by I'm working on putting together a complete set of instructions for backing
up SCO, and for rebuilding a new HD if you smoke the old one. I've put
together the following set of instructions from my (insufficient)
experience, from docs, and from posting questions on newsgroups.
**WARNING**
I have *NOT* tried this yet. I want to pass it by the people in this
newsgroup and get any obvious errors fixed. Then I will try it and post
results. Eventually I want to have detailed and dependable enough
instructions to add to a FAQ.
**NOTICE**
Will those of you who keep emailing me to tell me to just use a commercial
product please CUT IT OUT! I know all about the commercial products, and
I have purchased one. That is *not* the point of this exercise.
[rant mode off]
So, with that in mind, here's what I have. Please send comments and
(This is cut-and-paste from a WORD doc. I hope it comes out legibly)
-------
BACKUP AND RECOVERY
BACKING UP A SCO SYSTEM
AND
RECOVERING FROM A CRASH
BACKUP
1. Install all hardware and drivers, setup all network cards, install all
printers, set up all users. Essentially, get the system completely ready
for production.
2. Do a tape backup of the entire system. Use cpio, as tar doesn't save
important things like device inodes and empty directories. Use the
following:
echo "/" | cpio o Abd O /dev/rct0
Note that the cpio options should be customized for your particular
installation. The device name may not be the same, you may have to specify
volume size and block size, etc.
Also, you may want to unmount any filesystems that exist on another
physical HD before doing the tape (or be more specific than "/"). If you
ever need to recover the boot HD, you don't need to or want to restore this
data into the root partition.
3. Make the emergency boot and root floppies, using mkdev fd. You may have
to do some work to ensure that the floppies are complete. First, if you've
added devices, the script may not be making the floppy filesystem with
enough inodes. To change this, edit the /usr/lib/mkdev/fd file, and do the
following:
look for this line:
/etc/mkfs -y -f EAFS /dev/marry/tmp/ramdisk 5120:1000 >/dev/null 2>&1
and change the 1000 to something higher (I use 1500).
If you want to make sure specific executables are on the floppy, look for
one of these lines:
core_utils=
OR
extra_utils=
And add the executable name to the appropriate list.
4. Test the floppies by booting from them. Also try to access the tape
drive using:
cpio i I /dev/rct0 -t
DO NOT try to make the hard disk devices at this time. As part of the
"mkdev hd" process, you will probably get your existing filesystems
overwritten. This would be a Bad Thing. Although you could use mknod to
make the device nodes and access the hd at this time, you probably
shouldn't. During the recovery process you are going to have to mkdev,
fdisk, and divvy the new hard disk anyway, and having existing nodes will
just confuse things.
5. Make a backup set of emergency floppies finding out that one of your
emergency floppies is bad, right in the middle of an emergency, may cause
stress.
In addition, if you have another server or other system available, put an
image of the emergency floppies in a well-known location on the alternate
system. You can get a disk image of a floppy using this:
dd if=/dev/rfd0 of=/tmp/SomeFileName bs=512
Your floppy device name may vary.
6. Print the disk configuration. You can do this by executing "script" then
doing a "fdisk" and "divvy" for each HD, then closing the script and
printing the resulting typescript file.
7. Do a "hwconfig h" to printer. If you have to replace the entire system,
this will help you to remember the setup.
8. print the /etc/default/filesys file. This will indicate your various
filesystem mounts and mount points.
9. Put the floppies, the cpio tape, and the printouts in a safe place. One
idea is to make up a loose-leaf binder, labelled with the name of the
system, containing all the media and printouts necessary to rebuild the
system.
10. In some systems, there are other things that have to be documented or
backed up. For instance, EISA systems will likely need to have the cards in
a replacement system go into exactly the same slots. Also, you should have
a backup copy of the EISA config floppy.
And if you have to modify the CMOS settings, you should document this
(although the emergency root floppy contains a file with the settings
saved).
RECOVERING
This section assumes that you've had the system HD melt down, or have had
to replace the entire system. If the latter, the new system should be
configured identically to the old one.
1. Boot from emergency floppies.
2. Make the new disk, using:
mkdev hd
Part of this process will also involve fdisk'ing and divvy'ing the new hd.
You don't have to make the new HD the same as the old, but it would make
life easier to keep things the same.
3. make mount directories:
mkdir /mnt
mkdir /mnt/stand
4. mount the new hd:
mount /dev/hd0root /mnt
mount /dev/hd0boot /mnt/stand
Note that the name of the devices "hd0root" and "hd0boot" will be based on
what you name the filesystems when you divvy the new HD.
5. If your entire system was on one filesystem, this is enough. However, if
you had things like a separate /u filesystem, you will have to manually
mount these.
For instance, with a separate /u filesystem, you'd need to do this:
mkdir /mnt/u
mount /dev/hd0u
Note that the name of the device or devices will be based on what you name
the filesystems when you divvy the new HD.
6. Restore from the cpio tape:
umask 000
cd /mnt
cpio -i I /dev/rct0
7. shutdown:
haltsys
8. boot from floppy, but at the "boot:" prompt, enter:
hd(40)unix
9. Go into single-user mode. Execute:
dparam w
instbb hd /dev/hd0a
10. relink the kernel:
/etc/conf/cf.d/link_unix
11. reboot
There will probably be some cleanup at this point, but you should be
essentially up and running.
------------
OK, I'm sure there are gaping holes. Suggestions accepted gratefully.