Re: 3B2 Disks

On 2009-01-18, Bill Gunshannon <billg999@xxxxxxxxxxx> wrote:
In article <slrngn5e55.g95.dnichols@xxxxxxxxxxxxxxxxxx>,
"DoN. Nichols" <dnichols@xxxxxxxxxxx> writes:
On 2009-01-16, Bill Gunshannon <billg999@xxxxxxxxxxx> wrote:
In article <slrngn0af1.298.dnichols@xxxxxxxxxxxxxxxxxx>,
"DoN. Nichols" <dnichols@xxxxxxxxxxx> writes:
On 2009-01-15, Bill Gunshannon <billg999@xxxxxxxxxxx> wrote:
In article <slrngmtmo2.40g.dnichols@xxxxxxxxxxxxxxxxxx>,
"DoN. Nichols" <dnichols@xxxxxxxxxxx> writes:

[ ... ]


I keep trying to convince people that we really need to revive this email
system. It offers the best and quickest solution to the SPAM problem but
apparently I am the only one who can see that.

You know -- I think that would work! (Of course, it would bring
most uses of email to its knees given today's user base. :-)

Why? It could, thoretically reduce the amount of email traversing
the network by, easily, an order of magnitude as there is at least
that much junk. If anything, it would increase the performance.
The only shortcoming in the concept of UUCP based Email is eliminated
by the use of the INTERNET as the transport medium. Seriously, what
would you see as the shortcomings of this idea?

What I was suggesting is that the user base would have to be
totally re-educated

Why? this is a change to the method used by MTA's, not MUA's. The
users would continue to read email with Firefox, Netscape, Outlook,
even Pine or Elm, whatever their favorite program is.

How many MTAs automatically generate bang paths these days? And
you would still need a massive database to generate the bang path.

-- and a lot of them would fight it tooth and nail.

If done correctly, they would never even now a change occured!

Except for the greater number of messages falling through the
cracks as systems are shut down.

Aside from that -- bang paths are vulnerable to a single machine
in the path going out of service between the receipt of the incoming
e-mail and the sending of the reply.

Considering the nature of connectivity over the INTERNET, there should
be no one of consequence with any single path to any other location.

What would a bang path look like which used the internet for
connectivity? You still need to know names of systems at each end at
the least. And IIRC, uucp required that there be no two systems with
the same name -- making a naming problem with the number of machines in
use today. After all -- without domain names, it is difficult for a
database to distinguish between systems with the same name to generate
the proper bang path.

The example which I showed above only used uucp to get to uunet.
From there, things went by domain name.

The one thing to remember, is that unlike the days of doing this over
the phone lines there is no additional cost for connecting to any
other location. And you don't have to wait til after midnight when the
rates go down. :-)

So -- how is it done without domain names between the main trunk
servers at least? *Pure* bang paths required the name of every system
between the origin and the destination.

But it would sure make it easy to track down the originator of
spam -- or at least the compromised machine sending it out. :-)

It certainly would. Which is why I mentioned the ability to assign
civil liability. Because no one could send email without going through
a legitimate MTA (no more need for RBL's to try and figure out if that
machine is a dial-up residential connection) the responsibility for
controlling users falls on the service provider. If he wishes to
remain in the email network, he controls his users. If he chooses
not to, he will find himself with no one willing to peer with him.
He is out of business.

If the ISPs would simply block port 25 for all DHCP and dial-up
systems, you would have the same conditions.

[ ... ]

I dumped all my big disks (RA-series and Fuji Eagles) but VAXen can
use smaller, more modern disks. I have a number of VAXStation-3100's
that use SCSI and sit on a desktop running VMS quite well.

Sure -- but he had multi-bay machines which he could never run
as well. :-)

I have some VAX 4000's. I haven'y used them lately, but that is mostly
because of a lack of time and no current project requiring that kind of

O.K. Did you ever run ultrix on a system? My friend has the
tapes for that acquired from one of the systems he got.

Of course,
I think they were also the ones who gave us "Eunice". :-)

Oh yes -- the only experience with "Eunice" that I have is the
comment from Larry Wall's "configure" script "Oh good! You're not
running Eunice." which told me that it must be pretty bad. :-)

Probably the first attempt at what later became known as POSIX. :-)
Actually, it worked OK. It was near the resource hog the first Ada
compiler was!! :-)

Hmm ... I remember being shocked when I heard that DEC claimed
POSIX compliance for VMS. :-)

They wrote a POSIX system that ran on top of VMS (exactly the way EUNICE
worked) that met the requirements. As far as I know, they later abandoned
the project.

