶️ View Recording

---

_{📹 Recordings uploaded automatically from CI run #23032713812}

[JamesNK]

The status goes to Finished during rebuild. It would be better if there was a build status displayed in the dashboard.

[mitchdenny]

Summary of additional fixes beyond the rebuild command

While implementing and testing the rebuild command with replicated resources (.WithReplicas(2) + .WaitFor()), we discovered two pre-existing bugs in the hosting infrastructure that became visible during rebuild-triggered restarts. These bugs exist on release/13.2 today and affect any scenario where individual replicas are started or stopped independently — rebuild just makes them easy to reproduce.

---

Bug 1: WaitForDependenciesAsync clobbers Running replicas to "Waiting"

Files changed: ResourceNotificationService.cs

Root cause: WaitUntilStateAsync and WaitUntilCompletionAsync both call model-level PublishUpdateAsync(resource, s => s with { State = "Waiting" }) to indicate a resource is waiting for its dependencies. The model-level PublishUpdateAsync overload iterates ALL resolved resource names (all replicas) and applies the state factory to each one. This means starting replica N sets ALL replicas to "Waiting" — including replica N-1 which may already be Running.

Why this is permanent: The DCP status watcher (OnResourceChanged) publishes per-replica state updates when DCP resources change state. Once a replica reaches Running and the watcher has published that event, no further DCP events arrive unless the DCP resource state changes again. When a subsequent replica's WaitForDependenciesAsync clobbers the first replica back to "Waiting", there is no DCP event to correct the stale state. The replica is stuck in "Waiting" forever.

The event chain (traced through the code):

1. orchestrator.StartResourceAsync(replicaName) checks if state is "Waiting" — if not, calls DcpExecutor.StartResourceAsync
2. DcpExecutor.StartResourceAsync publishes OnResourceStartingContext
3. ApplicationOrchestrator.OnResourceStarting sets THIS replica to "Starting" (per-replica update), then publishes BeforeResourceStartedEvent
4. WaitForInBeforeResourceStartedEvent calls WaitForDependenciesAsync → WaitUntilStateAsync
5. THE BUG: WaitUntilStateAsync calls PublishUpdateAsync(resource, s => s with { State = "Waiting" }) — model-level broadcast to ALL replicas

Repro without rebuild: Start an AppHost with a resource that has .WithReplicas(2) and .WaitFor(someResource). Stop one replica. Start it again. The other replica may get clobbered to "Waiting".

Fix: Guard the state factory to only transition replicas that are actually in the startup phase (state is null, Starting, or Waiting). Replicas already in Running or terminal states keep their current state. This is a targeted fix at the two call sites (lines 278 and 342) — no signature changes, no new overloads.

await PublishUpdateAsync(resource, s =>
    s.State?.Text is null
    || s.State?.Text == KnownResourceStates.Starting
    || s.State?.Text == KnownResourceStates.Waiting
        ? s with { State = KnownResourceStates.Waiting }
        : s).ConfigureAwait(false);

---

Bug 2: Health monitor torn down when one replica stops

Files changed: ResourceHealthCheckService.cs

Root cause: ResourceHealthCheckService maintains a dictionary of health monitors keyed by resourceEvent.Resource.Name (the model name, e.g. "api"). However, WatchAsync emits per-replica events (e.g. "api-0", "api-1"). When ANY replica enters a terminal state, the service tears down the single health monitor for the model name — stopping health evaluation for ALL replicas, including those still Running.

Why this causes oscillation:

1. api-0 enters Exited (terminal) → monitor for "api" torn down → health checks stop
2. Next event for api-1 (still Running) → monitor recreated → health checks resume
3. Health check publishes results → triggers state updates → api-0 event arrives (still Exited)
4. Monitor torn down again → cycle repeats

On release/13.2, this bug was masked by Bug 1: the "Waiting" clobber set the stopped replica back to "Waiting" (which is non-terminal), so the monitor was never torn down. Our fix for Bug 1 exposed this bug because the stopped replica now correctly stays in its terminal state.

