Hello,

I need a complete list of the files in the /etc directory of a RedHat
8.0 server that configure hardware or networking characteristics of
the server.

To give you a better idea of what I need, let me tell you what I'm
doing.  I am migrating a RedHat 8.0 installation that is located on a
virtual server over to a dedicated server (also running RedHat 8.0). 
I wish to preserve not only user data, but all of the applications and
custom configurations that I have made (such as qmail, apache, horde
etc.)

In order to do this, I have set up rsync as a daemon on the new server
and written a script on the client side to synchronize the files.  My
script included the /bin, /command, /home, /sbin, /var, /lib,
/service, /usr, and /etc directories for syncing.  It does not include
the /proc, /tmp, /boot, /dev, or /root directories.

The problem arises when I sync the /etc directory.  Since I do not
have physical access to either server, I can't just insert a CD and
modify files like /etc/fstab, /etc/hosts, /etc/sysconfig/network etc. 
Hence the need for a complete list of files that directly relate to
hardware instead of software configuration so that I can exclude them
from the sync.

---

Clarification of Question by astanford-ga on 12 Aug 2003 09:49 PDT

Thank you for responding!  Unfortunately, that doesn't really answer
my question, as so many files are distro-specific.  To give examples,
on RedHat, I have isolated the following files (excluding those you
have already mentioned).

/etc/sysconfig/
/etc/hosts
/etc/hostname.eth0
/etc/sysctl.conf
/etc/hotplug/
/etc/grub.conf (same as lilo)

I would like to have reasonable confidence that the list is
comprehensive for RedHat 8.0.  If a couple of files were missing and I
had to go through the procedure once or twice, that would be fine, but
I'd prefer not to have to iterate more than a couple of times.

The dry run idea is great and I plan on implementing it, but if it
fails I have to reload the OS each time :-(

---

Request for Question Clarification by maniac-ga on 14 Aug 2003 16:40 PDT

Hello Astanford,

Are you still interested in a comprehensive answer? I have access to a
few Red Hat systems with different network set up's but the same list
of daemons enabled. What I would provide is:
 - the list of files different between two systems (and why different)
 - suggestions on how to ensure the list is correct in your specific
case
 - suggestions on tools (other than rsync) that can help configure /
clone systems that can account for different hardware / network
equipment
Is this the kind of answer you are looking for?

  --Maniac

---

Request for Question Clarification by maniac-ga on 21 Aug 2003 16:47 PDT

Hello Astanford,

Are you still looking for an answer?

If not, I suggest you close the question so you can't be charged for
it.

If yes, I have collected information from systems I am managing to
answer your question.

  --Maniac

---

Clarification of Question by astanford-ga on 25 Aug 2003 06:13 PDT

Sorry to not get back to you earlier, but I had to go out of town and
didn't have email.

To summarize the results I obtained, here is a complete list of the
directories and files in a RedHat 8.0 distribution that have strictly
to do with hardware and network interfaces:

/mnt/
/proc/
/tmp/
/boot/
/dev/
/root/
/etc/mtab
/etc/fstab
/etc/sysconfig/
/etc/host*
/etc/cups/
/etc/foomatic/
/etc/hotplug/
/etc/modules.conf
/etc/ppp/
/etc/print*
/etc/raidtab
/etc/sound/
/etc/vfs/
/etc/X11/

Thank you to everyone for clarifications; I hope this may help others
in the future.

I don't have access to a Redhat 8.0 machine so I can't answer this
completely.

Depending of the setup, you can often get away with this sort of
migration by moving /home (all user files) and /etc (less some files,
as you say) and depending on the software and setup some directories
in /usr and /var

The files on my Debian system in /etc that relate specifically to that
machine:
X11/*
apm
fstab
inittab (arguably)
lilo.conf
modules
modules.conf
network/*
passwd (well, not really, but it can pay to go over this manually
first)
serial.conf (maybe)
shadow (see passwd)

Also, potentially some of the startup script in init.d/ and rc?.d/

These are the ones I spotted in my config, which is a Debian 3.0
system.

If you are able, the best way to do this is to do a dummy run. Setup a
similar system locally where you can get physical access, move the
data from your shared server to that, and get it to work in a
controled environment, then when you have a plan worked out do it for
real. And make sure you have a backout plan :)

Regards,
Sycophant-ga