`module HealthChecker`

A health checker for managed services.

Definitions

`def health_checker(instance, timeout = @evaluator.health_check_timeout, parent: Async::Task.current, &block)`

Start the health checker.

If a timeout is specified, a transient child task will be scheduled, which will yield the instance if a block is given, then mark the instance as ready, and finally sleep for half the health check duration (so that we guarantee that the health check runs in time).

If a timeout is not specified, the health checker will yield the instance immediately and then mark the instance as ready.

Signature

parameter　instance　Object: The service instance to check.
parameter　timeout　Numeric: The timeout duration for the health check.
parameter　parent　Async::Task: The parent task to run the health checker in.
yields　{|instance| ...}: If a block is given, it will be called with the service instance at least once.

Implementation

def health_checker(instance, timeout = @evaluator.health_check_timeout, parent: Async::Task.current, &block)
	if timeout
		parent.async(transient: true) do
			while true
				if block_given?
					yield(instance)
				end
				
				# We deliberately create a fiber here, to confirm that fiber creation is working.
				# If something has gone wrong with fiber allocation, we will crash here, and that's okay.
				Fiber.new do
					instance.ready!
				end.resume
				
				sleep(timeout / 2)
			end
		end
	else
		if block_given?
			yield(instance)
		end
		
		instance.ready!
	end
end