lkml.org 
[lkml]   [2000]   [Jan]   [28]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: Corruption w/ 2.3.41-pre2
On Sun, Jan 23, 2000 at 09:18:06PM -0800, Andre Hedrick wrote:
>
> There appears to be an "EXTREME" high load bug that is hard to catch.
> Only journaling FS or HA and Performance test ever trigger this one.
> If you can tell me how you created the situation to trigger this bug,
> I will have a better idea how to catch and fix it.

I have a Tyan Thunderbolt, dual P3/450 (not overclocked), AIC-7896 SCSI.
I have been experiencing the same problem, with 2.3.31, 2.3.39, and 2.3.40.
The filesystems are ext2 with 4k inodes.

I first noticed it when trying to build Mozilla -- it untars fine but some
files produce "Input/output errors" when you try to access them (ls, cat,
rm, find, etc.). I moved the tree to another location on the same
filesystem and tried again, and the same files were corrupted in the new
tree (maybe not a perfect correlation, but pretty close). If you unmount
the filesystem and do an "e2fsck -f -y /dev/xxx", it clears up the problem
(IIRC, it deletes the affected files, but I could be wrong).

I have done a few kernel builds and other large-scale compiles without
having this pop up -- it seems to be pretty intermittent and random,
most often happens with large trees, but some small trees affected.
Not limited to a single filesystem, either. CPU, disk, network loads
are all minimal -- this is a single-user X workstation.

jafager

--

"On desperate ground, fight."
-- Sun Tzu, _The Art of War_
[unhandled content-type:application/pgp-signature]
\
 
 \ /
  Last update: 2005-03-22 13:56    [W:0.124 / U:0.332 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site