Restoring the Root Backup

[Hemalatha_1]

Hi,

Following a suspected disk failure (primary), we are going to replace the disk and load the OS all over again. (HP UX 11.00). The earlier version had several patches installed for SYBASE ASE 12.5, plus there were several OS parameters that were customised. Printer installations too were present. Going through all these processes once again is going to be time taking. I have a clean backup of the entire root file system. I intend to simply restore the root back up after the initial installation. Will it work that way?

Thanks

[Pete Randall]

How was the backup produced? An Ignite backup would be ideal in this case. Any other type (tar, fbackup, dd, cpio) would probably be quite troublesome.


Pete

[Dennis Handly]

Your root backup should have all of your OS patches, kernel and printer configurations.
(Assuming you backed up /stand and /var too.)

I don't know about your SYBASE patches, were these OS patches from HP? If not, they may not be backed up.

[Dennis Handly]

>ME: If not, they may not be backed up.

Of course, if these were on /opt, a disk that didn't go bad, you should be fine.

[Hemalatha_1]

Hi

I took the FS backup of root by unmounting all the other filesystems except the ones created during installation namely, /, /stand,/opt,/var, /usr, /tmp. These portions are apparently healthy. I used a tar to create the backup.

Hemalatha

[Hemalatha_1]

yes, the sybase patches were downloaded from the HP site. I am not sure about their location. They should be in one of the root directories, presumably.

[Hemalatha_1]

No Ignite backup is available

[Deepak Kr]

Hema,

Yes, you can try to restore these FS using tar restore in single user mode and see if you find everything working...

But, to avoid same situation; it is advisable to take ignite backup once you fix all these issue if any.

[OldSchool]

Is MirrorDisk-UX installed and configured? If the root volume group is mirrored, you shouldn't need to install anything.

[Hemalatha_1]

No, there is no mirroring, whatsoever.

[Hemalatha_1]

Apart from the patches and OS customisation, I discovered there are about 25 users on this system. If the FS restoration of the root through tar is performed, then will the user accounts, passwords and previliges also be restored? We have a second disk too, housing the applications. Will this disk be recognised too after the restoration without a vgcreate?

[OldSchool]

well, you didn't say you backed-up /home, so if that where all the users home dirs are, they'll be gone.

on untrusted systems, the account info is in /etc/passwd which is part of /, so that should be ok. printer stuff is in /var and /opt (and maybe /etc...i don't recall) so that'll probably be ok as well.

If you only suspect that a disk is failed, I'd look really closely at confirming before you proceed.

For example: What messages / errors make you "suspect" its failed?

[Hemalatha_1]

Hi,
Yes /home was backed. Disk failure is suspected because it is not able to write to the swap device. While start up the swap device is reprting I/O errors. Despite all our attemepts the swap device could not be enabled. That is what makes us suspect that the disk is amiss. We have now configured a secondary file system swap, which has marginally improved the performance. Further, we also have a backup of the root FS prior to the error being reported.

Hemalatha

[Prashanth Waugh]

Hi Hema,

Check the disk using eg./dev/rdk/c0t0d0
#diskinfo /dev/rdsk/c0t0d0

U use ur disk path .

Regards
Prashant

[Khairy]

hi hema,

i refer from your recent post:

"While start up the swap device is reprting I/O errors. Despite all our attemepts the swap device could not be enabled. That is what makes us suspect that the disk is amiss. We have now configured a secondary file system swap, which has marginally improved the performance. Further, we also have a backup of the root FS prior to the error being reporte"

Correct me if i'm wrong, after u configured 2nd swapspace, u able to bring up the system.

If you do, perform make_tape_recovery as recovery from this would be clean and straight forward.

# make_tape_recovery -Av

Rgds

[Hemalatha_1]

The system was coming up in either case due to the Psudo Swap. The startup takes 8 min, stuck for 5 min while trying to write to swap and finally ignores the swap and proceeds with the rest of the boot process to bring up the system. The secondary swap is just about 200 MB , the last available space on the disk.

I will do make_tape_recovery -Av right away, that is a good idea.

But how do I restore later?

Thanks
Hemalatha

[Khairy]

hi hema,

to restore:

1) reboot your system
# shutdown -ry 0

2) interrupt the boot loading, u can press any key to interrup the boot process.
You will see bch prompt
bch>

3) search for bootable devices
bch > sea ipl

..and look for sequantial device. This should be your tape drive

4) boot the sequential device. Assuming its the 1st entry P1
bch > boot p1

Choose no to interact with ISL. The restoration should start now.

Rgds

[Hemalatha_1]

make_tape_recovery -Av . Is this the command? There is no manual entry for this. I checked. How does one run this?


