I spent a fair amount on this and thought a few people on the list may
benefit so I'm going to post as a summary...This is my own sumary...Joe
may post his own after collecting all responses...
C
If you keep your backup report every night, it will tell you which tape
contains the image of the backup server's full backup. This copy should
contain the indexes for the systems as well. If not, it's just a matter
of rebuilding the indexes from tapes.
Disaster recovery to me is to have the hardware inventory for the machine,
the OS media, all notes on partition tables, special configurations, etc...
and the Networker media disk along with license enablers. All deposited
in a bank nearby somewhere.
Learn the disaster recovery process for Legato. Go to their web site,
you will see the process sheet. Basically, reinstall on same hardware,
input enabler code, insert the backup media and do an indexing operation.
Once you got the index for the backup server, you can restore its old
image. This old image is the exact set up of the machine before it
crashed...You may have to re-index all client's indexes but maybe you
don't need to because the clients don't neccessarily need a restore
right away.
Remove all this worries and boring process by buying and install a
second disk. Use vdump/vrestore and back up your root image. Note
that I emphasized the sizes of the partitions and the system configuration
information above. You should have these info written down and deposited
in a bank. Restoring for this scenario involves some pre-planning...
You need to bring down the machine, install the OS on the 'backup' disk--
the disk you are going to use for a second copy of your root drive. This
way the disk is bootable. You can skip doing all this if you can look
at the 'disklabel' command which write a boot block for you.
Now if you boot the system back to the primary disk, both disks should
have the same partition layout PLUS, the secondary disk is also bootable!
Feel free to set up a cron job to do this every month or so...
mount /dev/rzBACKUP /mnt/rootbackup && vdump 0vf - [/, /usr, /] |
(cd /mnt/rootbackup; vrestore xvf -) >/dev/null 2>&1
...
You can save this image on another NFS disk.
vdump 0vf - / | rsh remote "cd /dir; vrestore xvf -"
For simplicity, make it
a disk on the same subnet so you don't have to deal with the gateway.
To recovery from this scenario, you will still need the partition
table. Boot the system from the CD, install the OS...right after
it's done with creating the partitions the way you wanted and wanting
to load the OS, you can cut the power...(this is so you save time)
because by this time, it's done writing the boot block! and you also
got the partitions nicely created. Boot again in single user mode
with the CD-ROM, create the devices for the disk, mount it...and do
ifconfig tu0 IP netmask #experiment a little and also save the
#network info for the node before
hand.
Now you can mount the remote disk and do vrestore...
The procedures for AdvFS is a little more involved but doable. There
are also ways to improve the process so it can take the least time...
etc.
On a side node, I like Compaq/Tru64 in this regard...so you shouldn't
worry too much about bootstrapping. There seem to be some attention
to this area when they designed the OS.
I learned all this because I had some difficult customers so I backed
up everything before upgrading...In some cases, I had to completely
restore...In addition to this immediate data, I also had them on
backup tapes.
The more difficult OS I had to deal with was IRIX because it was all
GooEY (GUI)...But I also managed.
SUN is easy and there's a procedure on legato's web site.
HP is relatively easy too...
Don't know about AIX...there's this mysterious mksysb to deal with.
But I think legato also has a procedure for this.
-----Original Message-----
From: Joe Shields [mailto:jshields_at_post-gazette.com]
Sent: Thursday, September 07, 2000 5:34 PM
To: 'tru64-unix-managers_at_ornl.gov'
Subject: flaw in our backup/recovery
> ----------
> From: Joe Shields
> Sent: Wednesday, September 06, 2000 3:56 PM
> To: 'alpha-osf-managers_at_ornl.gov'
> Subject: flaw in our backup/recovery
>
> We inherited a Digital Alpha box running Digital UNIX v4.0d. The RAID5
> Scsi disk set and the system disk are being backed up using Legato
> NETWORKER. However, if we lose the system disk, we lose Legato NETWORKER
> and our way to get everything back.
>
> We are more experienced with VAX VMS, where we have the ability to do
> Image backups to spare drives. Our strategy for recovering our system
> disk and all the system software (like our backup/recovery app.) that
> resides there is to always have a fairly current image backup of our
> system disk on a spare drive that could be pulled into service.
>
> Is there a way to do something like this with Digital UNIX? We've tried
> using the dd command with identical drives, but the resulting copy would
> not boot and the copying was disappointingly slow.
>
> We are hastily reading Digital UNIX documentation and trying to figure out
> the best strategy for fixing this flaw, but could sure use, and would
> appreciate any help or advice that could be offered by this group.
>
> Thanks
> Joe Shields
> Pittsburgh Post Gazette
>
>
>
Received on Fri Sep 08 2000 - 15:13:07 NZST