2 ms·
> You can of course use a health check in addition to the normal endpoint in your circuit breaker, but then its just a fail-faster for the scenarios that the he
by derefr 7y ago
> You can of course use a health check in addition to the normal endpoint in your circuit breaker, but then its just a fail-faster for the scenarios that the health endpoint covers.
Er, yes, that was the thing I was trying to suggest. You want to fail fast (or, more precisely, keep the system in the "circuit open, degraded functionality code-path" state) so your users don't experience mysterious latency spikes when their request becomes the guinea-pig for the circuit-breaker. That'll still happen, but not as often if a health check can tell the service there's not even a chance the service is worth checking via user-request.