Service Health

Every MATIH service exposes health endpoints for Kubernetes lifecycle management and monitoring. Python services use FastAPI health routes, and Java services use Spring Boot Actuator. Health responses include checks for the service itself and its critical dependencies.

Health Endpoints

Python Services (FastAPI)

Java Services (Spring Boot)

Kubernetes Probes

Liveness Probe

Determines if the container should be restarted:

livenessProbe:
  httpGet:
    path: /health
    port: http
  initialDelaySeconds: 30
  periodSeconds: 10
  failureThreshold: 3
  timeoutSeconds: 5

Readiness Probe

Determines if the container should receive traffic:

readinessProbe:
  httpGet:
    path: /health/ready
    port: http
  initialDelaySeconds: 10
  periodSeconds: 5
  failureThreshold: 3
  timeoutSeconds: 5

Startup Probe

For services with slow initialization:

startupProbe:
  httpGet:
    path: /health
    port: http
  initialDelaySeconds: 10
  periodSeconds: 5
  failureThreshold: 30
  timeoutSeconds: 5

Dependency Health Checks

The readiness endpoint checks critical dependencies:

Dependency Check Response

{
  "status": "ok",
  "dependencies": {
    "postgresql": {"status": "healthy", "latency_ms": 2.3},
    "redis": {"status": "healthy", "latency_ms": 0.8},
    "kafka": {"status": "healthy", "broker_count": 3},
    "dgraph": {"status": "degraded", "message": "Not reachable"}
  }
}

Services operate in degraded mode when optional dependencies (like Dgraph) are unavailable.

Health Check Best Practices

• Liveness probes should only check the process itself, not dependencies
• Readiness probes should check critical dependencies
• Set appropriate timeouts to avoid false positives under load
• Use startup probes for services with long initialization times
• Never include expensive operations in health check endpoints