On Tue, Jan 10, 2012 at 3:35 PM, <m.roth at 5-cent.us> wrote: > >> >>> What we do is build one, then create /boot/new and /new on the next >>> server, rsync over to them, then mkdir /boot/old and /old, and (using >>> zsh with modules loaded) mv * old, mv old/lost+found ., mv >>> old/new/* ., make sure a few things are correct (for example, >>> ifcfg-eth*, /etc/ssh/), and sync, then reboot. All your other stuff is >>> fine.... >> >> Have you looked at http://rear.sourceforge.net/ (and in EPEL) as a >> potential backup/clone/rollout mechanism? It seems like something > <snip> > The one difference with the method we use is that you *don't* have to > format /, and so anything you have under it is still safe. We normally > have a few directories that are local, and so need to be saved (web, a > temp that everyone can use that is guaranteed *not* to go away, etc). But that also means you don't get to re-arrange your filesystem layout to set up a bigger /boot, change filesystem types, or fix something you've learned could be better in the years of running the previous version. -- Les Mikesell lesmikesell at gmail.com