Fix: Before tearing down the health monitor, check if ALL replicas of the resource are in terminal states. Only tear down when none are still running. Health checks continue through the shared DCP port as long as any replica is alive.

var allReplicasTerminal = resource.GetResolvedResourceNames().All(name =>
    resourceNotificationService.TryGetCurrentState(name, out var evt)
    && KnownResourceStates.TerminalStates.Contains(evt.Snapshot.State?.Text));

if (allReplicasTerminal)
{
    state.StopResourceMonitor();
    // ...
}

---

Why these fixes are in this PR

Both bugs are pre-existing on release/13.2 and affect any replicated resource with health checks or .WaitFor() dependencies. The rebuild command exercises the restart-replica-after-build path, which reliably triggers both bugs. Without these fixes, rebuild with replicas is broken. The fixes are small, targeted, and improve the overall replica experience beyond just rebuild.

[JamesNK]

I forgot to mention, projects in a waiting state should probably be rebuildable.

[JamesNK]

A few issues found.

[JamesNK]

Rebuilding waiting resources fails:

Waiting for resource 'messaging' to enter the 'Running' state.
Executing command 'rebuild'.
[build] Stopping resource for rebuild...
 Error executing command 'rebuild'.
k8s.Autorest.HttpOperationException: Operation returned an invalid status code 'NotFound', response body {"kind":"Status","apiVersion":"v1","metadata":{},"status":"Failure","message":"executables.usvc-dev.developer.microsoft.com \"orderprocessor-wuyntwbf\" not found","reason":"NotFound","details":{"name":"orderprocessor-wuyntwbf","group":"usvc-dev.developer.microsoft.com","kind":"executables"},"code":404}

   at k8s.Kubernetes.SendRequestRaw(String requestContent, HttpRequestMessage httpRequest, CancellationToken cancellationToken)
   at k8s.AbstractKubernetes.ICustomObjectsOperations_PatchClusterCustomObjectWithHttpMessagesAsync[T](Object body, String group, String version, String plural, String name, String dryRun, String fieldManager, String fieldValidation, Nullable`1 force, IReadOnlyDictionary`2 customHeaders, CancellationToken cancellationToken)
   at k8s.AbstractKubernetes.k8s.ICustomObjectsOperations.PatchClusterCustomObjectWithHttpMessagesAsync(Object body, String group, String version, String plural, String name, String dryRun, String fieldManager, String fieldValidation, Nullable`1 force, IReadOnlyDictionary`2 customHeaders, CancellationToken cancellationToken)
   at Aspire.Hosting.Dcp.KubernetesService.<>c__DisplayClass18_0`1.<<PatchAsync>b__0>d.MoveNext() in C:\Development\Source\aspire\src\Aspire.Hosting\Dcp\KubernetesService.cs:line 170
--- End of stack trace from previous location ---
   at Aspire.Hosting.Dcp.KubernetesService.<>c__DisplayClass28_0`1.<<ExecuteWithRetry>b__0>d.MoveNext() in C:\Development\Source\aspire\src\Aspire.Hosting\Dcp\KubernetesService.cs:line 483
--- End of stack trace from previous location ---
   at Polly.ResiliencePipeline.<>c__10`1.<<ExecuteAsync>b__10_0>d.MoveNext()
--- End of stack trace from previous location ---
   at Polly.Outcome`1.GetResultOrRethrow()
   at Polly.ResiliencePipeline.ExecuteAsync[TResult](Func`2 callback, CancellationToken cancellationToken)
   at Aspire.Hosting.Dcp.KubernetesService.ExecuteWithRetry[TResult](DcpApiOperationType operationType, String resourceType, Func`2 operation, Func`2 isRetryable, CancellationToken cancellationToken) in C:\Development\Source\aspire\src\Aspire.Hosting\Dcp\KubernetesService.cs:line 480
   at Aspire.Hosting.Dcp.DcpExecutor.<>c__DisplayClass88_0.<<StopResourceAsync>b__0>d.MoveNext() in C:\Development\Source\aspire\src\Aspire.Hosting\Dcp\DcpExecutor.cs:line 2493
--- End of stack trace from previous location ---
   at Polly.ResiliencePipeline.<>c__9`2.<<ExecuteAsync>b__9_0>d.MoveNext()
--- End of stack trace from previous location ---
   at Polly.Outcome`1.GetResultOrRethrow()
   at Polly.ResiliencePipeline.ExecuteAsync[TResult,TState](Func`3 callback, TState state, CancellationToken cancellationToken)
   at Aspire.Hosting.Dcp.DcpExecutor.StopResourceAsync(IResourceReference resourceReference, CancellationToken cancellationToken) in C:\Development\Source\aspire\src\Aspire.Hosting\Dcp\DcpExecutor.cs:line 2471
   at Aspire.Hosting.Orchestrator.ApplicationOrchestrator.StopResourceAsync(String resourceName, CancellationToken cancellationToken) in C:\Development\Source\aspire\src\Aspire.Hosting\Orchestrator\ApplicationOrchestrator.cs:line 593
   at Aspire.Hosting.ApplicationModel.CommandsConfigurationExtensions.ExecuteRebuildAsync(ExecuteCommandContext context, ProjectResource projectResource) in C:\Development\Source\aspire\src\Aspire.Hosting\ApplicationModel\CommandsConfigurationExtensions.cs:line 206
   at Aspire.Hosting.ApplicationModel.ResourceCommandService.ExecuteCommandCoreAsync(String resourceId, IResource resource, String commandName, CancellationToken cancellationToken) in C:\Development\Source\aspire\src\Aspire.Hosting\ApplicationModel\ResourceCommandService.cs:line 161

