Shell Programming and Scripting

You boot from DVD into single user.

You use 'format' to label and slice the disk. This determines what sectors on the disk the 's0' (slice 0) occupies. At this time there is no filesystem on that slice (and therefore no Lost+Found).

You use 'newfs' to write a new filesystem to s0 so that the OS can mount it. It contains a proper UFS filesystem with superblock, inode table, etc. This step creates the Lost+Found at the top level of s0 and nothing else.

Now you know the device node of the root filesystem that you are trying to restore to this new clone, eg, c0t0d0s0???, you need to mount this on /a mountpoint, eg,

Code:

# mount /dev/dsk/c0t0d0s0 /a

(As previously discussed, you cannot be in /a when mounting otherwise you get 'busy'.)

After the mount has completed without error, if you go into /a directory you should see the Lost+Found. If you don't, then either the newfs didn't work, or the mount didn't work.

/a will be empty if the mount didn't work because you are simply looking at the /a directory on the DVD. After the mount, you should be looking at the apex of c0t0d0s0.

Try issuing the command:

Code:

# mount

to check that c0t0d0s0 is indeed mounted on /a

1. You prepare your empty hard disk root filesystem and mount it on /a.

2. You plumb, configure and up your network interface and mount your remote NAS on /mnt.

These are separate operations and it matters not which order you do them in.

You then check that in /mnt you can see your NAS files.
You then check that in /a you can see Lost+Found on the empty new root filesystem.
If all is well then you stay in /a and issue the ufsrestore.

Thanks! It worked and I have a recognizable root partition. So- don't format and mount the other partitions now? Do this?:

10. 'sync' and 'umount' the NFS storage and the root hard disk and do an orderly shutdown.
11. System should now boot.

Note that if your /usr filesystem is separate from the root filesystem you should consider backing up that for emergency recovery too since without being able to mount /usr the recovered system will probably go into maintenance if it cannot mount /usr

Yes, is your /usr filesystem a separate filesystem????

If so, you'll need to ufsrestore the dump of that too, otherwise the clone will go into "maintenance mode" when it boots. All other filesystems you can restore (by whatever means, eg, ufsrestore, cpio, tar) after the system is on its feet.

To enable the system to boot from the disk you will need to install the boot blocks into sectors 0 & 1 of the disk (like MBR on Windows). There's an 'installboot' command on the DVD to do that. Search this forum for that. Any questions post back here.

Is the clone hardware identical to the production box? If not, there may be some adjusting to do, modifying /etc/vfstab, modifying /etc/system, setting up new device nodes eg, /dev/dsk/c4t1d1s0 if the hardware is different.

Unmount your NAS connection with:

Code:

# umount /mnt

unmount your hard disk root filesystem with:

Code:

# umount /a

Once you've installed the boot blocks close the system down in an orderly manner:

# init 0

I'm sure you'll have more questions but perhaps you should start a thread as this one's already quite long.

Ehh.. Don't know if I screwed up or not here- but I could not

Code:

umount /a

because it was busy. I made sure I was not in that directory as well. Did the

Code:

init 0

anyway and now the machine keeps rebooting over and over saying something like

Code:

cannot mount to root

Is this hardware the same as from where the dump was taken?

If not, it's quite possible that you need to recreate the root filesystem device node. Also, it's possible that because root did not umount cleanly it now needs to be fsck'd before it will mount. I suggest that you:

Code:

# cd /

to place you on the DVD before you attempt to umount /a or /mnt.

You can boot from DVD and see if you can mount the hard disk root filesystem on /a without it complaining.

Also, something that we haven't mentioned so far, have you allocated a swap slice?? Look at the production system VTOC to see how big that swap space is.