[Summary] Remote boot & remote diag on a SUN

From: tonytran@leland.stanford.edu
Date: Mon Oct 07 1991 - 21:41:15 CDT


My original question:

>Has anybody implemented remote booting /remote diag on a SUN?
>I need to be able to provide support to our critical servers
>remotely from home (such as booting, running diagnostics etc...)
>I've seen remote booting/diag done on VAX systems by DEC Tech
>Support from Colorado to assist their customers with hardware
>trouble shooting (using a special DEC modem), but I am wondering
>if it can be done on a SUN.

Many people reported they have been using "console server" (or
global console) to allow remote booting/diagnostics with a lot of
success. It allows you to remotely boot the server, boot the diagnostics ..
In addition, you can save a lot of footprint in the file server room
by not using several console terminals.

You basically connect multiple console serial ports to a dedicated
server with a ALM2 board (or similar config).

The problem/restrictions with this set up are:

1. The console should be near the fileservers
2. No scroll back facility
3. There is no overview
4. If something happened to the console server (power outage,
   unplug the console etc ...) everthing goes down.

Andrew Luebker <aahvdl@eye.psych.umn.edu> pointed out an old
issue of Sunspots where one can construct a special circuit
to prevent this from happening. All you need to do is to connect
a 4.7k resistor between pin 3 and 25 on the fileserver side as follows:

 
            tty |
            port | 2 ---------------- \
            on sun | \
                    | 3 ------x--------- > to customers equipment
                    | | /
                    | 7 ---------------- /
                    | |
                    | |
                    | | |
                    | | | 4.7k resistor, 1/4 watt
                    | | |
                    | |
                    | 25 -----x
                    |

        Be sure the resistor is attached at the Sun CPU connector, not the
        terminal end or switchbox end of the cable....

A better circuit like below will prevent all the servers from going down
even if the console server goes down:

[From: Malcolm Harper <mkh%prg.oxford.ac.uk@nss.cs.ucl.ac.uk>]

To terminal To Sun

pin 3 <----:-----------------------------------------------------------< pin 2
           | ________ _________________ ____________
           | | | | | | 2200uf 16v |
           :---| 470ohm |----| <- IN4002 diode |---:---| capacitor |--- pin 7
               |________| |_________________| | |____________|
                ________ _________________ | -ve +ve
               | | | | |
           :---| 470ohm |----| <- IN4002 diode |---:
           | |________| |_________________| | Types of diodes and
           | ________ ___________________ / transistor, and values
           | | | b| |/c of capacitor, are not
pin 2 >----+---| 4K7ohm |----| BC212L transistor | critical.
           | |________| |___________________|\e 5,6,8,20
           | _________________ \ wired together
           | | | |
           :-----------------| IN4002 diode -> |---:-------------------> pin 3
           | |_________________|
           | ________ _________________
           | | | | |
           :---| 22Kohm |----| IN4002 diode -> |----------------------- pin 25
               |________| |_________________|

pin 7 ------------------------------------------------------------------ pin 7

They use only three wire connections, so connect pins 5, 6, 8 and 20
together at the Sun end, and appropriate control line connections at the
terminal end. This circuit is permanently connected to the Sun end; any
disconnection must be at the terminal end.

If the terminal is unplugged, the transistor is turned on by the negative
reference voltage present at pin 25 of the Sun serial port, and hence
pulls pin 3 of the Sun port negative. This causes the Sun to believe
there is still a terminal plugged in.

Acknowledgements to Andrew Newman and Paul Williams who designed and built it.

The only thing you have to be real careful about this "console server"
set up is to use your "tip" command carefully.

If you remote log in onto the console server, and you need to halt
the system, use ~~#. Using ~# will halt the console server (which is
a NO NO).

Example:
                # rlogin console_server
                # tip starburst
                do something terrible and hang the system
                ~~# (Halts starburst)
>b (boots the system)
                ~~. (dis-connects the tip session)
remember that the second tilde will escape the rlogin shell and let
your ~. get into the tip session.

Dan Butzer <butzer@cis.ohio-state.edu> reported that he had a similar
setup at Ohio State. The console server is set up so that it is
also able to boot any of his other servers diskless. This combines
with the 3 hour UPS seem to be a valuable asset to any remote Sys Admin.
Tom Fines <fine@cis.ohio-state.edu> also wrote a program called "console"
which specifically manipulates the server console terminal remotely.

Geert Jan de Groot <geertj@ica.philips.nl> improved the console
server by using a dedicated PC, added a few multi-port serial-I/O boards
in it, and created a multi-port terminal emulator. With F1-F9, he can
switch between each server console port. With F10, he can switch to a
'summary' screen, on which all output is displayed (no output means all
is safe). Every screen has a scroll-back feature.

Michael S. Maiten <msm@energetic.com> mentioned that if you attach
a 9600 baud modem to ttya and make sure that the switch on the
back of the CPU is set to DIAG (the labels are NORM and DIAG), you
will be able to monitor the selftests and run MONITOR diagnostics from
the serial line. (I haven't tried this yet).

with regard to the remote diagnostics, Kevin Sheehan <kevins@aus.sun.com
(Kevin Sheehan {Consulting Poster Child}) mentioned that the DiagExec kernel
does have support for remote diagnostics. It should work fine from
frame buffer, terminal, modem, and network access. Under SunDiagnostic
Executive User's Guide version 1.2 page 42, there was some mention
of remote execution (using a program called "netcon") of the Exec program
over the Ethernet. Unfortunately, Sun Diag Executive is not included
free with the standard SUNOS. I haven't tried this program myself.

Thanks to the following people for comments and suggestions:

earl@division.cs.columbia.edu <earl smith>
geertj@ica.philips.nl (Geert Jan de Groot)
pomeranz@isis.dccs.upenn.edu (Hal Pomeranz)
Dan Butzer <butzer@cis.ohio-state.edu>
Andrew Luebker <aahvdl@eye.psych.umn.edu>
kevins@aus.sun.com (Kevin Sheehan {Consulting Poster Child})
Timothy G. Smith <tgsmith@spdev.east.Sun.COM>
Michael S. Maiten <msm@energetic.com>
Chris cc@dcs.ed.ac.uk

Tony Tran



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:06:18 CDT