All 15 Health Checks questions
More in Health Checks
How do you customize the JSON response format of the health check endpoint, since the default output is just plain text status? How do you add pre-built health checks for common external dependencies like SQL Server, Redis, or a URL endpoint, using the community AspNetCore.Diagnostics.HealthChecks packages? How do health checks interact with dependency injection scoping, particularly for checks that need a scoped service like a DbContext? Why should the detailed /health endpoint (with full dependency status breakdown) generally not be exposed publicly on the internet, and how do you restrict access to it? How would you exclude health check endpoints from other cross-cutting middleware like authentication, logging, or rate limiting that shouldn't apply to frequent automated health polling? How do you differentiate liveness and readiness health checks using tags, and why does Kubernetes treat these two probe types differently? How would you implement a health check that reports Degraded (rather than a binary Healthy/Unhealthy) based on a performance threshold, like elevated database query latency? How would you integrate health check results with the AspNetCore.HealthChecks.UI package to get a persistent, historical dashboard view of health status over time? How would you set a per-check timeout to prevent one slow or hanging dependency check from blocking the entire /health endpoint response indefinitely? How would you implement a startup-specific health check distinct from ongoing liveness/readiness, useful for slow-starting applications with lengthy initialization (like cache warming or large data loads)?
Advertisement
ASP.NET Core › Health Checks

How would you write an integration test verifying that your health check endpoint correctly reports Unhealthy when a dependency is down?

Intermediate

Answer

Using WebApplicationFactory, override the DI registration for the dependency being tested (replacing a real database connection with a fake that simulates a failure) via ConfigureTestServices, then issue a request to /health via the test client and assert both the HTTP status code (503 for Unhealthy by default) and, if using a custom JSON response writer, the specific failing check's name in the response body.
public class HealthCheckTests : IClassFixture<WebApplicationFactory<Program>> {
    [Fact]
    public async Task Health_ReturnsUnhealthy_WhenDatabaseDown() {
        var factory = _factory.WithWebHostBuilder(builder => {
            builder.ConfigureTestServices(services => {
                services.AddHealthChecks().AddCheck("database", () => HealthCheckResult.Unhealthy());
            });
        });
        var client = factory.CreateClient();
        var response = await client.GetAsync("/health");
        Assert.Equal(HttpStatusCode.ServiceUnavailable, response.StatusCode);
    }
}
Real-world example A test suite verifies the /health endpoint correctly returns a 503 status when a simulated database check fails, giving confidence the health check wiring itself is correct before relying on it in a production Kubernetes deployment.

Common follow-ups: How would you test the interaction between multiple health checks with mixed Healthy/Degraded/Unhealthy statuses?;Should health check logic itself have unit tests separate from this integration-level test?

Testing ASP.NET Core Applications;Error Handling
All 15 Health Checks questions