[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: e2fsck running for hours, printing out lists of numbers -- should I stop it?



Hi,

On Tue, 2005-04-05 at 17:13, Allen Ziegenfus wrote:

> However, at one point I forgot to pick the correct kernel at boot time
> and I ran the standard woody kernel instead which has this ext3 driver:
> 
> Linux version 2.4.18-bf2.4 (root zombie) (gcc version 2.95.4 20011002
> (Debian prerelease)) #1 Son Apr 14 09:53:28 CEST 2002
> EXT3 FS 2.4-0.9.17, 10 Jan 2002 on ide3(34,65), internal journal

Well, that _is_ old --- there have been a fair number of bugfixes over
the past 3 years, but nothing in ext3 itself that you'd expect to cause
instant data corruption just because you ran it once instead of a later
kernel.  And I certainly don't know of any incompatibility issues save
some involving features only present on later kernels, such as extended
attributes --- and for the most part ext3 handles that transparently,
anyway.

> The night after I booted with
> the older kernel, the machine had the following  in its logs. It must
> have been when trying to delete some older archives.
> 
> Mar 30 02:37:11 musicien kernel: invalid operand: 0000
> Mar 30 02:37:11 musicien kernel: CPU:    0
> Mar 30 02:37:11 musicien kernel: EIP:    0010:[journal_forget+170/400]  

Anything else in the logs?  You just hit a BUG(), and a bug or assert
failure message should have been emitted just prior to this.  Such an
error in journal_forget() sounds like one of the situations I fixed a
couple of years ago, where certain on-disk corruption could cause ext3
to oops internally rather than fail gracefully.  But that's not a
*cause* of the problem, rather just a less-than-ideal way of responding
to it.

Also, with a kernel that old, data corruption problems could be due to
something as basic as the old IDE driver not dealing properly with new
hardware in your system.

> So are the ext3 drivers not backward compatible? 

They should be fully compatible all the way back to the 2.2 kernel (2.0
if you force version 1 superblocks on disk.)

--Stephen



[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]