On Tue, Dec 10, 2013 at 3:26 PM, Kevin Fenzi <kevin(a)scrye.com> wrote:
On Tue, 10 Dec 2013 15:06:27 +0100
Miloslav Trmač <mitr(a)volny.cz> wrote:
> This seems like a nice installation path. For a "featured role" I'd
> expect more than a nice installation, though:
> - Automatic integration to the machine-wide monitoring system (at the
> very least, answering "is it up?", ideally also some key metrics)
> - Automatic inclusion into the machine's backup regime, if any,
> including any possible special tasks (creating a consistent DB
> snapshot to be backed up, for example)
> - Integration into the machine's configuration system (e.g. reacting
> to networking setup changes?)
So, would that imply that we need at least:
machine wide monitoring
roles first that all other roles depend on?
The way I think about it, no, these would not be installable roles -
there would be a basic infrastructure included in every Server install
(whether that is a daemon or only a .d directory to drop configuration
into is an implementation aspect); one would never decide to "install"
machine-wide monitoring, it should be available for every system out
of the box.
I'm not sure there would be a consensus for this approach, having
Server somewhat "fat", even if some subset of users would not use all
of the functionality - i.e. with a "really minimally minimal" setup a)
unsupported, b) delegated to Cloud, c) a second-class citizen.