[K12OSN] Info on Upgrading from RedHat 9 to Red Hat ES 3.0

[Glenn Arnold]

This weekend I upgrade the school customer I work for from RedHat 9 to
RedHat Enterprise Edition 3.0 Academic edition.  
Here is Summary of what I did to migrate to RedHat Enterprise.

	First thing I did is used Webmin to do a filesystem backup to
another samba server to have backup of the home, var, and etc.  Also, I
ran fdisk and did a print out the partition setup to a text file so I
knew what partitions needed to be added to mtab and fstab.  I copied
/var /etc on my /home partition since it was separate from the root
partition, so I could access my old server configuration files easily.
I also made an image backup of the server with ghost.  My server specs
are Proliant 7000 Dual 450MHZ Xeon II with 2MB cache with 1.5GB of Ram
with 108GB disk in a Raid 5 setup.  This server is a Samba PDC used for
file and print sharing (Cups and Samba), DNS,  and DHCP.  I did have one
luxury I have several Compaq Proliant 3000,  5500, and 6500 servers that
I could use to restore my ghost image file to do a test migration.  I
deleted the root partition on my test box.  Then I partition the root
partition to make it the same size as the original root partition.
Finished the install ES 3.0.   On booted up RedHat detected the
different RAID controller and network controllers.  I changed the
network card configuration to a different IP address, so I could run the
up2date.  When the server booted I proceeded to run Redhat up2date and
it would not download any updates. The reason was my /boot partition was
too small.  The partition was only 100MB.  Since I did this on my test
migration box I was not too concerned about this I new I could
reconfigure the partition size to a bigger size when I actually do the
migration.  I tried to resize the partition with Partition Magic 8.0,
but it keep giving me block errors and would not resize the partition.
So, I was going to restore my server image backup on my test box, but
user computer problems did not allow me enough time during the week to
do another test migration. My plan was to reconfigure the server on my
test server and then restore it on my production server, but it just did
not work out that way. Since the migration needed to get done before
summer school I took the server off line and made another image backup
and  dump of /etc /var /home on Friday night.  I came in on Saturday
Morning and boot off the Enterprise 3.0 Cd.  When anaconda came up I
went through the wizard and selected the manual disk portioning option.
If you choose automatic disk portioning it wants to wipe out all your
disk partitions which would added a lot of time to the process, because
you would have to restore all your data back.  Since I had /home and
/smbsrvr (My directory for netlogon ,windows shares, and apps) on
separate partitions I did not have to restore any data just server
configuration files in /etc and /var.  I manually delete /boot and /
with disk gruid.  I reconfigure /boot to 500MB and / to 9.5GB.  Both
partitions are EXT3.  Continue through the Anaconda wizard and feed the
installation CDs to the server.  When the server was finished and booted
into Enterprise 3.0 I login in and startx.
When X came up proceed to run up2date and register the server with
RedHat and get the latest updates.  For some reason it took an hour to
get the updates downloaded and installed.  If I had a speedier box it
might have went faster.   I rebooted the server after the updates where
installed to make sure the new kernel patch was running and the updates
did not cause any problems.  The server was running with now problems,
so I proceeded to edit the mtab and fstab to mount the /home partition.
Once I mounted the /home partition I open up the fstab and mtab config
files in my /etc backup on my /home partition and cut and pasted the
missing partitions in my Redhat ES fstab and mtab config files.   I
restarted the server to make sure all my partitions mounted.  The
partitions mounted fine.   Then I started to compare my RH9 /etc config
files with RH ES 3.0.  Most of the configuration files where the same,
so I backup RH ES config files in backup directory for each service I
was changing.  For Example:  under /etc/samba I made a directory
/etc/samba/backup and copied the config files from ES to the backup
directory and copy my old RH9 config files in the Samba directory.  I
also copied my DNS and SAMBA data into the proper directories /var.   I
used the gui services tool to set DHCP,DNS, and Samba to start on boot
and started the services.  Everyting transferred except for Samba.
DHCP, DNS, and CUPS worked fine.  All my CUPS queues where there and the
queues worked.  I was pleasantly surprised how well those services
transferred especially, since I have only been using Linux for two years
my Linux skills are not the great some of you experienced Linux Guru's
on this list.   My only problem was Samba would start, but I could not
connect to the server.  RedHat ES 3.0 has there version of Samba 3.02E
which I assume this is patch version with the printer patch.  My RedHat
9 server was running Samba 3.04.  Since I had a new version of Samba I
guess I could not down grade to ES version.   I used the RedHat 9 RPM
version of Samba from samba.org.  The RPM installed, but something in
the install script did not work. Even though the setup script bombed it
copied all the executables and Samba started to work.  The only side
effect I have seen so far is winbind does not show up in the services
gui and it does not start automatically.  I can start winbind from the
command line without any problems.  All my trust relationships work
fine, so I just need to edit the startup script and add winbind I think.
The other side effect of migrating to RedHat ES 3.0 is webmin 1.50 you
can not install the SSL piece for secure login into webmin.  I am not
sure how to fix this yet.   I hope this info helps this news group out,
because this news group has help me out a lot.   Please give me some
input if there is a better way of doing this migration.

Thanks
-Glenn