Hello all, I have a bit of an odd question. A friend of mine recently decided to show off his nice shiny 1TB usb hard drive (he's a git). While it looks nice I had some odd errors when mounting it.

The entire drive is supposedly formatted in xfs, when plugged in, udev does something odd. It creates 7 devices. /dev/sda and /dev/sda{1..6}.

Hmm. The only one that is readable is /dev/sda6. Which is a 1TB (give or take) xfs partition, containing all of his data.

I used cfdisk to look at all of the sda devices, no partition table in sight. Literally all that had been done was

When the disk was brand new. I'm not sure what the ramifications of this might be. For now Debian seems quite happy force mounting /dev/sda6. There are a few error messages when it is plugged in that I could dig up if anyone is interested. Anyone have any idea how to rectify the situation without destroying data? I was thinking about creating a partition table on /dev/sda, but this would most likely make the existing xfs filesystem (I was going to say partition, but without a patition table what is it?) invisible.

Cheers

EDIT: Oops wrong forum ... never mind

If your friend can mount the drive, I would advise leaving well enough alone. You don't want to screw up your friends data. My guess is that either he created an extended partition and the others simply don't exist, or that there is a problem with the udev rules creating devices that don't exist.

He might want to run fsck.xfs or xfs_check (both have manpages) on the partition to fix any errors there. There is an xfs_info command which may be illustrative. As well as an xfs_info manpage, there are also manpages for mount, xfs, mkfs.xfs, xfs_db and xfs_admin.

I was looking for a force or error=resume mount option but didn't find one. There may be more mount options listed in your kernel source documentation (xfs.txt).

Hello I apologise for not replying for some time, the disk is not mine and I don't have easy access to it. Anyway ...

All of the xfs tools I tried failed with strange "read-ahead" errors. Don't know exactly what that means but I have a good idea (see testdisk results).
So I found testdisk (fantastic program btw) and ran it on the drive.

It found a 1TB xfs partition that was in a bizarre position, it started roughly in the middle of the drive and ended beyond the physical limits of the drive itself (by a good 600 Gb).

God knows how this magical fairy partition came into existence. In any case all is now well, we managed to save all of his data and recreate a proper partition table. What surprises me is that the file system didn't fail earlier.

Thank you for your help jschiwal,

and hopefully this thread should serve as a warning for the future...

Ps. does anyone know what actually would have happened if the disk had tried to write at the back end of the phantom partition?

my housemates HFS+ formatted drive looks like taht to mee too, (not taht i cna mount it as i dont have HFS+ kernel support)