Can the failure of a post dependency fail the build?



Can I use post-dependencies as gatekeepers to promoting a build? I would write a script that curls several health-check endpoints in the app. If they succeed, then they’ve acted as a battery of integration tests.

Are there reasons against thins kind of health-checking as a gatekeeper? A host of services exist to monitor just health endpoints (Pingdom, UptimeRobot, etc, etc, etc). Running cucumber/selenium integration tests, in my experience, takes longer to run when compared with curling a battery of health endpoints.

I also recognize that theses are things often handled by different teams — but also I’d like to strive for centralized definitions (like this), which multiple teams & processes can use in a Continuous Deployment environment.