Hemalatha

[Dennis Handly]

>There is no manual entry for this. I checked.

Check the Ignite-UX Administration Guide:
http://docs.hp.com/en/5992-3336/index.html

This says the man pages are in:
/opt/ignite/share/doc/

This does NOT match the proper SVR4 layout, so I'm not sure how you would have to set MANPATH or use man -M.

More manuals:
Ignite-UX Quick Start Guide: Create an Ignite-UX Server and Cold-Install Clients
http://docs.hp.com/en/5992-4815/index.html

[Hemalatha_1]

That means, running make_tape_recovery requires the Ignite Server to be installed. I don't have it on my server. Can Ignite be installed on HP UX B 11.00 ? I also have a very old system running HPUX B 10.20. Is there a version restiction for Ignite Server?

Hemalatha

[Deepak Kr]

Hema,

You please check following link for ignite-ux for hp box you have.

http://software.hp.com

Also I am not very sure if ignite-ux product is still available for version 10.X u have.

Ignite-ux is a free downloadable from hp site as mentioned above.

You check for existing software in your boxes using following command:

#swlist |grep -i ignite

[Pete Randall]

Try here:

http://www.docs.hp.com/en/IUX/


Pete

[OldSchool]

I don't believe you can download Ignite for 10.x or 11.0 anymore, as both are out of support (10.x for years).

So, you've an issue with the swap partition, but the machine boots...so the question is how are your disks laid out today? what's in vg00 and what does ioscan report.

Maybe its possible to fix it without reloading the O/S.

What I would be concerned with is that you could take a running system and absolutely trash it if things go sideways....

[Hemalatha_1]

>I don't believe you can download Ignite for 10.x or 11.0 anymore, as both are out of support (10.x for years).

Checked the Web site http://docs.hp.com/en/5992-1977/5992-1977.pdf. Available only from HPUX 11i v1 only.

> So, you've an issue with the swap partition, but the machine boots...so the question is how are your disks laid out today? what's in vg00 and what does ioscan report.

The machine boots but takes a long time. Gets stuck while reading the swap configuration, output is reproduced from syslog.log

Sep 3 18:15:07 hpaclass vmunix: Swap device table: (start & size given in
512-byte blocks)
Sep 3 18:15:07 hpaclass vmunix: entry 0 - major is 64, minor is 0x2; DI
AGNOSTIC SYSTEM WARNING:
Sep 3 18:15:07 hpaclass vmunix: The diagnostic logging facility has started
receiving excessive
Sep 3 18:15:07 hpaclass vmunix: errors from the I/O subsystem. I/O error en
tries will be lost
Sep 3 18:15:07 hpaclass vmunix: until the cause of the excessive I/O logging
is corrected.
Sep 3 18:15:07 hpaclass vmunix: If the diaglogd daemon is not active, use th
e Daemon Startup command
Sep 3 18:15:07 hpaclass vmunix: in stm to start it.
Sep 3 18:15:07 hpaclass vmunix: If the diaglogd daemon is active, use the lo
gtool utility in stm
Sep 3 18:15:07 hpaclass vmunix: to determine which I/O subsystem is logging
excessive errors.
Sep 3 18:15:07 hpaclass vmunix: ignored - can't write to device
Sep 3 18:15:07 hpaclass vmunix: ERROR: dump device 64:0x2 cannot be used and w
ill be ignored:
Sep 3 18:15:07 hpaclass vmunix: device is read-only
Sep 3 18:15:07 hpaclass vmunix: WARNING: no swap device configured, so dump can
not be defaulted to primary swap.
Sep 3 18:15:07 hpaclass vmunix: WARNING: No dump devices are configured. Dump
is disabled.
Sep 3 18:15:07 hpaclass vmunix: Warning: file system time later than time-of-da
y register
Sep 3 18:15:07 hpaclass vmunix:
Sep 3 18:15:07 hpaclass vmunix: Getting time from file system
Sep 3 18:15:07 hpaclass vmunix: Starting the STREAMS daemons-phase 1
and then the boot proceeds

The disks are apparently fine. vgdisplay -v vg00 & ioscan reports are OK.

>Maybe its possible to fix it without reloading the O/S.

That is what I am looking for. Would appreciate if I can get help in this regard.

>What I would be concerned with is that you could take a running system and absolutely trash it if things go sideways....

ABsolutely. We never would like to tamper a running system. Nevertheless, we would like this issue to be sorted out

Hemalatha

[OldSchool]

it almost looks like the swap space has been deleted. what does

vgdisplay -v /dev/vg00/lvol2

report?