O.K. Before DEC abandoned doing business entirely.

And, IIRC, the major problem with running unix-inspired things
on VMS was that process creation on unix was dirt cheap -- just a
"fork", However, on VMS, it was quite expensive, so a program which was
quite happy on even an old v7 unix machine would bring VMS to its knees.

Worse than that, while exec() is doable on VMS a real fork() is not.

Right -- thus the computational cost of emulating fork().

[ ... ]

It grew gradually to become
the center of USENET and the non-AT&T UUCPNET. Then one day, someone
at the government realized that the machine, in a government site, run
by the government, consuming government resources was not doing government
business and ordered it to be shut down.

Hmm ... this reminds me of what happened to the system running
TOPS-20 -- a major repository of open source software. Ah yes -- I
remember the name now -- simtel20. Many's the time I telnetted to that
from work -- and even some from home eventually.

That was handled much worse. I rbnelieve the the guy who used to run
SIMTEL was actually let go when they shut down the machine. :-)

Ouch! Was he let go because of the shutdown, or did he retire,
and nobody else was competent to run the system?

[ ... ]

UUNET seems to be dead, but still is out there -- it
just hasn't been updated in years.

Domain belongs to Verizon. ;-(

Sigh! So does the copper that my T1 comes through. And they
keep trying to push me onto FIOS -- but won't offer me static IPS, and
especially a Class-C block of IPs.

Probably because there are none left available. :-) Have to wait
for the arrival of IPv6.

No -- because they are trying to push it on end users, who will
accept the combination of FIOS for phone, cable, and internet connection
all at the same time. And the box which connects to the internet only
knows how to NAT for a single (dynamic) external IP. I *need* static
IPs, because I run web servers and mail servers.

I probably could talk to their business account division and get
something more to my liking -- without the cable-TV and phone rolled in.
I particularly don't like the FIOS for phone because there is a limited
time that you can lose power before the local electronics run out of
battery charge.

[ ... ]

existed in NEPA. They could not see where anyone would be willing to
pay money for something like Email or News. :-(

Interesting. *I* certainly was willing to do so -- even while
working for an Army lab and getting access to Arpanet at the time.

Yeah, that was a great example of the shortsightedness of people around

Where is "here"?

Northeastern PA. Third or fourth largest metro area in PA.

O.K. You are not that far away, then. I'm just to the West of
DC in Virginia -- just outside the Beltway. And luckily, I don't have a
need to go into DC for the next few days. :-)

I had a similar experience with a local PC company who contracted
me to set them up in the ISP business. After many hours of work I showed
up one day, not too long before we would have gone live, to find the
servers I was building wiped and loaded with Windows. I was told they
decided not to pursue that line of business. They gave me a check for
what they thought my work was worth and told me goodbye. I could have
sued them for breach of contract in order to get the rest of my money
but decided it wasn't worth the headache. I have never done any contract
work around here since and never would.

Ouch! You should have used machines which would not support
Windows -- then they couldn't do that. :-)

They were their machines. One of the selling points was the very low
investment to get into the business with a very quick ROI. As an
interesting aside, two ISP's popped up shortly afterwards. One a small
two man operation in a hick part of town and the other a major non-Bell
phone company. Both suffered the same problem. Trying to keep up with
the demand. The two man operation ended out moving into the city as they
quickly exhausted all the available phone bandwidth into their hick town.
The phone company grew to more than 1,000,000 customers in the first year.
At $25.00 a month! Hmmm..... I make that out to be somewhere in the
neighborhood of $300,000,000 gross a year in addition to their already
existing phone revenue. Not bad for something people around here were
telling me there was no business viability and cetainly no liklihood of
profit in!

Hopefully the company which shut down your project is still
kicking themselves now for what they lost. :-)

[ ... ]

It *did* make it easier to break into a system which you had
lost the password for. Log in as guest (by default no password), send
yourself some e-mail, then click on the mail icon when it popped up, and
bang out of the mail program to a root shell. :-)

Yeah, I remember that. Kind of like iPhones that all over the world have
the same root password. ;-)

Ouch! I don't have one. Do they even give you access to change
the password?

