One of the reasons that etch 3 does everything on the server is that it gives the server the ability to generate a set of configs for a client at any time. I.e. the server does not need an active connection to a client to generate configuration for it as all state used to generate the configs for a client comes from the client's facts and node groups.

What this would allow us to do is to see what the impact would be of a change to the configuration repository. I.e. if you changed some files in the repository we could generate configurations for all clients and diff them against the last configuration sent to the client (as those are stored in the database). This could be used to help predict the impact of a change or to do a dry run of a change.

This might be particularly useful for changes which you expect to affect a small number of hosts. If the impact prediction is that the change will actually affect a large number of hosts then you probably need to double-check your work.

Moved from https://sourceforge.net/apps/trac/etch/ticket/20