Public bug reported:
I'm using on several servers (5 Servers, 3 have Problems, all with
Ubuntu Dapper) software raids (2-4 S-Ata drives with about 300-500 GB)
and an LVM over it (LVM volumes go from 10GB to 1,5TB). On each LVM
volume I'm using Ext3. The Problem I have is, that fsck reports too many
errors. Typical errors are Inode XXX, i_size is 938275410942854278,
should be 0 or Inode XXX has compression flag set but is not a
directory. After 5 days running fsck I stopped it the hard way and tuned
the file system to bring the system up again. I moved all data to other
systems, created a new ext3, checked it and found no errors, moved the
data back and there was the problem again. I can not reproduce the error
on command and can not find out if it is the software raid, lvm or Ext3.
The only thing I can say, that is not a hardware problem (same hardware,
but only one server has the problem, other hardware same problem). The
only way I can keep the servers running at the moment is to tune the
Ext3 that it will never need a check and then take a Knoppix CD (Version
5.1.0) boot the system with it and make an fsck. In most cases it will
find some errors (nothing big), correct them and on a second try there
will be no errors (Dapper will find errors until the end of days).
** Affects: ubuntu
Importance: Undecided
Status: New
--
Dapper lvm over software raid kills ext3
https://bugs.launchpad.net/bugs/247318
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
--
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
