Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 28 Sep 96 00:12:21 PST
From:      BRETT_GLASS@infoworld.com
To:        hardware@freebsd.org
Subject:   Migration to larger hard disk: How?
Message-ID:  <9608288439.AA843935686@ccgate.infoworld.com>

next in thread | raw e-mail | index | archive | help
I need to migrate the files of a production FreeBSD 2.1.0-R system from a
half-gig IDE hard disk to a 2.5-gig IDE hard disk. I have not tried this
before, and would like to (as much as possible) get the procedure right the
first time.

First off, I need to know the best way to avoid trouble with the large
(>4095 cylinder) IDE drive. (Before anyone says it: yes, I agree that IDE
is an abomination, but in this case I must proceed with this drive.) How
should I set the CMOS? (The BIOS *seems* to have support for disks with
large numbers of cylinders, though it only allows a maximum of 16 heads and
63 sectors.) 

Once I've configured the BIOS to understand the drive's geometry, should I
start with the new drive as the slave? Or as the master (with the older
drive as the slave)?

Next, I need to know how to prepare the drive. When I installed
FreeBSD originally, I just walked through the menus presented by the
boot diskette. But since FreeBSD's sysinstall utility doesn't seem to work
on a system which is already running, how do I find and execute the proper
utilities to partition the drive and create filesystems on it?

How should I allocate space between the paritions on the new drive?
Currently, there are four: root, swap, /usr, and /var. /home is symlinked
to /var/home. Should I continue to do things this way, or break /home out
into its own partition?

Finally, how do I copy everything (including links) to the new drive? How
do I temporarily mount the partitions on both the old and new drives and
copy the data?  How do I then set the mount points so that, when I remove
the old drive and make the new one the master, the system boots as before?
Can I mount the old drive read-only to avoid any chance of corruption?

Since users will be depending on the machine, and we can't have TOO much
downtime, I'd like to outline a complete plan before I start and be able
to back out of the procedure painlessly if it fails.  What do folks out
there -- the experts -- recommend?

--Brett




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?9608288439.AA843935686>