[lug] proper way to MD5SUM a cdrom?

[D. Stimits]

Sean Reifschneider wrote:
> On Sat, Jun 12, 2004 at 06:39:41PM +0000, D. Stimits wrote:
> 
>>Just a question on verifying CD-R's. I have the ISO, and have run md5sum 
>>on it, it looks good. Then I burn it with cdrecord. How do I verify the 
> 
> 
> Doing an md5 of the whole disc may work, but you may run into padding
> issues, where the CD has padding that the ISO doesn't.  You can md5sum
> the individual files within the ISO easily enough and that will solve
> the padding issue.
> 
> Unfortunately, simply doing md5sums on the invididual files won't tell
> you if a particular file was only readable after trying for several
> seconds or minutes.  So, an md5sum is a good first try at verifying, but
> we found that using that as our verification mechanism meant that we had
> a substantial number of users getting unreadable discs from us.

So how does one verify a whole CD? I found I get i/o errors if I try to 
md5sum the whole cd, even if it is on multiple cdrom drives and 
commercial cd's that I know are good. I was thinking that perhaps one 
could create an ISO file via dd of the CD? I'm not sure if I'd have to 
set a block size or not, so on, if I create a file via dd. If I did just 
a dd with if=/dev/cdrom and output to a file with no other parameters, 
can anyone guess whether the checksum of that file should be the same as 
the checksum of the ISO file used to create the CD? Or would the 
read/write processes alter this in some way that would alter the MD5 sum?

On a similar note, is it correct that the fedora core2 rescue disk 
requires disk1 of the install disks to be inserted once it gets past a 
certain stage? Or should the fedora core2 rescue disk be completely 
autonomous?

D. Stimits, stimits AT comcast DOT net