[NTLUG:Discuss] Multiple CD-R Burners / Observations & Questions

[kbrannen@gte.net]

David Simmons wrote:
> Guys,
> 
> For my company, I've created a CD that everyone wants me to reproduce
> for them - in an effort to reduce creation time, I've added a second CD
> burner to my Linux system.  Now the fun begins.  Here are my conclusions
> on what seemed to work best...and request for what others are
> using/doing and refinement of the CLI scripts.
...
> Questions:
> 
> 1).  What is everyone else doing with multiple CD-write drives for
> concurrent burning?

I use dedicated CD duplicators, with built-in inkjet printer and robotic arm, 
totally automated and very cool. :-)  Cost is $1600+ (depending on your 
needs); also requires Win2K or WinXP. :-(

If you want info on the little $1600 single burner unit or the $3400 2 burner 
unit that we use, email me offline and I'll tell you anything you want to know 
about them.

> 
> 2).  How can my CLI script be re-written for more reliable performance
> and indication of a PASS! or FAIL! of the burning process.

Be sure you not only look at STDOUT, also watch STDERR.  Other things to watch 
out for:  make sure your media supports your burning speed (i.e. don't try to 
burn at your writer's max 48x if the media only goes to 40x); also remember 
that your bus must be able to shove all the data down it you need *with time 
to spare* so you might want to try lowering your burn rate which should reduce 
your error rate.  Also don't forget that you read/write CD data at variable 
speeds:  the inner area can only be written at 24x no matter how fast your 
writer will go, you don't get to use the higher speeds until you get closer to 
the outside; so if all your ISO files are 300M or less, don't bother with 
anything over 24x.

You should also know that not all media is created equal, some is bette than 
others, so you may be having problems there.  Though my understanding is that 
the quality of the media generally affects useful shelf life, but it can 
affect other things too, or so I read.

> 
> 3).  Noticing that cdrecord -scanbus shows many empty 'slots', it seems
> like the theoretical limit would be 8 burners per machine - is anyone
> doing this?  What are the 'gotchas' and work-arounds?

You might could do this if all your burners were SCSI and you had an UW-SCSI-3 
card/cables/devices, but sending to more than 1 or 2 burners individually and 
simultaneously sounds problematic to me.  The decicated machines like I have 
have builtin buffers (probably a small embedded OS type computer) so you feed 
the ISO to it, which feeds the data to all the burners at the same time.  That 
reduces bus requirements (congestion).  Even so, the units still require a 
firewire connection, everything else (serial/parallel/USB) is too slow.

But for single unit burns on Linux, I've always had good luck with using 
cdrecord.  You might want to increase the buffer size to cdrecord too (e.g 
-fs=8m).

> 
> 4).  Per Note2 above - it was necessary to create different directories
> of files for 'locking'/permissions issues.  If I had 8 burners would I
> need 8 different directories of files and/or 8 copies of the same .ISO?

That suprises me, I've never be aware of that, but then with a single drive I 
guess I never would notice. :-)  Are you burning as root or another user?  I 
always do it as root to avoid "permission denied" errors when reading files; 
but then 90%+ of my Linux burning usage is for backups (tar -> mkiosfs -> 
cdrecord).

HTH,
Kevin