Health indicators
When spring-boot-actuator is on the classpath, jEAP Messaging automatically registers a health
indicator.
jeapKafka
The jeapKafka indicator checks broker connectivity for each configured Kafka cluster by calling
describeCluster() via the Kafka Admin API. It reports clusterId and nodeCount per cluster and is
available at /actuator/health/jeapKafka.
Probe integration
Probe integration is opt-in: jeapKafka can be added to the liveness and/or readiness probe groups.
management.endpoint.health.probes.enabled=true
management.endpoint.health.group.liveness.include=livenessState,jeapKafka
management.endpoint.health.group.readiness.include=readinessState,jeapKafka
The same down-after grace period applies to both.
Recovery grace period
The indicator reports UP with a recovering: true detail for a configurable grace period after a
failure is first detected, and only flips to DOWN after it elapses — giving Kafka's built-in
reconnection time to recover.
The total time before a probe sees DOWN = down-after + (failure threshold × check interval). With
the default down-after=35s:
| Platform | Probe settings | Time before DOWN |
|---|---|---|
| Kubernetes | failureThreshold=5, periodSeconds=5 | ≈ 60s |
| AWS ECS | unhealthy_threshold=5, health_check_interval=15s | ≈ 110s |
Properties
| Property | Default | Description |
|---|---|---|
management.health.jeap-kafka.enabled | true | Enable/disable the indicator |
management.health.jeap-kafka.response-timeout | 1s | Timeout per broker check |
management.health.jeap-kafka.down-after | 35s | Grace period before reporting DOWN |
management.health.kafka.enabled | false | Spring Boot's built-in single-cluster indicator, disabled by default |
spring.kafka.listener.auth-exception-retry-interval | 10s | Retry interval for auth failures in listener containers |