Subject: Re: Oooooh... IDE to Q-bus/Unibus adapters.
To: None <port-vax@netbsd.org>
From: Geoff Roberts <geoffrob@stmarks.pp.catholic.edu.au>
List: port-vax
Date: 07/08/1999 13:25:09
----- Original Message -----
From: Brian D Chase <bdc@world.std.com>
To: Michael Kukat <michael@camaronet.de>
Cc: emanuel stiebler <emu@ecubics.com>; <port-vax@netbsd.org>
Sent: Thursday, 8 July 1999 6:02
Subject: Re: Oooooh... IDE to Q-bus/Unibus adapters.
> AAAARRRGGGHHHHHH!!!
>
> How the did this end up in a SCSI vs IDE war? Who the hell really cares
> anyway. WE'RE TALKING ABOUT MICROVAXEN HERE!!! It would of course be
> very nice if everyone had Q-bus SCSI controller, or MCSP speaking UltraDMA
> IDE controllers for their Q-bus and Unibus MicroVAXen, but it's not going
> to happen.
Just on the subject of SCSI controllers in Vaxen,
I picked up, on behalf of the school here, 3 Vax 6000 machines on Monday,
from my favourite scrapyard. ($300 for the lot) This in now on topic, they
are 1989 build, I think.:^) though not micro I hasten to add, they weigh in
at 700lbs or 318Kg.....
Having searched for and never found a SCSI controller (apart from $$$$ new
ones) that would work in one, I was very pleased to find in a 6000-420 from
this lot, not one but THREE! CMD CBI-1000 VAXBI SCSI Controllers, 2 x /MB
(Disk only) and 1 x /TM Disk and/or Tape. I can now contemplate building
up one machine with all it's disk drives in the cpu cabinet. And my 4mm DAT
tape can now be cluster served again. Haven't been able to do that since
the strange Emulex Tape controller in the HSC50 died. (Well, the interface
in the enclosure with the 2 x DAT drives anyway, the HSC card is still ok I
think.
It emulated the DAT devices as TA78's to the Vax, (if you can imagine a 4 Gb
TA78) The surviving HP 4mm DAT is currently living in a VS4000-90, where it
works ok, but I'm not able to share it to the cluster, for some reason.
Included was:- 6420 with 128Mb of RAM and it's SCSI stuff - an XMI bus only,
(never seen one of them before) 6440 with a DSSI controller, CI and
Ethernet, 196Mb of RAM(!) and a very tired 6240 with 96mb.
The 6240 has been dropped from a great height at some point and is really
bent, twisted and beaten up, (front door torn off) though it would probably
still run.
The 6440 is somewhat dinged, bent and scratched, ("body rough" I guess you'd
call it) but otherwise ok. The 6420 is fine apart from a few minor dings
and scratches. I'm spare parting the 6240, and probably the 6440.
Next week I have to go and pick up a retired, complete, CI based cluster,
that only came off DEC Maintenance in September 98, 1 x 6410, 1 x 6240, 2 x
HSC70, 2 x TA79, 1 x Star Coupler and stack of RA9x and RA7x drives in an
SA6?? cabinet. Direct from the owners. So they will be in mint nick. No
coin, just come and get it.
This is gonna take at least 2 trips (130 miles each way BTW) since the
Triton 1 ton ute won't take it all in one go. It struggled a bit with the 3
x 6000's in fact.
Consequently, the school's lone 6320 is getting a major upgrade. It's now
fat with 256mb of RAM courtesy of the beat up 6440, and is going to get it's
CPU's and one of the CMD SCSI controllers and XMI based CI & Ethernet as
well. Plus we will have one complete spare machine ready to plug in.
The HSC50 will now retire to a working display I intend to set up and be
replaced by the HSC70.
I haven't figured out exactly what bit's will go where, but I'm counting on
several Multi-CPU machines when the dust settles.
Some of the excess then comes to my private collection, so I will score a
mult-cpu box of my own, a 6260 with 128Mb and a SCSI controller if it all
works out - I will probably just update my 6220 to that level, (since I've
already converted it to single phase) and have plenty of spare parts ;^)
BTW, anyone know how some good startup SYSGEN values for a 6000 that had
64mb originally and now has 256Mb? It runs Multinet 3.2 Rev B, WASD Web
Server/Proxy Server, MX mail, IUPOP3, Madgoat FTP, & only 2 or 3 interactive
users - me and the developer of WASD mostly - we're his beta site.
(The day to day file serving for the school is done by a Novell box.)
I can't see why I can't get it to a point where there are essentially no
page faults, considering the workload and available memory. I have bumped
up some of the PQL_xxxxxxxx Sysgen variables, but am really out of my depth
here. I have managed to make the correct changes in SYSGEN to have VMS 6.0
recognise all the memory, however, for some odd reason it won't accept -1 as
a variable in PHYSICALPAGES which should make it use (according to the VMS
FAQ anyway) all the memory it finds. I set it to 524288 and SHOW MEM says
256Mb so I guess it's happy.
But if I try SET PHYSICALPAGES = -1 SYSGEN says SYNTAX ERROR.
AUTOGEN? Last time I tried that, it hosed up the system so bad it took an
hour to get everything working again No idea what went wrong but I'm in no
hurry to try it again. When I get a second system up, I'll see if I can find
out what happened.
I'm reading the VMS tuning instructions, but the signal to noise ratio is
still rather on the high side at my level of comprehension.
Greetings from (Wet, coldish and raining) Country South Australia.
Geoff Roberts
VK5KDR
Computer Systems Manager
Saint Mark's College
Port Pirie, South Australia.
Email: geoffrob@stmarks.pp.catholic.edu.au
netcafe@pirie.mtx.net.au
ICQ #: 1970476