Sure. it's just a linux box in your pocket. Has an sshd available.

Yes -- but through what path? Does it have an ethernet jack?
Or are you stuck having to expose it to connections over the air?

It's a cellphone. Of course it's exposed over the air. It's also a
WiFi device. Without that it's not an iPhone.

But are logins over either allowed by default? If so, then they
should be forced to allow you to change the root password.

Do they warn you that it should be changed.

It can't be changed. If you change it, the phone don't work anymore.

That is terrible.

Luckily, there is no daemon accepting incoming connections by default,
but many people turn on ssh. And, one never knows what gets done when
you visit one of those whiz-bang web sites. :-)


Given the number of attacks that I see against ssh I would not
want a system which I had not been able to set a good unique password

[ ... ]

Yeah, I have a couple of QBUS M68K cards that would make great OS9-68K
machines. And, it's ROMable!!

I actually have such as set of Qbus cards too.

Integrated Solutions?

I don't know. The cards are out in /dev/barn01, with a lot of
stuff which I would need to move to reach them, and I don't have the
cage to plug them into.

You want one? :-) of course, you'll need more QBUS modules if you
actually want to use it.

Which -- the backplane? Yes, I would like one.

I have quite a few, including some that aren't even DEC.

There was a
suite of about four or five boards with it.

It would be interesting to know what they are.

IIRC, there is a ribbon
cable bus on the outside edge of the cards to supplement the Q-bus.

Actually, that's probably the cable from a disk controller to the disks
which would be in a differnt box and maybe even a different rack. 8"
floppies or RL02 disks.

Nope -- this just connected the several cards together -- and
the multi-connector cable was there with the boards, ending at the last
board on each side. I think that it was either implementing dual-ported
memory, or adding data bus width. What is the maximum address bus
width on Q-bus? Could it even approach the 16 MB of the 68000, let
alone the 4GB of the 68020 and later?

Now that I think of it -- it had apparently run Sys-III, and
I've still got a box of the OS manuals out in /dev/barn02 on top of a
rack full of stuff.

While not overly thrilled with having yet another version of Unix I
would still be interested in any software you might have.

No software -- just the cards and the manuals. My friend got it
for the Q-bus cage.

Sadly, if
these are in fact the same thing I have I fear your's are also with
Unix boot PROM's.

I suspect so. Any chance of getting sufficient documentation
about the boards so you could write your own boot PROMs?

I know that the COSMOS CMS-16/UNX was very well documented, but
it lived in Intel's MultiBus.

Right now -- even if I had lights in that building -- I would
not go out. It is bloody cold out there right now,

Not as cold as here. I had pipes wrapped with an electric heat-tape
and insulation that were frozen this morning. Got to move south!!

Ouch! Where is "here"? I'm in Northern VA, FWIW.

Scranton/Wilke-sBarre, PA

And how much colder did it get than here? We saw a low of 7 F
late at night (about 2:00 AM).

Of course, this house was re-built, expanded, and insulation
improved about fifteen years ago.

[ ... ]

If so, which PROMs? They had two. One booted a
version of Unix (which I have not been able to find) and the other was
MIKBUG (I think). Mine have the Unix boot PROMs. :-(

How do I tell? Especially without a card cage to run them in.

Well, you can read the label on the PROM if it's still there, otherwise,
you have to fire ut up and see what you get for a prompt.

Depends on whether the PROMs just had a label stuck on (which
may have faded or blown away) or whether they were metal-capped PROMs or
something with a MelInk stamped identifier. Hmm ... not EPROMs, but
bipolar single use ones?

Mine have EPROM's. I suppose that was the easiest as they offered more
than one version of firmware.

O.K. I'll see what I have when the weather warms up a bit.

[ ... ]

Yep! And the later CoCos actually had 128K of RAM, and ran
Level-2 OS-9.

Yeah, I got that, too. :-)

I never did pick up one of the expanded ones. About that time,
I was digging into multiple unix boxen, starting with the COSMOS

I've got more than enough Unix boxes. As nice as Unix ism some of the
other OSes had a lot to offer as well. Like OS-9 and RSTS.

