Consistency with EnableDetailedOutput about beatpulse HOT 4 CLOSED

Hi @hbiarge

Well, the idea of BeatPulse is to be used as a liveness uri from k8s/swarm and similar. Most of these use the status code to restart the pod/container not the content of the response.

When the service is not available ( 503 ) BeatPulse return the first not running checker and break the task. Are you suggesting don't break the task and check all liveness?

from beatpulse.

Yes, exactly that, but depending on the options.EnableDetailedOutput value.

My point is that the current behavior seems correct if you are not interested in the detailed information. The first failed checker break the Task and return a 503 so the orchestrator can kill the container and start another one.
But if you enable options.EnableDetailedOutput I would expect that all the checkers run and receive the complete information in the response body. As you say, may be a good point maintain the 503 status code if some check fails, in order to be consistent with the behavior of the library when options.EnableDetailedOutput is disabled, but get the information of ALL checkers.
I have been testing the library in the context of a middleware application in a complex system with a lot of dependencies and the information shown in the health endpoint is not consistent when different dependencies fail and options.EnableDetailedOutput is enabled.

from beatpulse.

Sound ok for me, labeled as up for grabs!

from beatpulse.

Commited, thanks @hbiarge for this suggestion!

from beatpulse.