[CLUE-Tech] ext2 file corruption

[Jed S. Baer]

On Thu, 25 Dec 2003 09:02:02 -0700 (MST)
ockers at ockers.net (Jim Ockers) wrote:

> Hi Jed,
> 
> > I know this sounds whacko. But I'm 99% certain I'm getting some
> > file-corrupting behavior on an ext2 filesystem. Regrettably, I have
> > really nothing much to go on here, as the last fsck on the filesystem
> > passed fine-- but I'll do another shortly.
> 
> It's not the ext2 filesystem.  Tell us about your block device.  If
> it's IDE, what is the motherboard chipset?  What kind of IDE cable
> are you using?  What kernel version, and does it properly support
> your motherboard's chipset?  What kind of hard drive?  How old is
> all the hardware?

It's a Biostar M7VIG-Pro mobo, AthlonXP cpu, Via chipset "VT8375
(KM266)/VT8235", Winbond W83697HF. Purchased 7-Sep-2003.

Drive is a Seagate ST330630A "Barracuda", purchased August 2000.

IDE cable is the std flat IDE cable, probably came with the drive or mobo
purchased at the same time.

Kernel is 2.4.18-14, custom compiled with proper chipset drivers included.
As to "properly" supporting the chipsets, IANAC (I am not Alan Cox ;-),
but I don't recall hearing of any issues with the VIA chipset except for
maybe the earliest kernels in which it appeared. It's a RedHat kernel,
meaning it includes whatever patches RH deemed appropriate.

I've run the smartmontools with results that are ambiguous WRT specifics,
but indicate that the drive itself figures it's OK.

On Thu, 25 Dec 2003 09:08:29 -0700
"Joe 'Zonker' Brockmeier" <jzb at dissociatedpress.net> wrote:

> Have you checked other files to see if this has happened with anything
> else? I would guess that if you're experiencing some kind of file
> corruption that it would affect more than one file. OTOH, if your disk
> has developed a bad block, it might affect only one file.
> 
> Also, did you check to see if the other file still exists? 
> 
> You also didn't mention what kind of files these were or what
> application you might be using them with -- perhaps it's not a
> filesystem issue, but an application-level issue? 

Well, the files are both plain text files. One would have originally been
saved from Pan (usenet article), and the other from Sylpheed (e-mail
message). Both have since been edited in vim -- in particular, the newer
file was intact last time I looked at it. AFAICT, the "bottom" file has
disappeared (i.e. it exists only as the content in the concatenated
"corrupt" file).

I have, on occasion, observed intermittent errors with JPEG files.
Specifically, "premature EOF when decoding JPEG file", which, for the same
file, is sometimes reported, and sometimes not. But, having no other
evidence of problems, I had dismissed this as an issue with the JPEG
library.

It could be the result of a bad block. Presumably, smartsuite will have
show some indication of that. On the same physical drive as the file I
noticed, is the kernel, and /usr, and I have no evidence of either kernel
or application instability.

> > I had thought that ext2 was rock solid.
> 
> AFAIK, it is -- but there might be some kind of special condition which
> causes problems. 

Yeah, visions of the 2.4.19 kernel (ext3 corruption, but only with a
specific chipset, IIRC).

> It might help to know what kernel version you're using, as well as the
> IDE controller, and size/type of drive. Have you used hdparm on the
> drive? There are several warnings about filesystem corruption in the
> hdparm docs, though the potential corruption is usually described as
> "massive" so I don't think it would corrupt only one file.

Nope, I've never made hdparm adjustments. I've been put off by the very
warnings you mention.

jed
-- 
http://s88369986.onlinehome.us/freedomsight/

... it is poor civic hygiene to install technologies that could someday
facilitate a police state. -- Bruce Schneier