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

serial console

Here's what the Western Scientific guys had to say about the console issues we experienced.

I know that those serial settings are the default for our sun hardware. Has anyone cranked up the baud rate on the serial ports to see if we get better interaction with the intel hardware?


Begin forwarded message:

From: "Michael R. Hanulec" <michael.hanulec@xxxxxxx>
Date: June 7, 2004 6:09:55 PM EDT
To: sweda@xxxxxxxxx, David Jensen <dave.jensen@xxxxxxx>
Cc: support@xxxxxxx
Subject: Re: Fw: eval/order


The requested configuration for the Serial Console of:

"Units must come preconfigured for serial console (9600, N E) usage so we
can install completely without attaching a monitor and keyboard."

is what is causing the serial port i/o delays. This number should be
bumped up higher, ideally to 115200, in order to kill they various delays
you are seeing before getting into the bios configuration screen and
during the Adaptec configuration.

By default we 'disable' the Quick Boot just so users can see the proper
amount of RAM installed. There is no reason it can't be re-enabled BUT I
would HIGHLY recommend you reevaluate the need for a Serial Connection
speed of 9600.

As for the Parity Check, this can be disabled due to the ECC RAM.

Please let us know if you have any other questions about the units.


On Jun 1, 2004, at 2:16 AM, Marcus Watts wrote:

At Sean's request, I experimented a bit more with bios settings
on the western scientific 3u machine.

I had previous tried out:
        Quick Boot              enabled -> disabled
        bootup numlock          on -> off
        bootup cpu speed        high -> low
        parity check            disabled -> enabled
of these,
bootup numlock doesn't matter if there's no keyboard
plugged in.

bootup cpu speed high -> low makes no difference I can measure.
Elapsed time in boot is identical.

I don't know if enabling parity check is a good thing or not,
given that these machines should have ECC memory and might
report memory errors differently.  This would be something
to ask the western scientific folks.  Linux probably sets
this how it pleases once it boots, so this probably only affects
memory while booting.

Quick boot seems to be the interesting option. With quick boot =
disabled, it takes about 9 minutes total to boot unattended (from reset
to login:) which includes a 6 minute window in which to send a <DEL>
and get into the bios setup menus. With quick boot = enabled, well,
the whole boot process only takes about 3 minutes, including about
20-22 seconds from initial system reset to the initial bios screen
redraw. I was able to get into setup by sending a repeating <DEL> for
the 10 seconds before the initial bios screen appears. Once the redraw
happens, it's definitely too late. One more gotcha which is true even
without quick boot is that even if the machine is going to go into
setup mode, there's no obvious feedback. Instead it's necessary to
first wait for the adaptec bios initialization, which takes about 20
seconds, before the bios setup screen appears. Without quick boot,
is also a "8 seconds until boot" screen display where it looks like
it's going to boot lilo, before the setup screen appears.
With quick boot = disabled, for attended operation, you can
short-circuit the 6 minute memory testing process (and the 8 second
boot delay) by sending a <ESC>.

So, um, if you're using a digi serial mux on solaris, I believe
you'll want
quick boot = disabled
and possibly you might want (ask western scientific):
parity check = enabled

If you're using a laptop or another serial connection that puts you
right next to the reset switch and doesn't disconnect when carrier
drops, don't mind unobvious behavior, and unattended boot speed is
important, then you might want "quick boot = enabled".