On Mon, 5 May 1997, Lucio Chiappetti wrote:
THE QUESTION
[about writing CD-ROMs using]
Elektroson's GEAR 3.2 software on a Sun, to which the CD writer is
attached. Our Suns and Alphas share (or may share) disks via NFS.
We create a physical image with "physvol", for reason of space on another
Alpha disk seen in NFS by the Sun. However although the test write
succeeds a real write gives
> a "50h" error (the manual says just "WRITE APPEND ERROR indicates that an
> append by a write command during writing failed")
--------------------------------------
SUMMARY IN SHORT (details below)
For a production environment, usage of a dedicated workstation for the
CD-writer and the physical image disks is definitely recommended.
In our case that is not applicable. It appears that using a physical
image via NFS is perfectly acceptable. However the physical image must
be kept short of the CD capacity (could it be an h/w problem ?)
--------------------------------------
SOME OF THE REPLIES
(the "best" ones at the end ...)
Some suggestions are unfortunately inpractical or unfeasible for economic
reasons in the current situation, like :
>3.) Replace your CD-Writer if you have a Phillips CDD2000 or a HP 4020i.
(we have indeed a Philips ...)
> You could also consider getting a GEAR license for your alpha.
> Attach the CD writer to the Alpha
all too costly (and untimely)
>it is recommended that the CD-Writer and a disk where the data are, to be
>written to the CD are alone on one SCSI-Bus.
What is written on our manual was that the disk and the CD had to be on
SEPARATE buses (as they are). Having those buses with ONLY those devices
implies having a dedicated machine for making CDs, which is uneconomic in our
arrangement.
--------------------------------------
Some other were not applicable like :
> 1.) lower your writing speed
we are already writing at the lowest speed (1x)
> If you feel that the SCSI bus is too active on the Sun, you could always
> drop in a second SCSI card and run the CD drive through that.
the CD-R is already on a separate bus than all disks. It is easy to
avoid using the other device (i.e. the DAT) which is on the same bus
when using the CD-R.
> use a different software like mkisofs
--------------------------------------
Consensus was against use of NFS by all respondents
Eberhard Heuser-Hofmann <vaxinf_at_chclu.chemie.uni-konstanz.de>
David Warren <warren_at_atmos.washington.edu>
Dave Golden <golden_at_falcon.invincible.com>
"WHITTAKER, Bruce" <bjw_at_ansto.gov.au>
Kevin Reardon <kreardon_at_cerere.na.astro.it>
Harald Baumgartner <hmb_at_rosat.mpe-garching.mpg.de>
MARCO_at_astbo3.bo.astro.it (Marco Lolli)
However it has to be noted that we do not use a virtual image over NFS (that
will require a lot of directory lookups and other activities which implies a
ragged performance), but a PHYSICAL image over NFS, so the flux should be
fairly constant (and low, at least as far as I can tell looking at the
blinking LED of the disk on the Alpha). Moreover a colleague of mine always
succeeded in running physical images over NFS), and after I've now produced my
9 CDs I also succeeded !
Our main argument against using a physical image on a local disk, is that it
would imply to dedicate a machine and its buses and disks to write CDs.
THIS MAKES PERFECTLY SENSE TO A SITE RUNNING ROUTINELY "PRODUCTION" OF CDs
(e.g. an observatory archiving systematically all data), and I would
definitely recommend that to anybody intending to run such a thing.
Some people reported succesful use of such a combination for 100's of disks.
In our situation where this archiving is occasional, such an arrangement is
wasteful for our resources
> We now feel there might be an hard limit on the capacity of the CD (possibly
> also depending on the size of the files, and the way they fill disk sectors)
In fact the problem occurred twice exactly in the same place (when I
burnt my first two disks of this lot)
I kept the physical image always shorter than 600 Mb (602 actually) and was
able to write the CDs successfully (9 out of 11 cases).
However in 3 out of 9 cases I managed to have the track completely
written, but I got a 44h error ("unexplained h/w error") during fixation.
In this case I exercised the following procedure :
- exit GEAR
- eject the disk
- reinsert the disk
- reenter GEAR
- issue "discinfo" command
- issue "fixation" command
- bring the disc to Alpha and compare with original data
In 1 out of 3 cases the disk was unfixated, and my manual fixation was
successful. In the other 2 cases the disc was apparently already fixated !
In all 3 cases a compare indicated that the disc was OK.
The comment on replacing the CD writer might be applicable, or otherwise it
might be an hardware problem ... anybody knows how these disks are physically
written ?
I suppose data are on concentric or helix tracks from the centre to the border
or viceversa. Since I'm having problems always at the end of the data, could
it be that some mechanical tolerances perturb the writing at the disc centre
or border ?
----------------------------------------------------------------------------
Lucio Chiappetti - IFCTR/CNR - via Bassini 15 - I-20133 Milano (Italy)
For more info :
http://www.ifctr.mi.cnr.it/~lucio/personal.html
----------------------------------------------------------------------------
Received on Fri May 09 1997 - 14:59:14 NZST