Everything start_managed does before starting the server: the download, and the cluster.
These are the slow parts (seconds to tens of seconds, the first time only), so code with a
time limit on its requests calls this first, somewhere without one, and start_managed is
then quick. Safe to call again and from several processes; it returns an error text.
prepare_managed :: (settings: Managed_Settings) -> error: string
Parameters
- settings: Managed_Settings
Returns
- error: string