SUMMARY: booting after recovery
- From: "spiro harvey" <spiro.harvey@xxxxxxxxx>
- Date: Sat, 5 May 2007 07:13:13 +1200
Thanks for all the rapid responses.
The solution was in fact that I was missing updating the /etc/path_to_inst file.
This was done by booting the working system, mounting the recovered
filesystem into /mnt/root and then running 'devfsadm -r /mnt/root'
devfsadm has the ability to update device trees and path_to_inst in an
alternate root path using the -r option.
However, I did notice that by doing this, path_to_inst was updated on
the *current* root partition, so I had to copy that over afterwards.
Another point to note is that devfsadm will only update files that
don't currently exist, so I had to delete all the /dev/dsk and
/dev/rdsk entries before devfsadm ran.
On 04/05/07, spiro harvey wrote:
Hi all..._______________________________________________
I've got an interesting problem trying to boot a hard disk using a recovered OS.
I'm building a test network, so the hardware I'm building on isn't the
same as the original hardware.
The OS is Solaris 9.
The source server is a 280R.
The destination server is an E3000. I know. I know. But it's all I've
got to work with.
The problem I've got is that all the device paths are completely different.
c0t0d0s0 has the root partition installed from CD
c0t3d0s7 has the root partition recovered from tape.
I couldn't overwrite the original root partition as it's not big
enough, so I boot the server with 'boot disk3:h' which works fine
until it gets to fsck the drive.
Then I was told that fsck couldn't stat the device. After that I
copied everything from the working root's /dev/dsk, /dev/rdsk and
/devices (specifically SBUS dirs) into the recovered partition using
cpio.
However, this didn't fix the problem. It won't read the disk as
c0t3d0s7. It tells me it can't stat the drive, and to run fsck
manually. I am then dumped into a shell prompt. When I run fsck from
here, I'm told it can't open the device (previously was told couldn't
stat the device). So I think I've progressed, but it still won't boot
properly.
I tried doing a boot -r but the fsck problem showed before it
reconfigured, hence the need to manually copy the devs with cpio.
my basic steps in the recovery were:
- restore root data onto spare partition using scanner|uasm (Networker)
- edit vfstab to list new physical devices (original server was using
metadevices)
- edit /etc/system and comment out all the MDDB lines
- rename hostname.ce0 to hostname.hme0 (different NICs)
- install bootblock
- cpio -p /dev/dsk /dev/rdsk and /devices to new partition
- boot disk3:h (c0t3d0s7)
I've got a bit of a tight deadline to build this test environment up,
so any ideas greatly appreciated.
Thanks.
sunmanagers mailing list
sunmanagers@xxxxxxxxxxxxxxx
http://www.sunmanagers.org/mailman/listinfo/sunmanagers
- Prev by Date: ZFS and SAN snapshotting
- Next by Date: Solaris 10 swap size question
- Previous by thread: ZFS and SAN snapshotting
- Next by thread: Solaris 10 swap size question
- Index(es):
Relevant Pages
|
|
