[xavierjohn]

Summary

• Add a one-command, always-on Project Tracker demo with the API, continuous traffic, recurring incidents, Collector, Prometheus, and Grafana.
• Show per-project availability and exact within-250-ms latency compliance, with healthy background traffic and meaningful no-data behavior.
• Use the library's Unknown sentinel for rejected project IDs, so unknown-customer diagnostics reach the dashboard without exposing arbitrary rejected values as metric dimensions. Document how to exercise the panel and why it is normally quiet.
• Keep continuous traffic running through transient request failures and timeouts. Log only the first failure of each kind per phase while counting every failure in phase summaries. Rotate all five containers' JSON logs at 10 MB with three retained files.
• Set explicit local Prometheus retention to 24 hours or 256 MB, with documented head/WAL and compaction caveats. Keep the observability services restarting with the continuous demo and document how to stop them.
• Make finite runs stop sending promptly on failure, cancel/drain outstanding work, and give requested shutdown precedence. Report unexpected and secondary cleanup failures without masking the initiating exception or producing an unhandled stack trace. Suppress the first stop signal before invoking cancellation callbacks, report callback failures, and allow another Ctrl+C or supported SIGTERM to terminate even if draining is stuck.
• Cap traffic at 32 in-flight requests, retire completed tasks, and skip missed send slots instead of creating catch-up bursts. Reuse the pacing deadline instead of creating a linked cancellation source and timer on every wait. Require digit-only, invariant CLI values within 1-1000 requests/second and 1-3600 seconds/phase, with friendly argument errors. Do not add retries or circuit-breaking policies that alter the intended incident traffic.
• Start traffic only after the API's HTTP healthcheck succeeds. Replace host-address autodiscovery with explicit --health-check <URL>; incorrect probe arguments return usage rather than starting another server. Share ${SLI_DEMO_PORT:-8080} across API binding, published-port target, probe, and traffic URL. Probe every five seconds during startup and every 30 seconds afterward using the existing .NET runtime.
• Cache restores separately and isolate each Docker image's project-graph inputs. Exclude worktrees, workflows, and unrelated documentation from the build context while retaining the two required package guides.
• Define sync modes in one status/delay/error table. Cover actual delay palettes, response contracts, both sides of the configured 250-ms threshold, bounded diagnostics, failure precedence, and race-free asynchronous scheduling with focused regressions. Avoid fragile assumptions that CI HTTP requests must finish below 250 ms.
• Align package repository links and the SBOM namespace, and remove redundant dependencies/version pins without changing resolved logging dependencies. Explain the test-only TimeProvider package's independent release cadence rather than trying to align it to an unpublished runtime-servicing version.
• Refresh the landing page with a checked-in dashboard screenshot and complete onboarding. Review and align all 10 product documents, including API references and the metric-contract design record.
• Retain focused Console, Minimal API, and API-versioned MVC examples with stable project dimensions; remove redundant weather/greeting demos, old Aspire launchers, and obsolete screenshots. Preserve custom-meter ownership guidance in the Console example.

Run the demo

From the repository root, with Docker Engine 25 or newer running Linux containers and Compose v2.20.2 or newer:

docker compose -f .\sample\Observability\Grafana\docker-compose.yml --profile demo up --build -d

Open the SLI dashboard and compare project-red with the healthy projects as incidents repeat.

The library's public APIs, metric names, tags, and millisecond units are unchanged. This is a loopback-only learning environment, not a production deployment.

[github-actions]

Test Results

467 tests  +122   465 ✅ +122   33s ⏱️ +13s
  6 suites +  3     2 💤 ±  0 
  6 files   +  3     0 ❌ ±  0 

Results for commit 3fd160f. ± Comparison against base commit e6ba362.

♻️ This comment has been updated with latest results.

[codecov-commenter]

Codecov Report

❌ Patch coverage is 86.79775% with 47 lines in your changes missing coverage. Please review.
✅ Project coverage is 92.71%. Comparing base (e6ba362) to head (3fd160f).

Files with missing lines	Patch %	Lines
sample/GenerateSli/Program.cs	83.75%	34 Missing and 5 partials ⚠️
sample/SliProductionDemo/Program.cs	85.71%	2 Missing and 2 partials ⚠️
sample/MinApi/Program.cs	92.30%	1 Missing and 1 partial ⚠️
...liProductionDemo/Controllers/ProjectsController.cs	96.87%	0 Missing and 1 partial ⚠️
.../WebApiVersioned/Controllers/ProjectsController.cs	0.00%	0 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #80      +/-   ##
==========================================
- Coverage   95.66%   92.71%   -2.96%     
==========================================
  Files          26       33       +7     
  Lines         600     1002     +402     
  Branches       90      173      +83     
==========================================
+ Hits          574      929     +355     
- Misses         13       50      +37     
- Partials       13       23      +10     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.