[TriLUG] CentOS image weirdness

[Brian Henning]

Hi Folks,
   A while ago I obtained the CentOS 4.2 Server CD iso via bittorrent. 
Verified the md5 checksum, burned the disc, all seemed well.  Got around 
to actually using the disc yesterday and it failed media check.  So I 
tried to checksum the disc itself to see if maybe somehow the burn 
process went wrong quietly....and I got an I/O error from the device. 
Hmm..  Okay, bad burn.  Chuck.  So yesterday I re-verified the iso's 
checksum (md5sum -c CentOS-4.2.ServerCD-i386.iso) and burned a new disc. 
  Immediately I went to checksum the disc, and lo and behold, another 
I/O error on the device.  Hm.  Okay, maybe it's because of the 
buffer-underrun prevention at 40x.  So I burnt a third disc, this time 
on CD-RW just in case, which automatically forced the speed down to 4x. 
  Again, I went to checksum the disc and it gave me yet another I/O error.

I took a look at /var/log/messages and it turns out the I/O errors occur 
at EXACTLY THE SAME PLACE each time!  Four sets of messages in /v./l./m. 
just like the following, only differing in timestamp:

Jan 26 16:12:27 localhost kernel: hda: command error: status=0x51 { 
DriveReady SeekComplete Error }
Jan 26 16:12:27 localhost kernel: hda: command error: error=0x54
Jan 26 16:12:27 localhost kernel: ide: failed opcode was 100
Jan 26 16:12:27 localhost kernel: hda: command error: status=0x51 { 
DriveReady SeekComplete Error }
Jan 26 16:12:27 localhost kernel: hda: command error: error=0x54
Jan 26 16:12:27 localhost kernel: ide: failed opcode was 100
Jan 26 16:12:27 localhost kernel: end_request: I/O error, dev hda, 
sector 1143396
Jan 26 16:12:27 localhost kernel: hda: command error: status=0x51 { 
DriveReady SeekComplete Error }
Jan 26 16:12:27 localhost kernel: hda: command error: error=0x54
Jan 26 16:12:27 localhost kernel: ide: failed opcode was 100
Jan 26 16:12:27 localhost kernel: end_request: I/O error, dev hda, 
sector 1143400
Jan 26 16:12:27 localhost kernel: hda: command error: status=0x51 { 
DriveReady SeekComplete Error }
Jan 26 16:12:27 localhost kernel: hda: command error: error=0x54
Jan 26 16:12:27 localhost kernel: ide: failed opcode was 100
Jan 26 16:12:27 localhost kernel: end_request: I/O error, dev hda, 
sector 1143396

The two sets of sector numbers are always the same.

Could this be a problem with the image?  I'm already downloading it 
again, but if the image were somehow wonky, I was of the impression that 
md5 collisions are rare enough that a few bits out of place in a single 
file have essentially zero chance of hashing to the same value as the 
uncorrupted file.  But three discs all having errors at exactly the same 
place just makes me wonder...  I suppose it could also be a problem with 
the hardware, but I've not seen it foul up any other big discs.

Thanks for the input!
~Brian


-- 
----------------
Brian A. Henning
strutmasters.com
336.597.2397x238
----------------