Why in liveness health check group only StorageLivenessCheck is depending on status?

[FediNOP]

We are using kafka as storage and happened a couple of times when kafka is not available or certificate is expired in run time, but pod still pass livness probe. I looked in the code and noticed that StorageLivenessCheck is depended on alive status, but ResponseErrorLivenessCheck or PersistenceExceptionLivenessCheck is always up, although ResponseErrorLivenessCheck error counter is more than maximum

In logs, we can see that error counter is 2 of maximum allowed but live health check status is up

2023-11-09 09:13:22 INFO <_> [io.apicurio.registry.metrics.health.liveness.ResponseErrorLivenessCheck] (executor-thread-1) After this event, the error counter is 2 (out of the maximum 1 allowed).         │
│ app 2023-11-09 09:13:22 ERROR <_> [io.apicurio.registry.services.http.RegistryExceptionMapperService] (executor-thread-1) [500 ERROR DETECTED] : Not authorized to access topics: my-topic

{
    "status": "UP",
    "checks": [
        {
            "name": "PersistenceExceptionLivenessCheck",
            "status": "UP",
            "data": {
                "errorCount": 2
            }
        },
        {
            "name": "ResponseErrorLivenessCheck",
            "status": "UP",
            "data": {
                "errorCount": 2
            }
        },
        {
            "name": "PersistenceTimeoutReadinessCheck",
            "status": "UP",
            "data": {
                "errorCount": 0
            }
        },
        {
            "name": "StorageLivenessCheck",
            "status": "UP"
        },
        {
            "name": "PersistenceExceptionLivenessCheck",
            "status": "UP",
            "data": {
                "errorCount": 2
            }
        },
        {
            "name": "ResponseErrorLivenessCheck",
            "status": "UP",
            "data": {
                "errorCount": 2
            }
        },
        {
            "name": "PersistenceTimeoutReadinessCheck",
            "status": "UP",
            "data": {
                "errorCount": 0
            }
        },
        {
            "name": "StorageLivenessCheck",
            "status": "UP"
        }
    ]
}

[jsenko]

Thanks for pointing this out, this is a bug accidentally introduced when updating major Quarkus version. Fixing it in 2.5.2 and later.

[FediNOP]

Thanks for the answer