You will receive plenty of replies on this - so I hope I don't blow it
with something wrong or stupid... anyway, on to the questions.

| There are five connectors and two switches that look "stock": (1) a
| DB25M, labeled with arrows pointing left and right; (2) a DB15M,
| labeled with a silhouette of a keyboard, something that is probably
| supposed to be a monitor screen, and an arrow with the head circled;
| (3) a DB9M, labeled with what is clearly supposed to be a printer; (4)
| a BNC connector labeled with a strange little crosshatch and the letter
| A; (5) a DB15F labeled with a similar crosshatch and the letter B; (6)
| a two-position switch labeled A and B, near connector 5; (7) what
| appears to be a momentary-contact pushbutton labeled with a triangle
| inscribed in a circle.

The DB25M is I believe a serial port.  The DB15M is a custom connector
to a cable harness that bundles a monitor, keyboard and mouse (thus
the three symbols).  The switch slects either "A", the BNC (10Base2)
Ethernet connector, or "B" (10Base5) connector.  The momentary contact
switch is the "reset" button.  The DB9M is your friend.  DEC has two
cables for this, one is labelled "printer" and the other "console."
The console cable shorts pins 8 & 9 I believe which tells such a system
that it is a console and not printer cable and thus the PROM monitor
will talk to this port (i.e. this is how to get a console without a
"head" on the unit).  Your uVAX-II/GPX console cable will work here!

| There are also three spaces at the bottom back, labeled A, B, and C.  A
| has a 50-pin Centronics connector, B has a DB50F, and C has a blank
| plate behind it.  I was unable to disassemble the machine very far, but
| I did determine that the DB50F at position B has a small board behind
| it marked "ST506/512 INTERFACE ADAPTER", with two ribbon-cable
| connectors with nothing plugged into them.  The Centronics connector at
| A is attached to a ribbon cable that disappears into the machine.

The DB50F is a custom ST412 MFM cable to an external box that looks
like the VS2000 box but instead contains an MFM drive (e.g. RD53, RD54).

The Centronics connector is for an external SCSI based TK50 drive.
This is a neat box because it provides a SCSI to TK50 card inside.
But this is also an access to the NCR 53C80 SCSI onboard too.

| - How do I get the thing apart?  I managed to take off the base, but
|    further disassembly seems to require snapping off either the front
|    or back plastic piece, neither of which wants to come off.  Or do I
|    just have to be firm and force the sides to spring apart enough to
|    lift the metal cover off the top?

Remove the bottom "pancake" unit.  It is the cable expander that adds
the SCSI and MFCF "ports." This 1" thick "box" that is attached to the
bottom covers the 4 or 6 screws that hold the cover on.  Please do not
"rip" the ends off or you'll be very unhappy.  The cover wraps around
the two sides and slides off.  you'll find things are fairly crammed in
and if the unit was used much is likely full of dust and such.

| - How do I install NetBSD on it? :-)  The hardware list lists the
|    VS2000, so it should be possible.  If I'm lucky enough for the
|    Centronics connector to be SCSI, can I maybe build a boot disk on
|    another machine, or do I have to netboot for the initial install?

It will netboot very well from my past experience.  This involves
setting up RARP, MOP, BOOTPARAMS, TFTP and NFS I believe, but the
port web page has a great "netbooting" item you'll want to peruse
that details all this.

My personal comments:

The VS2000 is a frightenly fast box (just kidding, 0.9MIPS at best) with
a rather unique bus architecture (the disks, both MFM and SCSI go through
a 16Kb "buffer" memory - I presume that fitting the logic to allow DMA
to main memory was just to expensive or no space - the Ethernet can do
direct DMA though) and as a result tends to make development a rather
"dedicated" activity 8-)  Bartram did the original work on the vs2000
port I believe, but the MFM and SCSI support was removed 6 months or more
ago to make way I believe for an "improved" version.

I sure wish SCSI controller chips were no so difficult to write drivers for!

-- Ken