SUMMARY: Daisy Chained CD-ROMs

From: Danny O'Mahony (danny.omahony@lineone.net)
Date: Mon Mar 20 2000 - 07:21:48 CST


Hi,

Firstly, apologies for the way too late summary of this problem - it's
taken (sort of) this long to get it sorted.

-----------------------------------------------------------
This was the original mailing:

I've got an old sparc 20 acting as a CDROM server. It
has 3 CDROMS attached in a daisy chain, terminated at
the last device.

The problem I have is that only one of the CDROMs is
automounted when you put a cd in, even after a boot
-r. The targets for the devices are: 0, 2, and 5. The
two internal disks are on 1 and 3 (being the boot
disk). I've wiped the /etc/path_to_inst and done a
boot -ar to rebuild it.

So now, even though I load 3 cds, only one shows up
with a df -k. Before someone rebuilt the machine (I
know I know, but I was on holiday!) all three loaded
up when inserted....

I've tried mounting each individually and they
failed....
------------------------------------------------------------

Most of the replies asked me to check out the config of /etc/vold.conf.
My vold.conf was exactly like the standard, default, vold.conf file that
is built when the OS is loaded.
All the devices showed up on a probe-scsi-all, at least so I
thought.......(more later)
Thanks to Matthew Stier who pointed me in the direction of
/etc/rmmount.conf, not the solution but taught me something about
Solaris that I didn't know!

Although all these suggestions were areas that were of use as a
potential starting place, they did not, however, point me to the
solution!

What had actually happened was this. While I was away on an induction
course one of the other sys-admins had brought the machine down to
rebuild it as a pure X-server. While he had done this he had added an
~**internal**~ cd-rom, which no-one had noticed because the side of the
machine is very near a wall (meaning that to get to load a cd you have
to move the sparc-20!! - I know, don't ask!). For some reason he had
also unchained everything on the scsi bus but what he didn't notice was
that in his meddling he had inadvertently assigned two hardware devices
the same target number (i.e. two cdroms on target 1). Probe-scsi-all
worked, I thought, as it found 4 cdroms, but what I failed to notice was
that one of the cdroms had now become target 6!

It took ages to find this one out, after having messed with vold.conf
and rmmount.conf to no avail. It was only when I had a sort of free
afternoon that I decided to pull the machine away from its corner and
really investigate. What essentially this all means is that once all the
target numbers were reset back to some sort of order, there was more
than 7 devices on the scsi bus, which you can't have without fitting
another controller as we all know! The outcome, we kept the internal
cdrom because it was faster and removed an old cdrom (1-speed or 2-speed
I think) to bring the number of devices to an working amount!

Phew!

So thanks to all that helped out, as an aside one 'helper' asked a
question that maybe we all should think about when posting:

"With what errors did the operation on the cdroms fail with?"

How many, and I include myself here, posters post a problem with no
error messages? Just a thought.

Cheers,
Dan O



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:14:05 CDT