Implement Health Checks for Weight API

[willvelida]

Parent Issue

Part of #119 - Implement Custom Health Probes and Dependency Health Checks

Overview

Establish the health check pattern by implementing comprehensive health probes for the Weight API service. This is the foundation issue that will establish the pattern for all other services.

Scope

• src/Biotrackr.Weight.Api/
• infra/apps/weight-api/main.bicep

Tasks

1. Install Health Check Packages

Add to Biotrackr.Weight.Api.csproj:

<PackageReference Include="Microsoft.Extensions.Diagnostics.HealthChecks" Version="9.0.0" />
<PackageReference Include="AspNetCore.HealthChecks.CosmosDb" Version="8.0.1" />
<PackageReference Include="AspNetCore.HealthChecks.AzureKeyVault" Version="8.0.1" />

2. Implement Health Check Endpoints

Update Program.cs:

• Add health checks for self, Cosmos DB (lightweight connectivity check), Key Vault
• Implement custom authenticated Fitbit API health check
• Configure /health/live, /health/ready, /health/startup endpoints

Health Check Configuration:

builder.Services.AddHealthChecks()
    .AddCheck("self", () => HealthCheckResult.Healthy("Service is running"))
    .AddAzureCosmosDB(
        sp => sp.GetRequiredService<CosmosClient>(),
        name: "cosmosdb",
        failureStatus: HealthStatus.Degraded,
        tags: new[] { "ready", "database" })
    .AddAzureKeyVault(
        sp => sp.GetRequiredService<SecretClient>(),
        name: "keyvault",
        failureStatus: HealthStatus.Degraded,
        tags: new[] { "ready", "secrets" })
    .AddCheck<FitbitApiHealthCheck>("fitbit-api", 
        failureStatus: HealthStatus.Degraded,
        tags: new[] { "ready", "external" });

3. Create Custom Fitbit API Health Check

Create Services/FitbitApiHealthCheck.cs:

• Use authenticated lightweight endpoint (e.g., /1/user/-/profile.json)
• 5-second timeout
• Return Degraded (not Unhealthy) if API is unreachable
• Log warnings for failures

4. Update Bicep Configuration

Update infra/apps/weight-api/main.bicep:

probes: [
  {
    type: 'Liveness'
    httpGet: {
      path: '/health/live'
      port: 8080
    }
    initialDelaySeconds: 10
    periodSeconds: 30
    failureThreshold: 3
    timeoutSeconds: 5
  }
  {
    type: 'Readiness'
    httpGet: {
      path: '/health/ready'
      port: 8080
    }
    initialDelaySeconds: 5
    periodSeconds: 10
    failureThreshold: 3
    timeoutSeconds: 5
  }
  {
    type: 'Startup'
    httpGet: {
      path: '/health/startup'
      port: 8080
    }
    initialDelaySeconds: 0
    periodSeconds: 5
    failureThreshold: 30  // 150 seconds total
    timeoutSeconds: 5
  }
]

5. Add Integration Tests

Create health endpoint tests:

• Verify /health/live returns 200 when service is running
• Verify /health/ready returns 200 when all dependencies are healthy
• Verify /health/ready returns 503 when Cosmos DB is unavailable
• Verify /health/startup checks all dependencies

6. Document Pattern

Create documentation for:

• Health check implementation steps
• Probe configuration rationale
• Testing approach
• Lessons learned for replication to other services

Acceptance Criteria

• Health check packages installed
• /health/live, /health/ready, /health/startup endpoints implemented
• Cosmos DB health check validates connectivity (lightweight, minimal RU consumption)
• Key Vault health check validates secret client connectivity
• Fitbit API health check uses authenticated endpoint
• Bicep probe configuration updated with 150s startup timeout
• Health endpoints return JSON response with detailed status
• Integration tests validate health endpoints
• Documentation created for pattern replication
• Deployed to dev environment and validated with cold start scenarios

Dependencies

• None (this is the foundation issue)

References

• ASP.NET Core Health Checks
• Container Apps Health Probes
• Reducing Cold Start Time

Notes

This is the pattern-establishing issue. Lessons learned here will simplify implementation for the remaining 7 services. Focus on creating reusable code and clear documentation.

Design Decisions:

• Cosmos DB health check: Lightweight client connectivity only (minimal RU cost)
• Fitbit health check: Authenticated endpoint (validates OAuth flow)
• Health endpoints: Public (no auth required for Container Apps probes)
• Startup timeout: 150 seconds (generous for cold starts)
• Probe intervals: Liveness 30s, Readiness 10s, Startup 5s