CAMEL-19648: readiness/liveness probes #1010
Conversation
| management.endpoint.health.group.readiness.include=readinessState,camelReadinessState | ||
| ---- | ||
|
|
||
| Using camel specific readiness and liveness health indicators, the probes will be augmented with camel components |
| @@ -0,0 +1,34 @@ | |||
| package org.apache.camel.spring.boot.actuate.health; | |||
| @@ -0,0 +1,39 @@ | |||
| package org.apache.camel.spring.boot.actuate.health.liveness; | |||
| @@ -0,0 +1,51 @@ | |||
| package org.apache.camel.spring.boot.actuate.health.readiness; | |||
| if (!HealthCheck.State.UP.equals(result.getState())) { | ||
| isUp = false; | ||
|
|
||
| result.getError().ifPresent(error -> log.warn(result.getCheck().getId(), error)); |
There was a problem hiding this comment.
Hmm do you want WARN logs every time this is invoked and there is some kind of error.
I think either this need to be an option your can turn on|off
There was a problem hiding this comment.
I can lower the log level to debug, I think it makes sense to log the error since by default the readiness/liveness probe shows only UP or DOWN without further information.
Anyway I'll add an option CamelHealthCheckConfigurationProperties to enable/disable the log
There was a problem hiding this comment.
Does spring boot not have more details you can set/enrich to the object your return. It should not be every custom check that has to do their own kind of logging or whatever - SB should really deal with this generally.
There was a problem hiding this comment.
Well, you can configure management.endpoint.health.group.liveness.show-details=always but, it will only add information regarding the beans status, ex:
{
"status":"UP",
"components":{
"camelLivenessState":{
"status":"UP"
},
"livenessState":{
"status":"UP"
}
}
}
nothing really interesting for debug, anyway, kubernetes probes does not need it
There was a problem hiding this comment.
What about the message itself, is the id (result.getCheck().getId()) good enough as message content?
There was a problem hiding this comment.
@davsclaus in case of a disconnected kafka consumer, the error is logged only when http://localhost:8080/actuator/health/readiness or http://localhost:8080/actuator/health/ are invoked, but the console is full of warns from kafka, do you think it is worth to have a flag to enable/disable the warn for the probes?
| public class CamelAvailabilityCheckAutoConfiguration { | ||
|
|
||
| @Bean | ||
| @ConditionalOnMissingBean(CamelLivenessStateHealthIndicator.class) |
There was a problem hiding this comment.
In which case, can it be useful to allow a custom creation of a CamelLivenessStateHealthIndicator?
There was a problem hiding this comment.
I do not think there is a real usecase, but we have the ConditionalOnMissingBean for almost all the beans in camel (health check one for example), but I agree with you, I do not see you someone should use a custom implementation, I will remove it
| @Bean | ||
| @ConditionalOnMissingBean(CamelLivenessStateHealthIndicator.class) | ||
| public CamelLivenessStateHealthIndicator camelLivenessStateHealthIndicator(ApplicationAvailability applicationAvailability, | ||
| CamelContext camelContext) { |
| } | ||
|
|
||
| @Bean | ||
| @ConditionalOnMissingBean({CamelReadinessStateHealthIndicator.class}) |
There was a problem hiding this comment.
ditto + the braces can be removed here
| protected AvailabilityState getState(ApplicationAvailability applicationAvailability) { | ||
| Collection<HealthCheck.Result> results = HealthCheckHelper.invokeLiveness(camelContext); | ||
|
|
||
| boolean isLive = CamelReadinessStateHealthIndicator.checkState(results, LOG); |
There was a problem hiding this comment.
Calling a method from CamelReadinessStateHealthIndicator in CamelLivenessStateHealthIndicator seems confusing, maybe the method should be moved into an utility class instead, WDYT?
|
|
||
| private static final Logger LOG = LoggerFactory.getLogger(CamelLivenessStateHealthIndicator.class); | ||
|
|
||
| private CamelContext camelContext; |
|
|
||
| private static final Logger LOG = LoggerFactory.getLogger(CamelReadinessStateHealthIndicator.class); | ||
|
|
||
| private CamelContext camelContext; |
| if (!HealthCheck.State.UP.equals(result.getState())) { | ||
| isUp = false; | ||
|
|
||
| result.getError().ifPresent(error -> log.warn(result.getCheck().getId(), error)); |
There was a problem hiding this comment.
What about the message itself, is the id (result.getCheck().getId()) good enough as message content?
Croway
force-pushed
the
CAMEL-19648
branch
2 times, most recently
from
5d4fee3
to
c080185
Compare
|
thanks @essobedo for your comment, I refactored the logging and the class structure, this is the log in case of failing kafka consumer: where kafka route is the id of the route. |
|
This is a small reproducer |
|
@Croway we can look at that WARN thing later. Lets first complete this work and ensure CSB works nicely with SB k8s ready/live checks. You are welcome to create a JIRA so we wont forget about that other thing |
No description provided.