Inconsistent timing of checks

#### Overview of the Issue

I'm using Consul for service discovery and health check of Docker services.
I register the service with Gliderlabs Registrator and I use a custom script that reports the correct execution via webhook to an Healtchecks.io instance.
The problem is that Consul is not respecting the timing of the execution of the check. Looks like Consul doesn't respect any timing as it runs the checks with variable timing of 20 sec to 10 or more minutes.

#### Expected behaviour

1. Register a Service with script check with 30 seconds of interval
2. Consul should trigger the check every 30 seconds

#### Actual behaviour

1. Register a Service with script check with 30 seconds of interval
2. Consul runs the check at random intervals, this varies from 20 sec to more than 10 minutes

This is a wrong behaviour, as this affect the true availability of the service. Could happen that the service goes down and Consul misses this downtime because is not executing the check.