Hi,

I am new to this forum and hope someone can help. Does anyone know how to restore a Solaris 9 backup tape from server A to a completely different server B hardware. Both boxes are sun 64bit.

Your help is much appreciated.

Thank you

Jan.

boot from cd and use "ufsrestore"

Restoring Server A over server B is not going to work. Lets assume server B has the same OS as server A. With this in mind. What is safe to restore using ufsrestore and what should I omit from the restore selection?

Thank you.

there are so many posibilities.... what kind of backup to you have? what is the disk layout? which filesystem/s do you use? what hardware do you use? what was the type of the installation?

I have a tape backup using ufsdump and I know the order of the dump.

The backup is of a sun v440 with solaris 9 and configured for a SAN

I want to create a test environment on a sun e420 with local disks. I have installed solaris 9 on the e420 and created the disk/slice layout the same as the v440. the e420 will not be connected to SAN.

Filesystem kbytes used avail capacity Mounted on
/dev/md/dsk/d0 8160366 6358828 1719935 79% /
/proc 0 0 0 0% /proc
mnttab 0 0 0 0% /etc/mnttab
fd 0 0 0 0% /dev/fd
/dev/md/dsk/d6 6192138 815529 5314688 14% /var
swap 14817784 96 14817688 1% /var/run
/dev/dsk/emcpower0a 20649993 11612265 8831229 57% /l
/dev/dsk/emcpower0e 20649993 13510511 6932983 67% /oa
/dev/md/dsk/d12 4127845 688996 3397571 17% /home
swap 14848472 30784 14817688 1% /tmp
/dev/md/dsk/d9 6192138 268382 5861835 5% /spare1
/dev/md/dsk/d24 8154588 2429 8070614 1% /spare4
/dev/md/dsk/d15 10323610 9 10220365 1% /spare5
/dev/md/dsk/d18 8258597 6102185 2073827 75% /spare2
/dev/md/dsk/d21 8257152 2455957 5718624 31% /testoa
/dev/dsk/emcpower2b 82595927 31018913 50751055 38% /onbackup
/dev/dsk/emcpower2a 20649993 7661554 12781940 38% /l/trng
/dev/dsk/emcpower1b 7220997 1947893 5200895 28% /l/bi
/dev/dsk/emcpower0d 20649993 11459971 8983523 57% /l/demo
/dev/dsk/emcpower1a 8253532 1255023 6915974 16% /l/ai
/dev/dsk/emcpower0f 20645897 12072108 8367331 60% /l/var
/dev/dsk/emcpower0b 20649993 8920722 11522772 44% /l/test
#

I am at the stage where I can start restoring. What do I restore that is not server specific. I want to transfer the applications and system settings but not the hardware drivers. After the restore I want the e420 to boot.

Hope you can help.

Kin regards,

I found a document that looks to be a set of instructions for restoring a sun backup on to different hardware. It reads like it might work. Can someone take a look and fill in any missing or unclear gaps. I only have basic unix skills. I can install/admin etc but thats it.

Who wants to be first to publish a guide using this as the frame work?

thank you.

==================start======================

Installing the operating system


Turn system on and from the “ok>” prompt run: -

Ok> probe-scsi-all

This will check that the system can see all the attached (physical) SCSI devices.

If all ok then insert the Installation CD and run :-

Ok> boot cdrom

This will start the installation and will prompt you warning messages (wrong magic numbers if disks not labelled)

Set up

Local UK-English
Hostname NDR
Networked No (YES for Legato/Netbackup )
NIS None
Subnet no
Timezone Europe/GB
Client Services no
Installation Entire Distribution including OEM support







**Point to remember:-

Any disks, which were previously used in a non sun system, will mean that the disk labels have been overwritten. The installation routine may exit due to this!

From the console window run “format” - to label each disk

Once each disk has been checked for a Solaris Label then the installation can be resumed by running “suninstall” from the same console window.

When prompted select “Auto Layout” and then “Customise” (to create file system layout)

Now set up the system as per their “df -k” ( Note this must be kept up to date)

Mount Remote Filesystems no
Begin Installation
Auto reboot
Set Root password None (This will be recovered from backup)
Power Saving no*

*You must respond to “no” for Power Saving otherwise the system will shutdown when there is little or no activity.




