What methods can be used to determine which microservice has stopped or is unresponsive?

To determine which microservice has stopped or is unresponsive, several approaches are used:

1. Health Checks — microservices provide special HTTP endpoints, such as /health or /status, which return the service's status. Monitoring periodically polls these endpoints.

2. Heartbeat — services periodically send signals (heartbeats) to a monitoring system or message broker. The absence of a signal for a certain period indicates that the service is unresponsive.

3. Monitoring and alerting — using monitoring systems (Prometheus, Grafana, Zabbix) to track metrics and service availability. If a service stops responding, an alert is generated.

4. Service Discovery with availability check — in systems with dynamic service discovery (Consul, Eureka), built-in health check mechanisms are available.

5. Logs and tracing — analyzing logs and distributed tracing helps identify failures.

Example of a simple health check in Go:

func healthHandler(w http.ResponseWriter, r *http.Request) {
    w.WriteHeader(http.StatusOK)
    w.Write([]byte("OK"))
}

http.HandleFunc("/health", healthHandler)
http.ListenAndServe(":8080", nil)

Monitoring will poll /health and consider the service alive if the response is 200 OK.