Having read through and executed the Badblock HOWTO, I am at a standstill on how to proceed.

My setup is a Dell Studio running Ubuntu 8.10, with 2 640GB SATA drives in an LVM2 RAID1.  I am running VMWare 2 with a few Windows VMs and a few Fedora/Ubuntu VMs.  My issue is with the WinXP VM which has a 16GB .vmdk on the VMWare-emulated LSILogic SCSI controller.  All was well until apparently my batteries in my UPS died unbeknownst to me.  With a few storms and other power interruptions, I now have some bad sectors on both disks.  Also, there was a Samsung firmware update for these drives from Dell to fix an issue with these intermittently detaching from the SATA bus.  I've run all the way through the HOWTO and find that /dev/sda and /dev/sdb have different blocks which smartctl lists as bad, but the 1st bad block on each although at a different offset are both underlying my XP.vmdk file, so I don't want to zero out these blocks as that means I lose the VM, right?  At this point, I'm looking for some guidance of the following possible choices:

(1) boot with Ubuntu 9.10 Live DVD and try to `tar zcvf` the XP VM tree onto a USB HD - FAILS with Input/output errors and read errors
(2) is there a way to mount the VM and extract any files that I want to keep?  Acronis, Ghost, VMWare converter?
(3) should I take out a drive and try to repair the other?
(4) can I break the RAID and zero out the bad blocks on /dev/sda and dd the data from the same offset block on /dev/sdb, and vice versa?
(5) LVM snapshot?
(6) boot the VM with the Ubuntu Live CD and copy off the files to keep?

The VM boots but eventually (minutes) will lockup, cause the LVM to fault, and lockup the host OS since the LVM is then unreadable.

Once I can either recover the files I want to keep -or- have thrown in the towel on doing so, how to recover the bad blocks and mark for re-use or to forever skip?  Am I destined for a complete rebuild?  Ideally, I think I'm headed for a 3Ware controller with battery back-up for write-through, and continue to use LVM but put the VMs' *.vmdk files on an XFS or ReiserFS partition separate from the root of the host OS, if for no other reason so that a similar issue doesn't hose the whole box, but hope for better performance too.

Oh, yes, of course I've replaced the UPS batteries and confirmed that the UPS can carry for long enough to get it all shutdown and unmounted cleanly before the batteries go flat.

Thanks very much for any valuable insight!!!  I've spent about a month googling and RTFM'ing but seem to keep going in circles...

Rob Fegley