EXT3 - Detected aborted journal

[Gabor Heja <kakaopor@kakaopor.hu>]

Hi,

I have an EXT3 filesystem where I periodically get messages like these:
Apr 28 14:24:06 zaphod kernel: journal_bmap: journal block not found at
offset 30732 on dm-0
Apr 28 14:24:06 zaphod kernel: Aborting journal on device dm-0.
Apr 28 14:24:07 zaphod kernel: ext3_abort called.
Apr 28 14:24:07 zaphod kernel: EXT3-fs error (device dm-0):
ext3_journal_start_sb: Detected aborted journal
Apr 28 14:24:07 zaphod kernel: Remounting filesystem read-only

When I fsck that filesystem, it finds some errors (deleted inodes, bitmap
differences), repairs them, but I am still unable to remount it in
read-write mode. In the kern.log I see the following:
Apr 30 00:22:11 zaphod kernel: ext3_abort called.
Apr 30 00:22:11 zaphod kernel: EXT3-fs error (device dm-0): ext3_remount:
Abort forced by user
Apr 30 00:22:11 zaphod kernel: ext3_abort called.
Apr 30 00:22:11 zaphod kernel: EXT3-fs error (device dm-0): ext3_remount:
Abort forced by user

If I reboot the system it gets mount correctly, and after some time the
error appears again.

I have destroyed and recreated the filesystem several times (two or three),
but the error is still there. I have updated my kernel too (was 2.6.28.3,
now it is 2.6.29.1).

This filesystem is actually on a crypto device over a RAID-1 mdadm array.

root@zaphod:~# uname -a
Linux zaphod.kakaopor.hu 2.6.29.1 #1 SMP Sun Apr 5 04:36:09 CEST 2009
x86_64 GNU/Linux

I would appreciate any help you can provide.

Thank you!

Best regards,
 Gabor