Reactive Health Service

[ojcarfr]

A health endpoint can contain a lot of health check policies, so we need a service in order to run all of such checks, reacting to every health check status change. This strategy will allow avoid denial of service throttling among check operations.

Every health check can be declared either to be executed on its own polling process (buffering the result) or to be executed on demand by the runner. Therefore, seems like the runner should be declared in the same way, buffering all the results whether is declared to poll checks.

The expected health check result is defined as:

• Health check name.
• A boolean value indicating health success.
• Health check status (from a defined status set).
• Result message.
• Execution cost (in timestamp).
• Tags? (could be useful on monitoring pulling).
• Additional info?

The runner result will group every health check result, defined as:

• Is health (boolean).
• Service status (useful on warm up).
• Execution cost (in timestamp).
• Check result collection.