[CentOS] Re: questions on kickstart

William L. Maltby CentOS4Bill at triad.rr.com
Fri Mar 28 23:13:01 UTC 2008


On Fri, 2008-03-28 at 15:50 -0700, Bill Campbell wrote:
> On Fri, Mar 28, 2008, Les Mikesell wrote:
> >Joshua Baker-LePain wrote:
> >>On Fri, 28 Mar 2008 at 4:32pm, Ross S. W. Walker wrote
> >><snip>

> I used to use the separate /boot partition, but quit when the 1024 sector
> problem was solved, mostly because OS upgrades or installation of alternate
> distributions in a different partition for ``/'' would frequently result in
> a less than useful /boot setup.  Having /boot on the ``/'' file system
> isn't as vulnerable to poorly written installation and upgrade scripts.

I still use it just because I can mount it ro, if I want to keep it
mounted and that reduces chance of FS corruption if we should an
unexpected disastrous event. Although I don't anymore, I also used to
have separate ro partitions for the infrequently changing "critical"
parts of the root file system. With the robustness of things these days,
I've found that less useful.

But because I still do a lot of dinking with stuff, the separate root is
worthwhile (see below).

> 
> Being a belts and suspenders guy, I don't boot from raid or lvm file
> systems as there are too many ways things can go bad.

Only through the grace of high metabolism, I'm still just a belt guy,
even at my advanced physical age. Mentally, I'm suspenders sometimes
though.

> 
> I generally build systems with two identical ext3 partitions for ``/'' and
> ``/backroot', swap, and the remainder in ``/home''.  Once the system is
> installed and configured, the ``/'' is copied to ``/backroot'' with the
> ``/backroot/etc/fstab'' file edited appropriately and ``/boot/grub/menu.lst''
> set up to allow booting from the ``/backroot'' partion (which isn't
> normally mounted).

I do the same with one minor change, ignoring naming conventions. All
systems have two bootable drives and my backup boot and base root are
duplicated, with minor mods, to the second drive. That gives me a
fallback not only for my screw ups, but also for HD failures.

> 
> This provides the ability to boot a damaged system from ``/backroot'', and
> a fallback position if an upgrade goes south by refreshing the copy just
> prior to doing the upgrade.

Great minds think .... I know, I know. Trite.

> 
> Bill
> --
> <snip sig stuff>

-- 
Bill




More information about the CentOS mailing list