SUMMARY: Big problems restoring a dump to another machine

From: Jeff Kennedy (
Date: Wed Jan 13 1999 - 12:31:35 CST

I finally got it up and running with the following procedure (well,

"Even though a sparc 5 and a sparc 10 are sun4m architecture they have
different hardware
device drivers for their scsi chains. What I have done in the past when
presented with
this conversion is the following -

     1. boot cdrom -s (make sure you the same OS release)
     2. mount the / partition on /mnt
     3. cd /mnt/dev/dsk; rm *
     4. cd /mnt/dev/rdsk; rm *
     5. (cd /dev/dsk; tar cf - *) | (cd /mnt/dev/dsk; tar xfBp -)
     6. (cd /dev/rdsk; tar cf - *) | (cd /mnt/dev/rdsk; tar xfBp -)
     7. cd /mnt/etc; rm path_to_inst
     8. boot -ar and take the default answers to the questions except when
        asks you whether or not create a new /etc/path_to_inst, answer yes.

This should get your system up and running."

I removed /a/dev/* instead of /a/dev/dsk/* and this might have been the
cause of my following problem. When running the tar command it would fail
at the fd directory every time, for both dev and devices. So I tar'd
everything but the fd dir. It took 2 boots to get everything copacetic but
it's up now.

I had several suggestions regarding the bootblk but that was not the issue
as I had done that already. Someone also suggested roughly the above
procedure only using the ufsdump command instead of tar, this did not work.
The dump aborted every time because /dev was not a "locally mounted
filesystem". Even though /tmp is supposed to be a local filesystem it
wouldn't pull it off. It was also mentioned that moving from several
slices to one slice was invalid but it seems to be running fine so far,
although I haven't put it through it's paces yet. There might have been
some confusion as to what was restored and where; I did not try to put all
the files in each filesystem into / but actually had the usr, var, and opt
dir's in / and restored the corresponding filesystem to what was it's
original mount point.

Thanks to:
spillar (who's procedure it is)
Danny Johnson
Mohd Abdul
Richard Smith
Aki Sasaki
Yvonne Herman
Scott Howard
David Staggs
---------------------- Forwarded by Jeff Kennedy/NDS on 01/13/99 10:12 AM

"Jeff Kennedy" <> on 01/08/99 05:50:33 PM
 cc: (bcc: Jeff Kennedy/NDS)
 Subject: Big problems restoring a dump to another machine

Hello All,

I am trying to restore a full dump of a firewall machine to a different
machine for upgrade testing purposes.

The OS is Solaris 2.5.1 originally on a SS5. I am trying to restore it to
a SS10. The disk partitions on the SS5 were old style (/, /usr, /var,
/opt, and /home all seperate) and the new machine has been restored to just
/ (and swap of course). I have changed the vfstab on the restored machine
to reflect the current disk.
My procedure was this:
Full dump of SS5
Format the SS10 as one / partition and swap
newfs the disk
restore the dump files of all filesystems (each in their respective
change /etc/vfstab

Here's the problem:

I can't get the system to recognize the new boot disk. I have done;
boot -r
boot cdrom -s and mount the disk and try to drvconfig--disks--devlinks

Nothing seems to help. When it boots it will tell me it can't stat the
disk, run fsck manually, and go into single user mode. Once in single user
mode I do a df -k and it shows the original (SS5) partitions mounted and
all at the same utilization and all of the same size. Any ideas what I'm
doing wrong?

The new disk is c0t0d0s0, the original disk was c0t3d0sX and c1t0d0sX.
These latter partitions are what's showing mounted.


Jeff Kennedy

This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:13:13 CDT