I just burned 3 CD's (at speed 16 on a 40x capable burner) of "CentOS4
i386-bin CD1" (md5sum of image on HDD is OK, after padding it changes but
who cares)
One straight from iso, the other with appended 300KB zeroes, the third
with cdrecord -pad.
CD A: iso
CD B: iso with builtin 300KB pad (150 sectors of 2KB, IMHO 15 is enough)
CD C: iso with cdrecord -pad (15 zeroed sectors of 2KB = 30KB)
cmp -l /dev/cdrom file.iso-unpadded
CD A: IO error (within the last 30KB)
CD B: EOF on file.iso
CD C: EOF on file.iso
cmp -l /dev/cdrom file.iso-prepadded
CD B: OK
[wonder why I'd expect an
IO error (within the last 30KB of the padded image)]
running md5sum on CD A/B gives IO ERRORS, while C works (think this was
just a matter of luck).
My theoretical explanation which isn't perfect:
Why? Cause the burner can only burn in 32KB blocks (16 sectors) and
anything leftover doesn't burn (is left of the disk, so it can't be read)
(my drive: [LITE-ON ][DVDRW SOHW-832S ][VS0A])
CD A: Mediacheck FAIL
CD B: Mediacheck OK
CD C: Mediacheck OK
This proves pre-padding CD iso images before distribution with a few (30KB
at least, 300KB suggested) KB worth of zeroes fixes burn issues and
retains a working mediacheck - even if the last part of the burned image
is unreadable (since it'll be in the unimportant padding...)
Thus images should be padded after implanting the md5sum and before
calculating the md5sum for distribution.
ie.
mkisofs > file.iso
impland-md5iso file.iso
dd if=/dev/zero bs=2048 count=150 >> file.iso
md5sum file.iso > MD5SUMS
upload file.iso MD5SUMS
Cheers,
MaZe.
On Thu, 3 Mar 2005, Maciej ?enczykowski wrote:
> I'm pretty sure we should just prepad the ISO's with 300KB worth of zeroes
> before calcing their md5sums and distributing them, I'll check whether
> this works (what's a CD worth? 30cents?) in a moment and report back.
> If this only passes mediacheck (all reports are it should) then this will
> eliminate _all_ non-badburner and bad-media related burn errors (some 95%
> I'd say) and save a lot of trouble for a lot of people (ie I wasted a CD
> and 10 minutes burning the 4 centos4 cd's).
>
> Cheers,
> MaZe.
>
> On Thu, 3 Mar 2005, Collins Richey wrote:
>
> > On Thu, 3 Mar 2005 14:27:02 -0500, Lamar Owen <lowen@xxxxxxxx> wrote:
> >
> > > This is my last post on the subject since people seem to be more interested in
> > > splitting hairs than in helping people get a good burn on their hardware.
> > > And whether their hardware is broken or not is irrelevant; there is a
> > > recommendation that works for the vast majority of people that have the
> > > problem, and I simply looked through my archives to try to help people get a
> > > good burn. If I had known that people were going to nitpick my post I
> > > would not have bothered looking through my archives, nor would I have
> > > posted it.
> > > --
> >
> > Please don't consider my questions on the subject "nitpicking." I'm
> > just enterested in learning "the rest of the story," most especially
> > since it runs counter to several years of my own experience.
> >
> > Thank you for replying at length. Helping people to understand "why"
> > is somethimes just as important as understanding "how" or "what."
> >
> >
> _______________________________________________
> CentOS mailing list
> CentOS@xxxxxxxxxxx
> http://lists.caosity.org/mailman/listinfo/centos
>
