Currently, we only calculate cluster health once.
With the addition of the watch_api, we need a way to update it
when new data is received.

Changelog: fixed