[fdaniel02]

The Kubernetes probes only check that port 8080 is open, and Uptime Kuma (backlog item 32) needs an endpoint that
says whether the app works.

Changes

• AddHealthChecks().AddDbContextCheck<AppDbContext>() (new package
  Microsoft.Extensions.Diagnostics.HealthChecks.EntityFrameworkCore 10.0.13, lock files updated).
• /healthz/live: no checks, only "the process answers". For liveness: a database outage must not restart the pod.
• /healthz/ready: with the database check (EF's "can connect"). For Uptime Kuma and the planned Docker smoke test.
• Plain text Healthy / Unhealthy with 200 / 503, no details, since the app is reachable from outside.
• Telemetry.IsTraced also leaves out /healthz, so probes don't become traces (the check's SQL span is dropped
  with its request).

The probes in cluster-gitops switch to these endpoints in a separate PR once this image is deployed. Kubernetes
readiness will use /healthz/live, not /healthz/ready: with one replica, taking the pod out of the Service during
a database outage would only replace the app's own responses with Traefik's "no available server".

Testing

• StartupTests (whole app, SQL Server): both endpoints return 200 Healthy. With the database dropped after
  startup (the app can't start without it), live stays 200 and ready returns 503 Unhealthy. The check doesn't
  hang on EF's retries: the test takes about 0.5 s.
• TelemetryTests: /healthz/* isn't traced.
• Checked that the test fails when liveness includes the database check.

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
100.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud