SUMMARY: serial console hangs whole system

Hi everybody !

Sorry for the late summary, but I got tons of work to do at the moment (who
hasn't ??)

Anyway, it doesn't look so unfamiliar that a serial console could hang a
whole server. 
Here is what other people told me:

 If you have the eeprom settings such that the hardware
flow control on the serial port is paid attention to AND the connector
either a) has those pins unwired so they float or b) the connector has
the hardware flow control pulled down so it is saying "stop sending
data" then the machine will appear to hang, waiting for the "go ahead"
to send data again.  You can set the tty settings in the to ignore CD
and DTR, see if that helps.

Brett Lymn
---

Hello,

you (and I) might assume, that if a keyboard and monitor is connected to a 
machine these should be used as system console and the serial port A should 
be usable as a general purpose serial IO. But the OBP at least sometimes 
tries to switch the console to ttya after detecting a BREAK character.
The speed is set to 9600 baud, regardless of the eeprom settings, and the
port 
is very sensitive in detecting a BREAK: sending a simple RETURN with a 300 
baud or slower terminal will do, and of coarse a malfunctioning terminal.

With the eeprom settings shown in the attachment a Blade 1000 is sensitive 
only during the boot procedure. What are the eeprom settings (especially 
input-device, output-device) of your machine?

The attachment is part of a discussion with SUN (Subject: "Serial port a ist

des Teufels")

Regards, Martin Huber
---

It shouldn't but there are a lot of signals that can be sent via a
serial console...one of them being a break signal but that would cause
the OK prompt.....I am guessing here but I think that the console is
screwed up and sending incorrect signals to the server.

Marco Greene 
---
Hi Gabel,

Sun machines are very sensitive in their serial ports for some reason.  
Odd things like a line buffer connected to a terminal server filling up 
can cause a machine to halt.  Cisco's older version of their 2511/4 
terminal servers would send break signals to all Sun machines connected to 
them when powered down or rebooted and halt all the machines, coming back 
up when the terminal server comes back up.  

It's quite madenning but that's how they work.  I've seen more then a few 
issues caused simply from serial port problems/misconfigurations/crossed 
wires.

Tony Bourke
---


thanks to everybody !!

cu
may

martin.gabel@ffm3.siemens.de
_______________________________________________
sunmanagers mailing list
sunmanagers@sunmanagers.org
http://www.sunmanagers.org/mailman/listinfo/sunmanagers