[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Need "console switch" - any ideas?




 
> John Geletej wrote:
>> I find myself in need of a "box" that will allow several UNIX
>> workstations/servers to be controlled at a single
>> keyboard/mouse/monitor. My primary need is for SGI equipment, but I may
>> need to incorporate a couple of Sun SPARCStations into the mix as well.
> 
> My preferred way of doing this is to use a serial console server, as
> follows:
> 
> Stick a 32 or 64 port serial card into a PC (Cyclades makes good ones),
> run Linux or *BSD on the PC, and install Paul Vixie's excellent rtty
> package to serve up the serial ports (available from ftp.vix.com).

	rtty looks interesting --- but it won't compile
	under my copy of Linux:

/usr/local/src/rtty/rtty.c:218: undefined reference to `FD_ZERO'
/usr/local/src/rtty/rtty.c:219: undefined reference to `FD_SET'
/usr/local/src/rtty/rtty.c:220: undefined reference to `FD_SET'
/usr/local/src/rtty/rtty.c:242: undefined reference to `FD_ISSET'
/usr/local/src/rtty/rtty.c:252: undefined reference to `FD_ISSET'

	.... I see that macros of the form __FD_* are defined in 
	/usr/include/gnu/types.h so creating a "mytypes.h" with
	a set of #defines (which merely fill FD_* with __FD_*)
	seems to get it to compile.

	I don't have time to test that "solution" right now
	--- but Paul does say that he only expects it to work on
	BSD in his notes.  

	The README contains this gem:

``
Since there's no documentation, there's no harm in telling you that the
"console" script just runs "rtty" with some obscure arguments to grease
its skids.  While in "rtty", a "~" after a carriage return (\r or 0x0D)
is magic; "~?" will explain the magic to you somewhat tersely.  The source
code will explain it in more detail.
''

	... so anyone that's planning on trying this under Linux
	may need to to a bit of extra work.
 
> The only downside of using serial consoles with Suns is that Suns have
> the annoying habit of dropping into the ROM monitor when power to the
> console server is lost (they interpret this as a "break" condition).
> As best I can tell this is the case for any serial console hooked up
> to a Sun, even a dumb terminal.
	
	It is.  It's happened to me many times.  (Seems silly to
	*have* to put your rarely used serial console terminal
	on your UPS --- but that's Sun's idea of a joke on 
	all of us).  What were those little Stanford boys thinking?

 
> The workaround is to never shut down the power to the console server
> (reboot is fine), and if you absolutely must, disconnect all the
> serial cables first.  Older Suns also sometimes drop into the monitor
> if you just disconnect the cable, but we've found that this can be
> prevented by putting a resistor across the right pins.  Actually one
> can prevent break conditions altogether by putting a resistor across a
> certain set of pins, but then you can't drop into the boot monitor
> anymore, so that's not desireable.  If anyone has an optimal solution
> to this problem, I'd love to hear it...
 
> - Jürgen

	I heard a rumor that one can cut a trace on older
	Sun motherboards to disable this "feature" --- and
	I've heard other rumors that there might be an OpenBoot
	(FCODE) nvram setting that can be tweaked.  However,
	I don't use headless Suns often enough to worry about it.

--
Jim Dennis  (800) 938-4078		consulting@starshine.org
Proprietor, Starshine Technical Services:  http://www.starshine.org