It won't work, in most instances.  The Cisco web page is Enormously
misleading, and any sites that echo the Cisco page are just as
misleading.  (Celeste's page is the least misleading of the ones
I've seen so far)

The resistor between pin 3 and 25 method originated in an e-mail message
sent to Sun Support several years ago (ca. 1992 or 1993) by a Sun customer.
It was something that the customer tried and enjoyed some success with.
Sun support has given the details of the method out from time to time,
but they have never promised that it would work.  Apparently some folks
at Cisco got hold of the method and re-published it on their web site.

Here are the three most important facts about the resistor method,
followed by a more technical explanation of its effects:

  1) It was never designed to *block* break signals sent by a terminal,
     computer, or terminal server connected to the Sun machine.

  2) It was never *possible* for the resistor to block break signals
     transmitted from the terminal/computer/terminal-server.

  3) It was only intended to suppress glitches produced by plugging or
     unplugging an RS232 cable, so those glitches would not look like
     a break signal to the Sun.

Unplugging or plugging an RS232 cable causes the pins to make and break
contact a few times as you slide the connectors together.  This can
cause the signal on the Sun's RxD (Receive Data) pin to jump from 0V
to -5V and back as contact is made and broken.  Capacitance and/or
inductance in the cable can cause the voltage jump from -5V back to
0V to overshoot and produce a glitch of positive voltage.  If the cable
conditions are right, the glitch can rise above +3.5V, and can last
longer than a couple thousandths of a second (milliseconds).  This kind
of glitch looks exactly like a break signal to the Sun machine.

Glitches caused by cable capacitance or inductance like that usually
don't have a lot of energy behind them.  The resistor we're talking about
was designed to drain off enough energy so the cable-induced glitches
can't rise high enough or last long enough to look like a break signal
to the Sun.  So the resistor can sometimes help when unplugging or
plugging the console cable causes your Sun to halt.

However, it can't help the other kinds of break signals.  These are
often caused by powering the attached terminal/computer/terminal-server
down or up.  When this happens, the RS232 driver circuitry in the attached
device forces the voltage on the pin to +3.5V (or higher), and backs it
up with a LOT of energy from its power supply.  Since this kind of glitch
has plenty of energy behind it, the resistor can't drain enough away to
prevent the signal from looking like a break to the Sun.  The resistor
won't help.  (it was never intended to help this situation anyway)

The only solutions when you have equipment that glitches its outputs like
that are to (a) re-design the equipment to clamp the output voltage to
zero volts when the power is cycled, (b) replace the equipment with other
equipment that doesn't let its outputs glitch, or (c) reconfigure the Sun
to not halt on the break signal.


Infodoc 21258 on sunsolve.sun.com describes the patches and procedure
to make Solaris 2.6 and later halt on an "alternate" abort sequence
instead of the RS232 break signal.

     
-Greg Andrews