HP HP-UX 10.2 disk rescue

Hello,

I have a similar problem as in HP UNIX HFS 10.2 system corrupted - #25 :
have an HP workstation (Visualize C200) on HP-UX 10.20 that won't boot anymore (given the symptoms, I guess it's a power supply problem) and I'd like to try to recover the data from it.
Fortunately, I have another HP-UX (HP Visualize J5600) machine (but with HP-UX 11) that still works. I tried to plug the hard disk of the broken machine with external SCSI, it is well detected and added in /dev/dsk/c3t12d0, but I can't mount it to recover the data. The command :

mount /dev/dsk/c3t12d0 /mountpoint

respond :

/dev/dsk/c3t12d0 : unrecognized file system

It was the only disk on the workstation that failed (then it contains system and data) and I think (but I am not sure) that there were logical volumes on this drive.

Before doing anything that may cause data loss, I have done a disk copy with the above command (don't know if I need to give more args to obtain a better image):

dd if=/dev/dsk/c3t12d0 of=dumpfile.img

Is it possible to work with this dump instead the drive (to avoid any data loss) ?
How do I determine the partitioning and the filesystem of this drive and then mount it ?

Thanks for any help you can provide.

Welcome on board Dany,
Please open your own thread even if/specially if same subject as we expect sometimes the thread owner to respond... Generally speaking thread hacking is not permitted in this forum.

about your image, I have no idea if you can use your dumpfile instead, as when I used to use dd in such case I used same size discs (when possible, or greater knowing I loose the extra size...) and used the raw device so I had a true copy which even could be bootable...
How to determine the partitioning I have no idea, too far back for me, as since 10.20 all HP went LVM, so if you were not using LVM, truly, finding present days somebody still familiar with HP-UX 9 is going to be challenging... If you were in LVM you are better off, there are ways to recover, but of course the easiest would have been to have the lvm config files which would have let you use vgimport (1M) command, look at:
https://community.hpe.com/t5/lvm-and-vxvm/vgimport-disks-without-mapfile/td-p/2987959#.Xtr1vy1h2L8

4 Likes

The link you gave works perfectly to solve my problem (I was convinced that the mapfiles were indispensable for reconstructing the volume).

Thanks a lot for your help (and sorry for the thread hacking ).

1 Like

hahaha you were lucky to have LVM... the config files are almost indispensable if you had given fancy names to you logical volumes, compared to lvol1, lvol2 etc, as when is doesn't have it will restore in the order the system finds them ( generally following creation time) and call then lvol1, lvol2 etc I did that mistake once and never since did I use anything else but lvolXX and customised the group name to its minimum like r5vg02 meaning group02 in Raid5 subsystem

Be careful with your D230, very reliable boxes de D series for the time, I have a few in my garage I kept for spare for CERN lab, they have one spare you can't find: the fan in the power supply box... So when you power off try to give it a little clean there don't let dirt get in...

All the best

1 Like

This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.