VTOCs & Partition Tables

How fussy is Solaris about what gets written to the VTOC using fmthard ?

If I use prtvtoc on a machine with a SUN4.2G drive I can create a text file containing the partition table.

If I then juggle the figures to suit the geometry of a generic unlabelled SCSI drive, assuming that the slices are sized to start/end on a cylinder boundary. Can I then use fmthard to write the VTOC to the "new" SCSI drive & still have a working system at the end of it?

Do I need to label the generic drive using format first? And if so, am I better off using format to partition the drive as well?

If the new drive has come from Sun presumably it will be already labelled, do I actually need to write a VTOC at all?

( I'm working on a disaster recovery system & want to be able to restore the system to a new disk as close to the original configuration as possible)

Many thanks

Kevin

[931 byte] By [Kevin_N] at [2007-11-25 23:26:17]
# 1
I forgot to mention that I'm intending replacing the original drive with the "new" one then restoring the original system using ufsrestore, having used ufsdump to take an image of the original diskKevin
Kevin_N at 2007-7-5 18:12:54 > top of Java-index,General,Talk to the Sysop...
# 2
HiFmthard is used to write the VToc of a disk with vtoc already saved in a file passing as a parameter with fmthard ,i believe it can not be done for a new disk i'm not shure.Cheers,
machiunixGILLI at 2007-7-5 18:12:54 > top of Java-index,General,Talk to the Sysop...
# 3

The prearranged label that Sun sends will probably not be the same as yours. The file taht prtvtoc saves is just text, so presumably you could vi it and then use the fmthard command, but again, I have never done this either. Also, be very careful if you do decide to edit the file, it has to be exact for the type of drive you have (or leess, never more).

If this is going to be a DR image, then wouldnt the same size of disk do?

bmacdo at 2007-7-5 18:12:54 > top of Java-index,General,Talk to the Sysop...
# 4

Yes, the same size of disk would be ideal.

However in a network of mixed hardware we'd rather not have to have a stock of spare disks of more than one type. What we're aiming for is to just have a few that are at least as big as any individual drive on any machine. The if any machine "Get's hit by lightning!!" or any other disaster that totally destroys the drive we can ...

1) label the disk,

2) rewrite the VTOC (Or a suitable approximation to it)

3) restore an image of the operating system

4) restore any user files using ordinary backup/restore software

5) Carry on as if nothing had ever happened (Apart from the scorch marks on the case :-) )

We're hoping that using suitable Jumpstart scripts to do 1) 2) & 3) over the Network will be quicker than a re-install & re-patch of the OS from scratch

Any comments on whether what we're trying to do is possible / practical / sensible ?

Many thanks

Kevin

Kevin_N at 2007-7-5 18:12:54 > top of Java-index,General,Talk to the Sysop...