Restoring the customers backup



It is better to bring back the O/S first ie /, /usr, /var, /opt, /export/home then reboot single user mode to bring the rest of the user filesystems. This provides confidence that the system is ok using the backup of the O/S.

Login as root and bring system into single user mode.

* The following will alter depending on the type of backup that has been performed

Tape device as /dev/rmt/0

Ufsdump

These are created on a “per” filesystem (or per slice basis). Multiple usfdumps are written to the same tape and can be accessed sequentially using the “n” – non rewind switch.

You will now need to look at the Customers backup script to see what order they were created.
You may need to skip using “mt –f /dev/rmt/0n fsf X” (X=No backups to skip) This will position the tape correctly for the filesystem you wish to recover.

There are excludes needed for root (/) and the (/usr) filesystems. All others can be restore non-interactively.
















Root filesystem restore:-

boot cdrom –s

Mount root filesystem on temporary mount point /mnt ie :-

# mount /dev/dsk/c0t0d0s0 /mnt

# cd /mnt

# ufsrestore –ivf /dev/rmt/0n

 ls
 add
 delete /cdrom /dev /devices /kernel /platform /proc
 cd /etc
 delete devlink.tab device.tab mnttab name_to_major path_to_inst
 delete driver_aliases driver_classes vfstab wtmpx utmpx

If the /usr directory is part of the / directory do :-
 cd /usr
 delete kernel
 extract
 volume # : 1
 set owner / mode? no
 set modes on directories? No
 quit

If the /usr is separate filesystem do the following :-
For the /usr filesystem:-

Mount usr filesystem on temporary mount point /mnt ie :-

# mount /dev/dsk/c0t0d0s6 /mnt

# cd /mnt

# ufsrestore –ivf /dev/rmt/0n
 ls
 add
 delete kernel
 cd lib / ufs
 delete ufsrestore
 extract
 volume # : 1
 set owner / mode? no
 set modes on directories? No
 quit
Restore the /var and /opt filesystems these can be fully overwritten :-

For the /var filesystem:-

Mount usr filesystem on temporary mount point /mnt ie :-

# mount /dev/dsk/c0t0d0s1 /mnt

# cd /mnt

Make certain the tape is at the right partition

# ufsrestore –rf /dev/rmt/0n


A standard restore can be performed for the remaining filesystems.








CPIO recovery

Command:

# cpio –icvdum -f “*****” “*****2” “*****3” </dev/rmt/0

Create a file containing this it makes it easy to check/correct typos before execution!!

The patterns used when excluding files are the same as those used when performing a selective restore. e.g. “etc/utmp” or “./etc/utmp” (if relative) and “/dev/*” to exclude whole directories

TAR recovery

Command:

# tar –xvfX /dev/rmt/0 /exclude.list

When creating the exclude list you will need to set it up as “one file per line” and you don’t need to include “*” when specifying directories*

* don’t forget to use “./” if needed!

e.g.

/cdrom
/dev
/devices
/etc/utmpx
.
.
/usr/sbin/tar

Post restore

If Solstice Disk Suite (SDS) or Veritas Filesystems and Volume manager (VX) are being used then you will need to edit /etc/system and comment out any reference to both.

Disk Suite:-

SDS looks like /dev/md/….

Disable Veritas Volume manager: -

VX looks like /dev/vx….


*rootdev:/psuedo/vx….
*set vx…..
cd /etc/vx/reconfig.d/state.d
rm root-done
touch install-db

**Don’t forget to change the password file as this has now been restored from tape as customers.





SUN EXCLUDES LIST:



/cdrom
/dev
/devices
/kernel
/platform
/proc
/etc/driver_aliases
/etc/driver_classes
/etc/devlink.tab
/etc/device.tab
/etc/mnttab
/etc/name_to_major
/etc/path_to_inst
/etc/vfstab
/etc/utmpx
/etc/wtmpx
/usr/kernel
/var/run
/etc/sysevent
/usr/lib/fs/ufs/ufsrestore (If this is used for recovery)
/usr/sbin/tar “
/usr/bin/cpio “

=================end=================