Matthew Palmer wrote:
On Fri, Dec 09, 2005 at 12:43:04PM +1300, Steve Wray wrote:
It is still useful to be able to collect data on the existing state of a
machine and pass that back to the cfengine server. I can think of *lots*
of reasons one might want to do this.
Please share. I can't think of anything that cfengine can do on the server
that it can't do on the client.
Not necessarily for cfengines direct benefit.
Eg. I'm in the process of converging a set of debian hosts to management
under cfengine.
I have a partial system for doing package upgrades and at the moment I
would very much like to pull in, to the cfengine server, each hosts
current package selection states so that these can be compared and I can
use these to work toward convergence, detect implicit host classes etc.
I agree that once a set of hosts are under cfengine control and
converged the need is much reduced, but what a lot of people tend to
forget is that bringing hosts to convergence takes a lot of careful
examination of how each host is configured; finding the implicit classes
and making them explicit.
--
There is nothing more important to good government than good education.
_______________________________________________
Help-cfengine mailing list
[email protected]
http://lists.gnu.org/mailman/listinfo/help-cfengine
