On 9/4/05, Ralph Shumaker <[EMAIL PROTECTED]> wrote:
> Carl Lowenstein wrote:
>
> >On 8/31/05, Tom Gal <[EMAIL PROTECTED]> wrote:
> >
> >
> >>On 8/31/05, Ralph Shumaker <[EMAIL PROTECTED]> wrote:
> >>
> >>
> >>>[EMAIL PROTECTED] wrote:
> >>>
> >>>
> >>>
> >>>>----- Original Message -----
> >>>>From: DJA <[EMAIL PROTECTED]>
> >>>>Date: Monday, August 29, 2005 7:35 pm
> >>>>Subject: CD Tests in Fedora Install
> >>>>
> >>>>
> >>>>
> >>>>
> >>>>
> >>>>>The Media test in Fedora Core's Installer, at least since FC3, is
> >>>>>broken
> >>>>>and its results thus unreliable either way. Don't use it. In fact,
> >>>>>the
> >>>>>FC maintainers are going to remove that test in future versions. It
> >>>>>was
> >>>>>supposed to have been removed from FC4, but was missed.
> >>>>>
> >>>>>
> >>>>>
> >>>>>
> >>>>Can you recommend a good test, from linux and/or windows that would
> >>>>ensure the iso has been burned correctly?
> >>>>
> >>>>
> >>>>
> >>>>
> >>>On a whim, I decided to use my rh9 install CD to "test the media" of my
> >>>fc3 discs. The test as done by fc3 gives "pass" for discs 1 & 4 but
> >>>"fail" to discs 2 & 3. But the results from using the rh9 install CD
> >>>gives "pass" to all four fc3 discs. Can I rely on this?
> >>>
> >>>
> >>If it doesn't work now, why would it have worked back then?
> >>
> >>T
> >>
> >>
> >
> >Converted from top post, to preserve the sequence of thought.
> >
> >Presumably somebody changed the way that the media test works.
> >So I don't understand "worked back then". The observation is that new
> >CDs tested with the old test routine pass. The same ew CDs tested
> >with the new test routine sometimes fail.
> >
> >
>
> And consistently so. Using fc3's media test, if a CD fails, that CD
> fails every time with that tester. It is the same if a CD passes under
> that tester. And all the CDs that failed under the fc3 tester, pass
> using rh9's media test. No matter which tester is used, that tester
> always gives me consistent results for any given CD. The two testers
> just don't agree with each other. So perhaps rh9's media test gleefully
> awards "Pass" to every CD whether accurate or not. Or maybe fc3's media
> test is broken, although in a *very* consistent way. But the only thing
> from this that *I* know for sure is that the two testers are certainly
> different. There certainly has been some change between the two.
I just ran through some test. Using my CDs of FC3, which have been
checked for correct data both by the K3b checksum and by direct cmp(1)
comparison with the .iso file from which they were produced. The
results were as follows, mostly using laptop computer.
1) boot FC3_disc1, enter "linux mediacheck" for the boot-time command.
takes about 30 seconds to come up to the mediacheck screen
takes about 4 minutes to check disc1, it passes.
2) insert FC3_disc2, run media check. Stalls at about 50% finished.
reboot computer, take disc out, look at it, find that lower
surface is smudged.
clean disc with alcohol and lint-free tissue. Note, rub gently
in a radial direction.
3) check newly cleaned disc2 in desktop computer, read the whole thing
into a file and
use cmp(1) to check against original .iso file.
[EMAIL PROTECTED] tmp]$ dd if=/dev/hdc of=/var/tmp/FC_3disc2.img bs=2k
326426+0 records in
326426+0 records out
[EMAIL PROTECTED] tmp]$ cmp /var/tmp/FC_3disc2.img \
/media/20041205/Fedora/FC3/FC3-i386-disc2.iso
[EMAIL PROTECTED] tmp]$
# note -- cmp is silent if comparison succeeds
4) reboot laptop computer "linux mediacheck"
put disc2 in drive, check. It passes.
Conclusion: your media may vary. I myself tend to use the Fedora
mediacheck at installfests to check for compatibility between my CD
burner and someone else's CD reader. If disc1 passes, the drives are
most likely compatible. This takes on the order of 5 minutes, which
is a small time investment compared to the 45 to 90 minutes of a
reasonably full installation.
I will say again that CD burning software may put additional zero-fill
runout after the data copied from the .iso file. This will change the
MD5 sum of data as read from the disc. I know that there is an
expected MD5sum value recorded on the Fedora discs, I dont know
whether the expected data length is also recorded there.
carl
--
carl lowenstein marine physical lab u.c. san diego
[EMAIL PROTECTED]
--
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list
