With the cron-synchronized scripts, if they do some communication with
our servers, do a random-time sleep, so they don't rush in all at once.

How on the earth did this ever work?

Using the atsha204cmd command line utility.

So the old ones are not sent again.

Omit the old lines.

Log the error to restart a service instead of crashing.

Without de-duplication, but works.

Especially the updater. If the updater restarts cron, the updater was
killed before. This is meant to avoid that.

They are needed, since this is global configuration.

A script to be run periodically and check for health of daemons.