Data corruption on IDE disk via USB.

Data corruption on IDE disk via USB.

[Joshua Ross]

I have an external hard drive with an IDE interface.  This goes through
a USB2.0-IDE adapter.  I'm getting silent corruption of data when
copying big files to the drive, but not consistently.  The partition was
originally ext3, but I'm now mounting it ext2 and sync.  Still getting
the corruption.  

Running on: Linux albert 2.6.8.1 #5  i686 Mobile Intel(R) Pentium(R) 4 -
M CPU 1.80GHz GenuineIntel GNU/Linux.

Any suggestions about stopping the corruption/diagnosing what's going
on?

Thanks.

p.s.  Please cc me in replies since I read the list through hypermail.

Re: Data corruption on IDE disk via USB.

[Rogério Brito]

On Sep 29 2004, Joshua Ross wrote:
> I have an external hard drive with an IDE interface.  This goes through
> a USB2.0-IDE adapter.  I'm getting silent corruption of data when
> copying big files to the drive, but not consistently.  The partition was
> originally ext3, but I'm now mounting it ext2 and sync.  Still getting
> the corruption.  

Which kind of corruption are you seeing? I was seeing some non-silent
problems with an IDE drive connected in a Firewire/USB2 enclosure.

Using it with my iBook on MacOS X is perfectly ok, but trying to use the
drive connected to a USB 1.1 desktop (my main computer) under Linux also
presented problems when copying a large file (one of Dijkstra's talks).

This error, differently from yours, was consistent and produced a lot of
messages regarding the media of the "USB drive" being removed (!) from the
system (while it obviously wasn't) and it left me with corrupted
filesystems (I tried ext2/3, vfat and hfsplus, which are things that both
MacOS X and Linux can read).

I usually tried to use rsync to copy the files, but, if I remember
correctly, the problem would also manifest when using plain cp.



-- 
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
  Rogério Brito - rbrito@ime.usp.br - http://www.ime.usp.br/~rbrito
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

Re: Data corruption on IDE disk via USB.

[Joshua Ross]

On Wed, 29 Sep 2004 07:40:49 -0300
Rogério Brito <rbrito@ime.usp.br> wrote:

> On Sep 29 2004, Joshua Ross wrote:
> > I have an external hard drive with an IDE interface.  This goes
> > through a USB2.0-IDE adapter.  I'm getting silent corruption of data
> > when copying big files to the drive, but not consistently.  The
> > partition was originally ext3, but I'm now mounting it ext2 and
> > sync.  Still getting the corruption.  
> 
> Which kind of corruption are you seeing? I was seeing some non-silent
> problems with an IDE drive connected in a Firewire/USB2 enclosure.
> 
	Well, I'm currently stuck in the middle of a long fsck loop.  I
successfully ran fsck once, correcting errors.  I thought I'd double
check and ran it again.  The following is a short section from it:

Restarting e2fsck from the beginning...                                 
      
Pass 1: Checking inodes, blocks, and sizes
Inode 8 has illegal block(s).  Clear<y>? yes                            
      

Illegal block #4254 (1043168817) in inode 8.  CLEARED.
Illegal block #4255 (1779650960) in inode 8.  CLEARED.
Illegal block #4256 (119610448) in inode 8.  CLEARED.
Illegal block #4257 (1023971144) in inode 8.  CLEARED.
Illegal block #4258 (2687440014) in inode 8.  CLEARED.
Illegal block #4259 (1952817553) in inode 8.  CLEARED.
Illegal block #4260 (4225006687) in inode 8.  CLEARED.
Illegal block #4261 (885351209) in inode 8.  CLEARED.
Illegal block #4262 (2972213350) in inode 8.  CLEARED.
Illegal block #4263 (367138787) in inode 8.  CLEARED.
Illegal block #4264 (1016997915) in inode 8.  CLEARED.
Too many illegal blocks in inode 8.
Clear inode<y>? yes

Restarting e2fsck from the beginning...                                 
      
Pass 1: Checking inodes, blocks, and sizes
Inode 8 has illegal block(s).  Clear<y>? yes                            
      

Illegal block #4265 (3534702556) in inode 8.  CLEARED.
Illegal block #4266 (1881325444) in inode 8.  CLEARED.
Illegal block #4267 (965385379) in inode 8.  CLEARED.
Illegal block #4268 (1938219089) in inode 8.  CLEARED.
Illegal block #4270 (2567249910) in inode 8.  CLEARED.
Illegal block #4271 (339569508) in inode 8.  CLEARED.
Illegal block #4272 (3827439844) in inode 8.  CLEARED.
Illegal block #4273 (3719372914) in inode 8.  CLEARED.
Illegal block #4274 (423250370) in inode 8.  CLEARED.
Illegal block #4275 (1503941510) in inode 8.  CLEARED.
Illegal block #4276 (3125961630) in inode 8.  CLEARED.
Too many illegal blocks in inode 8.
Clear inode<y>? yes

Restarting e2fsck from the beginning...                                 
      
Pass 1: Checking inodes, blocks, and sizes
Inode 8 has illegal block(s).  Clear<y>? yes                            
      

Illegal block #4277 (4003997203) in inode 8.  CLEARED.
Illegal block #4278 (1927319268) in inode 8.  CLEARED.
Illegal block #4279 (1988850371) in inode 8.  CLEARED.
Illegal block #4280 (1493009126) in inode 8.  CLEARED.
Illegal block #4281 (96127493) in inode 8.  CLEARED.
Illegal block #4282 (1023388584) in inode 8.  CLEARED.
Illegal block #4283 (407114002) in inode 8.  CLEARED.
Illegal block #4284 (3769922052) in inode 8.  CLEARED.
Illegal block #4285 (1805259680) in inode 8.  CLEARED.
Illegal block #4286 (189165963) in inode 8.  CLEARED.
Illegal block #4287 (700103729) in inode 8.  CLEARED.
Too many illegal blocks in inode 8.
Clear inode<y>? yes


> Using it with my iBook on MacOS X is perfectly ok, but trying to use
> the drive connected to a USB 1.1 desktop (my main computer) under
> Linux also presented problems when copying a large file (one of
> Dijkstra's talks).
> 
	I've got it on USB 2.0.  I have a 1.1 port that I'll see whether I can
reproduce it on, once the fsck has finished!

> This error, differently from yours, was consistent and produced a lot
> of messages regarding the media of the "USB drive" being removed (!)
> from the system (while it obviously wasn't) and it left me with
> corrupted filesystems (I tried ext2/3, vfat and hfsplus, which are
> things that both MacOS X and Linux can read).
> 
	I'm not getting anything logged.

> I usually tried to use rsync to copy the files, but, if I remember
> correctly, the problem would also manifest when using plain cp.
> 
	I was also trying rsync, but am now restricting myself to cp to see if
that helped.  It didn't seem to.

Re: Data corruption on IDE disk via USB.

[Geoff Mishkin]

[-- Attachment #1: Type: text/plain, Size: 468 bytes --]

I have a similar problem with a hard disk enclosure.  However, instead of 
getting corruption, the device just stops responding and the activity light 
on the enclosure stays pinned.  Then, even after power-cycling the disk, I'm 
not able to remount it; I have to reboot.  Removing all of the USB modules 
and re-loading them doesn't seem to help either.

I'll post usb-storage verbose debugging output when I have some.

   --Geoff Mishkin <gmishkin@bu.edu>

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]