Disk Failure

[L. Younes]

Hi everyone, I have a disk failure on a lv of 3 disks (stripe)(HP_UX 11.11, online jfs);I do not have tape backup, the data on the disk is not at all critical. But I would like to know, if there is a way to backup as much data as I can, then recreate the lv with 2 disks (without the failed one) then restore the data. I know that striping doesn't make it easy cause already part of the data is on the 3rd failed disk due to the stripe mechanism.
Another question, is if I do a dd on a disk, destroy the lv it belongs to, put it in another lv and restore the data with dd, would it destroy some critical information put on the disk by the lvm during it's migration ?

Thanks for your responses.

[Stefan Farrelly · Accepted solution]

Seeing as your lvol was striped its likely most files exist on >1 disk which means you have lost everything. I dont see any point in trying to backup what you can - chances are a lot of work for almost no return. You may only be able so save some small files (smaller than the stripe size).

If you dd an lvol to another lvol it doesnt destroy any critical info - if both lvols are in different vg's its no problem, your data will dd over in a perfectly usable condition. We do it all the time.

[L. Younes]

Thanks,
I'll just do everything all over again.
Concerning the second response, is it necessary that the lv's be in different vg's, what happens if not. and does your explanation apply to moving physical disks
from one vg to another ?

[Stefan Farrelly]

No, its not necessary the lv's are in different vg's. Its also fine to move physical disks from one vg to another or to a different server. The only thing to watch out for when using dd is that it completes without any errors. If you get an I/O error you have to redo it.