Yes. Though I have no experience with RSTS. What hardware
would I need to run it? Dual-wide LSI-11, 64 MB of RAM, quad serial
port board, and floppy controller board? Plus the card cage which just
barely hold all of that. I need to hook a power supply to it, and see
what it does.

[ ... ]

T-368? :-)

Probably. I don't remember the model, as I never saw it. I
only heard about it years later.

I used to be a radio teletype operator int he Army and remember that
one well. I know a number of them have ended out in surplus and went
to ham operators. There's a group that meets on 80M that runs AM and
I know at least one of these guys is using one. Draws more power than
my biggest old iron computer!!

Hmm ... how much power did it put into the final?

Don't remember exactly, several hundred watts.

Not a full gallon, then.

And could run keydown all
day long. We used to send some really long teletype traffic and it never
even flinched.

O.K. And that was FSK, so constant broadcast of a pair tones,
flicking back and forth between them.

[ ... ]

Speaking of SUN 3's..... Know anybody who might be interested in a pair
of Deskside pedestal model 3's?

Which ones? The three slot 3/140, or the 12 slot 3/[12]60?
I've got examples of both. And I can't find anyone who wants them.

Don't know, have to go look. I may roll them up here later today.

If they roll, it is the 12-slot ones. You could carry the
three-slot one, and had to put it in a stand to keep it vertical if you
did not want to lay it down.

These are both on wheels. About 3' high less than 1' wide.

O.K. The 12-slot ones then. 3/160, 3/260, etc. The 3/180 and
3/280 were rack-mount versions of the same.

And the Sun3 systems would *not* boot from a drive with the
parity enabled, while the SPARC systems *required* it. :-)

only part I really want is the 9-track.

Hmm ... I've got one here -- and nothing running the full VME
bus still up, so I can't load the interface card for it.

I would assume these are VME. I even have a board floating around here
that adapts VME to the smaller form used in some of the Sparc servers.
(I think they called it 2U or something.)

But I also have a SCSI-Interfaced HP one -- front loading self
threading machine. Not as much fun to watch as two spinning reels, but
it works rather well -- or did when I last used it.

I have a SCSI one that I use quite a bit. The other would be usefull
as Pertec shows up as a different device than the SCSI one.

Pertec was also a front-loading self threading system, IIRC. I
had one of those connected to my first unix box -- the COSMOS

The controller card could move to the VME bus Suns using adaptor
frames to VME.

[ ... ]

Hmm ... You know that they could be upgraded to SPARC based
systems with the 4/??? cards -- still use most of the existing cards,
except for the CPU card itself. And you could replace a lot of big RAM
boards with a single one full of 4Mx9 SIMMs.

Of course, I think that they could not run anything past perhaps
Solaris 2.6 -- if that far up the tree. :-)

This place has abandoned all of that. We used to have a bunch of Sparc
here but I just took the last one (an UltraSparc) home. All the rest
were either given away or trashed. :-(


No ouch, They were just SPARCStation-2's.

Oh -- no great loss. :-)

Which Ultra was it that you got? I've got Ultra-1
Ultra-5, Ultra-10, Ultra-60, the SF-280R, several SB-1000s and one
SB-2000 at present.

Does SUNBlade 100 sound familiar? I think that might be what it is.

O.K. One of the ones which used IDE drives. I have one, with
not enough RAM to really make it happy.

Mine I was running SunOs 4.1.4 for a few years after the other
boxes started to get Solaris 2.x in them.

I never liked Solaris. I think they should have stayed with SunOS.

Well ... there was a tradeoff. On the older machines, I
preferred SunOs 4.1.4 (or SunOs 4.1.1_U1 for the Sun-3 machines), but
for the Ultras, Solaris 10 is quite nice -- especially with zfs in there
giving me a very nice RAID-5 or two on a couple of systems.

zfs could have been used with a BSD based system.

Yes -- if it was written into the kernel.

I just never like
SYSV and I worked with it from the very beginning. And RAID is better
handled in hardware, anyway.

Hmm ... I remember several hardware RAID systems at work which
were rather fragile -- starting with the Sun one which had three sticks
of ten SCA interface drives each. We had to do corrective work every
time power went down for longer than the UPS could supply, and we
typically had that happen two or three times per year -- including with
snakes comitting suicide in the big transformer farms on post. :-)

The only one which we tried which seemed bulletproof was the one
by DEC.

