Solaris

I use fssnap and ufsdump for regular backups by way of Amanda. I have run trials of recovering a system from ufsdumps. Boot off CDROM, plumb network interface, temporarily open up rsh between the two machines on our local private network, and do a ufsrestore from a remote tape drive. Instructions for all of that are in the Systems Administration documentation collection for Solaris on docs.sun.com. If you have a local tape drive, of course, it is slightly easier. I actually had to do the disaster recovery for an E450 running Solaris 8 using a remote tape drive a few years ago. It worked.

For an additional option, I'm in the process of setting up a custom boot CD that contains the Amanda client for a special recovery identity (so it is already set up on our DNS with an IP and doesn't conflict with any machine already on the net), that will allow me to boot off CDROM and then do an amrecover, accessing the Amanda server and tape drive to recover from.

If I am wrong plz correct me.

when using #ufsdump tvf /dev/rmt/0 /
then can backup as required. after that when it is nessesory u can restore using following command.

#ufsrestore ivf /dev/rmt/0

The question was about reinstalling on another disk. Flar is a much easier solution for that. ufsdump goal is to allow restoring a filesystem.
flar goal is to allow installing Solaris from a reference machine.

ufsdump isn't going to make a bootable disk.

that was and is my point...

ufsdump/ufsrestore is slow.
mirroring the disks is one option.
another method is to use flar (flash image installation) - once you create the flar image, copy it somewhere. When you need to install on a new disk, insert an installation DVD, follow the prompts, then choose "flash installation".

You can still refer to this site: Solaris 10 Installation Guide: Solaris Flash Archives (Creation and Installation) - Sun Microsystems

Thousand times thanks for every one. I got new knowledge. Thank you again friends

Just to be clear about procedures -- and not to put down flash archives as a way of doing things -- ufsdump/restore has been used for a long time to recover systems. It may not "create" a bootable disk on its own, but standard procedure is to do a ufsrestore of the partitions, and then install boot blocks. You will either be booted from a Solaris CD or from a hard drive and recovering to another hard drive, and it is a one line command. No big deal.

I once had a script running off cron at 1am on a particular machine that copied the boot drive format to another drive, ufsdump/restored all the boot drive partitions to the other drive, installed boot blocks, and used sed to fix things like /etc/vfstab and move the clone script, so, if the first drive failed and the system rebooted, it would just boot off the second drive successfully (and not try to clone itself again until I got a new drive). In that particular configuration, the edit of the vfstab was because it was originally configured to boot off, say, c0t0 and if the second drive became the active boot drive it would be booting off, say, c0t1. I could swap the drive, but the idea was to be able to enable a quick remote or automatic restart.