Repo tests:

1. Stop docker
2. Run TestShop
3. orderprocessor should be in a waiting state because messaging isn't running
4. Rebuild orderprocessor
5. See error

[mitchdenny]

Addressed all three feedback items in c7c0cb9:

1. Building state stuck on cancellation — Added an outer catch (OperationCanceledException) when (context.CancellationToken.IsCancellationRequested) block that sets the resource state to Exited and returns an error result. This handles cancellation at any point in the rebuild flow (during StartResourceAsync, WatchAsync, or replica restart). Also removed the dead code (if (context.CancellationToken.IsCancellationRequested) check that was unreachable).

2. XLF formatting — Fixed all 13 XLF files: corrected indentation from 8-space to 6-space for the RebuilderResourceNotFound entry, and removed the extra translate="yes" xml:space="preserve" attributes that were artifacts of manual editing.

3. Trailing newlines — Restored trailing newlines on the .resx file and all 13 XLF files.

[mitchdenny]

Re: Should this be Finished?

Good question. The docs (polyglot-apphost-testing.md) define the convention as: Exited is for executables/projects, Finished is for containers. Since this is a project resource, Exited is the correct one by convention.

That said, looking at the codebase, the two states are functionally identical — same dashboard rendering (stop icon, info color), same terminal-state behavior, no branching logic differentiates them. Could be worth considering consolidating them in a future PR, but that would touch containers, projects, dashboard, and docs so probably not in scope here.

Happy to switch to Finished if you prefer though — it would behave identically.

[JamesNK]

Everything is working. However, I see this error in apphost:

basketservice-rebuilder-udhatzzq resource is not found.

