The HQ server considers an agent not starting as the platform being
unavailable. The unavailability of a platform is determined by the
fact that an agent does not report available for it. Therefore,
what's important is the HQ server be running without interruption to
ensure that our "backfiller" process fills in the missing available
data points as unavailable to give you an accurate accounting of the
percentage of time that a platform is available.
To monitor the availability of a website, you can set up an HTTP
Platform Service by manually creating a new Platform Service for any
platform in your infrastructure. The HTTP service has an
availability metric already, so there's not too much to it. It does
depend in the proper functioning of the HQ agent to collect
availability, and would be considered unavailable if the agent was
not running.
Charles