On Thu, 2003-05-29 at 09:28, Troy Dawson wrote: > *Troy thinks out loud as well, though a little better rested than Erik* > Hmmm .... > Very interesting idea. > So it COULD be sortof like a kickstart. > A person could do a bare bare install, enough to have the kernel and yum > installed. Then point their yum runscript at this file, and it would build it > up to whatever they really want. > (I'm just thinking of some of our people using 200Mhz Pentium Pro's. The > installer sucks up so much memory these things take hours to do a minimal > install. Though this is the exception, not the rule around here) > *Troy stops thinking ... out loud at least* > that was the majority of why. We're changing the layout for how we do additional packages to the distro here at duke. instead of integrating them all into the install tree so anaconda can install them we're just going to setup some sensible yum repositories and let people install the extra stuff that way. But I want to do that simply - and that means not having 40 yum install foo bar baz statements in my %post. > Yes ... but what if you could somehow get the template to point to it's own > yum.conf, or even multiple yum.conf > > Or something like > > installgroup Groupname > installgroup AnotherGroupname -c http://path/to/yum.conf > updategroup Groupname > remove pkgname > install pkgname > update pkgname -c http://path/to/yum.conf > # comment > ; comment > well I'd suggest then that you pass it on the command line, not INSIDE the template/script. ie: yum -c configfile template file the problem with including the config inside the template is that it would mean reloading and rerunning the entire shebang to make sure you had the right data. And I'm not sure how I would calculate what needed to be installed in those places. ie: what if there were two config files pointed to in there that had the same serverid? or had the same packages, do you want each line parsed as a single transaction? -sv
