On 9/20/10 10:45 AM, "Jiri Horky" <jiri.ho...@gmail.com> wrote: > On 09/16/2010 12:02 PM, Jan Kundrát wrote: >> I also got the feeling that it's normal to copy "all files" into each >> node's staging area, even if such files are not needed on that >> particular machine. >> >> .... >> and the second one is why to copy all files for all services to each >> machine. > I too would be interested in the answer for this question. Personally, I > don't see any good reason for distributing all the configuration to each > node, whereas I can think of at least two disadvantages of such an > approach (wasted traffic, possible security problem).
It's a matter of configuration. You've already solved something similar if you have dev/qa/stage/etc. through which you promote new policies. Here, we do that by declaring a $(masterfiles) variable early on which maps to different checkouts on our cfengine servers. This means policies all "copy from the same place", but that place is a variable which can changed based upon things like FileExists or a hostname regexp (if you want to keep it simple) or a web call to an inventory database keyed on something unique like serial number which returns class data (like host/cluster type that can map to a CVS tag or some other collection of files). You can segregate as much as you like through configuration, so 'all files' on client X does not necessarily mean the same as 'all files' on client Y. Hth! _______________________________________________ Help-cfengine mailing list Help-cfengine@cfengine.org https://cfengine.org/mailman/listinfo/help-cfengine