[JamesNK]

fail: Aspire.Hosting.Dcp.DcpExecutor[0]
      Error streaming logs for basketservice-rebuilder-udhatzzq.
      k8s.Autorest.HttpOperationException: Operation returned an invalid status code 'NotFound', response body {"kind":"Status","apiVersion":"v1","metadata":{},"status":"Failure","message":"executables.usvc-dev.developer.microsoft.com \"basketservice-rebuilder-udhatzzq\" not found","reason":"NotFound","details":{"name":"basketservice-rebuilder-udhatzzq","group":"usvc-dev.developer.microsoft.com","kind":"executables"},"code":404}

         at k8s.Kubernetes.SendRequestRaw(String requestContent, HttpRequestMessage httpRequest, CancellationToken cancellationToken)
         at Aspire.Hosting.Dcp.DcpKubernetesClient.ReadSubResourceAsStreamAsync(String group, String version, String plural, String name, String subResource, String namespaceParameter, IReadOnlyCollection`1 queryParams, CancellationToken cancellationToken) in C:\Development\Source\aspire\src\Aspire.Hosting\Dcp\DcpKubernetesClient.cs:line 76
         at Aspire.Hosting.Dcp.KubernetesService.<>c__DisplayClass22_0`1.<<GetLogStreamAsync>b__0>d.MoveNext() in C:\Development\Source\aspire\src\Aspire.Hosting\Dcp\KubernetesService.cs:line 345
      --- End of stack trace from previous location ---
         at Aspire.Hosting.Dcp.KubernetesService.<>c__DisplayClass28_0`1.<<ExecuteWithRetry>b__0>d.MoveNext() in C:\Development\Source\aspire\src\Aspire.Hosting\Dcp\KubernetesService.cs:line 483
      --- End of stack trace from previous location ---
         at Polly.ResiliencePipeline.<>c__10`1.<<ExecuteAsync>b__10_0>d.MoveNext()
      --- End of stack trace from previous location ---
         at Polly.Outcome`1.GetResultOrRethrow()
         at Polly.ResiliencePipeline.ExecuteAsync[TResult](Func`2 callback, CancellationToken cancellationToken)
         at Aspire.Hosting.Dcp.KubernetesService.ExecuteWithRetry[TResult](DcpApiOperationType operationType, String resourceType, Func`2 operation, Func`2 isRetryable, CancellationToken cancellationToken) in C:\Development\Source\aspire\src\Aspire.Hosting\Dcp\KubernetesService.cs:line 480
         at Aspire.Hosting.Dcp.ResourceLogSource`1.GetAsyncEnumerator(CancellationToken cancellationToken)+MoveNext() in C:\Development\Source\aspire\src\Aspire.Hosting\Dcp\ResourceLogSource.cs:line 91
         at Aspire.Hosting.Dcp.ResourceLogSource`1.GetAsyncEnumerator(CancellationToken cancellationToken)+System.Threading.Tasks.Sources.IValueTaskSource<System.Boolean>.GetResult()
         at Aspire.Hosting.Dcp.DcpExecutor.<>c__DisplayClass54_1`1.<<StartLogStream>b__1>d.MoveNext() in C:\Development\Source\aspire\src\Aspire.Hosting\Dcp\DcpExecutor.cs:line 661
      --- End of stack trace from previous location ---
         at Aspire.Hosting.Dcp.DcpExecutor.<>c__DisplayClass54_1`1.<<StartLogStream>b__1>d.MoveNext() in C:\Development\Source\aspire\src\Aspire.Hosting\Dcp\DcpExecutor.cs:line 661

[mitchdenny]

Re: error output in apphost logs (basketservice-rebuilder 404 NotFound)

Root cause: The rebuilder is a short-lived DCP executable. When dotnet build completes, the process exits but the DCP resource persists with follow: true log streaming still active. When the resource is later deleted (on the next rebuild via EnsureResourceDeletedAsync, or during app shutdown), the follow stream's HTTP connection gets a 404 NotFound. Previously this hit the generic catch (Exception) in StartLogStream and was logged as an error.

Fix: Added a specific catch for HttpOperationException with NotFound status in DcpExecutor.StartLogStream — downgrades it from LogError to LogDebug since resource deletion during log streaming is a normal lifecycle event for short-lived resources. This is the same pattern already used elsewhere in DcpExecutor (e.g., EnsureResourceDeletedAsync).

Pushed in 7d51c9c.