Scope orphaned-event cleanup to recent ingest window to eliminate 8-hour ES CPU spikes

dogfood-output/pr-2279-cluster-cpu-spikes.md

@@ -0,0 +1,54 @@
+# PR #2279 local verification
+
+Date: 2026-08-05
+Branch: `copilot/investigate-cluster-cpu-spikes`
+Scope: recent-ingest orphan cleanup and cleanup-job health state
+
+## Source and review
+
+- PR was confirmed `OPEN`, non-draft, and `MERGEABLE` at head `5a53ae9c5d08405269dc55197070f79fcd9d089f` before the merge-forward.
+- `origin/main` was 52 commits ahead of the PR branch, so it was merged with a normal merge commit and pushed as `f24d4f5f4`.
+- `git diff origin/main...HEAD` contains only:
+  - `src/Exceptionless.Core/Jobs/CleanupOrphanedDataJob.cs`
+  - `tests/Exceptionless.Tests/Jobs/CleanupOrphanedDataJobTests.cs`
+- `git diff --check origin/main...HEAD`: passed.
+- Surgical review found no additional high-confidence defect requiring a source change; no unrelated files were modified.
+
+## Build and tests
+
+Passed:
+
+```text
+dotnet restore tests/Exceptionless.Tests/Exceptionless.Tests.csproj
+dotnet build tests/Exceptionless.Tests/Exceptionless.Tests.csproj --no-restore --configuration Release -m:1
+Build succeeded. 0 warnings, 0 errors.
+```
+
+The focused suite compiled and launched, but all 27 tests were blocked during the shared `AppWebHostFactory` fixture startup:
+
+```text
+dotnet test --no-restore --no-build --configuration Release --max-parallel-test-modules 1 -- --filter-class Exceptionless.Tests.Jobs.CleanupOrphanedDataJobTests
+```
+
+Result: 27 failed during fixture initialization, 0 test bodies executed. Aspire reported:
+`Container runtime 'docker' was found but appears to be unhealthy.`
+
+## Aspire and dogfood
+
+- Initial `aspire run` hit `MSB4166` (parallel MSBuild child exited prematurely).
+- A serial AppHost build passed:
+
+  ```text
+  dotnet build src/Exceptionless.AppHost/Exceptionless.AppHost.csproj --no-restore -m:1
+  Build succeeded. 0 warnings, 0 errors.
+  ```
+
+- `aspire run --no-build --non-interactive` started a local dashboard, but endpoint discovery showed the backend API waiting for infrastructure.
+- Aspire resource state: Elasticsearch, Redis, Mail, and Storage were `RuntimeUnhealthy`; API and Jobs were `Waiting` on those resources.
+- API health and browser dogfood were not claimed because the API never became reachable. No production URL or data was used.
+
+## Residual gates
+
+- Local focused test execution and changed-surface runtime dogfood remain blocked by the unhealthy Docker runtime.
+- The earlier green PR CI run `30577192830` passed API, client, E2E, Docker, and version jobs at the pre-merge head.
+- Fresh CI run `31013563388` was triggered for the pushed head `19b09a3b6`: `version` and `test-client` passed; `test-api`, `test-e2e`, and `docker-build` were still pending at the final bounded poll. The CLA check was also pending with the contributor agreement unsigned.