But I have three A-1000/D-1000 boxes -- one converted to JBOD,
the other two hardware RAID -- and the latter need special management
programs in Solaris which are not upgradable to Solaris 10 (and not
available at all for OpenBSD.

But the zfs seems to be *very* reliable on either a Sun Blade
2000 or a Sun Fire 280R.

And the hardware ones are tailored for a specific range of drive
sizes. With zfs, I started with a set of five 18 GB SCA drives (plus a
hot spare or two), got a tray for fibre channel and a similar number of
36 GB drives, and was able to migrate one drive at a time from the
command line -- and when the last new drive was made active, the
capacity of the array suddenly doubled. Same experience replacing the
36 GB drives with 72 GB drives. It was a clean and smooth transition.
When I get a few more 72 GB FC drives, I'll do the same to the array on
the system which I am typing this on.

The Ultra-5 and Ultra-10 machines are running OpenBSD, so they
feel more like SunOs 4.1.x, except that you can have 15 partitions
(exclusive of the whole disk) instead of only seven.

Run BSD at work all the time. Never saw a reason for 15 partitions.

Well ... perhaps with the new 1 TB and 2.5 TB drives. :-)

And the Ultra-5 and Ultra-10 machines also use IDE drives.

But Solaris 10 seems to be faster on the newer Ultras, and
OpenBSD has just barely begun to support more than one CPU on the
UltraSPARC machines.

Oh yes -- I just remembered another couple of machines which I
forgot to mention in the list. Solbourne S4000 and S4000DX. Solbourne
was runing mult-processor on their SPARC systems back when Sun was
claiming that they had to move to Solaris so they *could* run
multi-processor. And for quite a while, I was running a SparcStation 10
with two ROSS double CPU cards in it. Four CPUs -- and to use them I
had to run SunOs 4.1.4 -- Solaris would not touch them because they did
not have on-card cache. :-)

Yet another reason why I thought the move to Solaris was a bad thing.

Well ... not necessary for the stated reason, at least.

I think that machines like the Sun Blade [12]000 and Sun Fire
280R are not going to work as well on any available BSD as they do on
Solaris 10.

[ ... ]

But -- later versions of Windows keeps everything possible
hidden -- especially how to run things like the WinModems (which remind
me of the floppy drives used in the Apple-II. :-)

I saw winmodems supported under Linux so it couldn't have been all that
much of a secret.

Two possible ways:

1) Lots of digging to figure out all the hardware functions in
the cards.

2) Accept a "blob" -- a pre-compiled lump which you have to run
inside a wrapper to adapt it to your OS -- and you have no
access to the source to do a security audit on it. This is why
OpenBSD will not support certain cards, including NVidea,
because they can't do a proper security audit on the source
code. The vendors absolutely *refuse* to release the hardware
documentation so the open-source types can write their own
drivers from scratch. And they require an excessively
restrictive NDA for using the blob.

A lot of linux distributions have signed such NDAs -- and have
been hit by security holes which they could not detect without
the source to analyze. OpenBSD *refuses* blobs.

The big question was why would you want to give up
CPU time for soemthing that was handled so well in hardware within the
modem itself.

I don't -- but if I want to use a modem from a laptop infected
with WinModems, then I may be stuck -- unless I carry a separate modem
as well. (Hmm ... the PCMCIA modem cards -- are they true stand-alone
modems, or are they really WinModems?

I stayed with external modems right up until I shut the
last one down. (Which was last Thursday when I finally pulled the plug
on the departments dial-up IP machine.)

And I've *never* used a WinModem on any hardware which I control.

Now, if the PC had started with the 6809 and OS-9 it would have
been a *much* more attractive machine to me. :-)

6809 was already deadend technology by that point.

It was more powerful than the 8088 which the PC started with.
And it had as much a claim to being a 16-bit processor as the 8088 did. :-)

But suffered fromt he same problem that IBM found with their idea to use
the M68K. :-)


Yes. Came from the same vendor as the M68K.

So they did not like the Vendor? Nothing to do with the


Email: <dnichols@xxxxxxxxxxx> | Voice (all times): (703) 938-4564
(too) near Washington D.C. |
--- Black Holes are where God is dividing by zero ---