Ing. Milan Kocian wrote:
> Hello,
>
> I played with xfs on crypto-loop device (aes-loop on sf.net) on sw raid5
> (log v2 on raid5 device). I think that I found a bug. I created loop on
> raid5, formatted it with xfs, mounted it and all was ok. Then I
> unmounted loop and with next mount I saw this:
>
> XFS: log inconsistent or not a log (last==0, first!=1)
> XFS: empty log check failed
>
> I had to repair loop with xfs_repair. And after next umount the same
> bug.
>
> I tried log v2 and then v1. It had no effect. On normaly partition (with
> loop) I had no problem. Only on raid5. I didn't test raid1. Original
> loop device had the same problem.
>
> Checked with xfs 2.4.20-2003-03-19_04:55_UTC
> and linux-2.4.21-xfs-2003-05-23-cvs.patch.bz2 and with 2.4.21_xfs.
>
> Thanks for answer.
>
> Best regards Milan Kocian
>
>
>
>
I am currently running a loop-aes encrypted filesystem over XFS on raid1
with no problems. I also tried running it over raid5, but suffered fs
corruption upon umount and was unable to remount it afterward. I assumed
that the way that data is laid out over a raid5 volume is incompatible
with a loop-aes encrypted filesystem. I remember reading some caveats
with journaling filesystems indicating that writes have to be guaranteed
to be written in order in order for them to work. Perhaps this isn't the
case when using raid5?
-Walt
|