Not solving your current problems, but there are reportedly severe
differences in longevity depending on the dye (and thus the brand
of CD-R blanks). I saw a recommendation for Kodak, so I record
backups onto Kodak and also at least one other brand.
I had a longer discussion about this with a friend of mine.
Unfortunately Kodak stopped making CD-Rs some years ago. Alternatives
are rare - TDK was mentioned, but we both have packs of TDK disks
bought 2 years ago, unused, which are distinctly cloudy on the
recording surface now, obviously some chemical reaction with the disk
surface and who knows. They have been stored in their sleeves as they
were manufactured, shrinkwrap not even opened. Good dye maybe - rubbish
otherwise. There are many factors which go into longevity, the dye is
only one. The fact that manufacturers change processes, materials etc
every few months makes it impossible to get anything decent in the long
run. You're also restricted to what you can actually buy, no good of
some obscure place has something better. Most shops here sell rubbish -
they tell you to expect 10-20% failure rate before it's even burnt
(excuse me?). Every single one sells Verbatim = Imation as "top end of
the scale", when I asked whether they had something better it was
obvious that none of them had any clue whatsoever. Very few
manufacturers have published their longevity test results, Kodak + TDK
included, the others were obviously only selling promises in the first
place. It's impossible to do anything similar yourself. Bottom line:
we're stuffed.
Your burner is also a big factor - it'll have to be well-calibrated. It
can be observed that e.g. Sony CRX19x/2xx models lose their calibration
optimum and start burning errors after about 300 disks burnt. The other
stuff on the market is almost certainly the same, with the cheap stuff
being worse.
The best would probably be to record everything twice on different
brands and types of media, on short + long strategy types for example.
Use your best guess as for what's good quality at the time, there isn't
much more real info to go on than how professional the top print looks.
Think ahead and record md5 sums. Operate on the assumption that you
can't buy decent quality media today.
As for your problem, there is no iso9660.fsck. Try as many different
drives, especially burners, to read off your data. Use new, not used,
equipment. Try a very very old cdrom drive. Use dd to try and read
single blocks, piece your important data together with a hex editor if
need be. Unfortunately, it seems CD-Rs aren't any better than tapes.
Copy your disks onto new media every few years...
For a long time I used "Imation" because it got a good review on contrast
in several articles. Lately I use what I can get cheap... Doing a c2scan
after burning seems a good idea, and I generate a table of md5sums which
I add to the CD before burning (and keep on disk). I've had very good results
burning using the audiomaster option, but the sample size is not large enough
to mean anything but "I've been lucky so far." I have had zero error c2scans
using that option, however, so maybe it does something useful. Then again
it may do nothing for data and just be a no-op, as I say, the sample size
is only a few dozen CDs.