In message <200112252005.fBPK5Qm07295@grok.beer.org>Herb Peyerl writes
>"Thomas Michael Wanka" <Tom@Wanka.at>  wrote:
> > I have a Supermicro 370DLE that (it claims so) can be controlled on 
> > BIOS level through a serial port (have not tested it until now). I 
> > got no response from either the Bios or board manufacturer if other 
> > models support this too. 
>
>The problem with a motherboard that supports serial console is that
>it solves only part of the problem.
>
>You don't have a hard reset other than using a power-switcher,
>so you still have to buy hardware to get remote maintainability,

Depends which flavour of "serial console" you get.

The Tyan motherboards (also some Dells) provide "serial console" by
having a background loop in the BIOS, continually reading the VGA
textmode console area and writing it to a serial port.  This yields
considerable flicker and fairly slow update rates -- POST memory tests
can take a *long* time.

For this flavour, Herb is correct.

Several `Intel' server motherboards (N440BX, L440GX+, STL2, ...)  have a
microcontroller on the motherboard, doing the screen reads and smarter
cursor updating. (the output loks just like a BIOS screen) which can
force a remote reset independently of the main CPU(s) The protocol
isn't documented; one has to use something like the VA-Linux VACM
package (yuck!!) for wihch a takes two serial cables -- one for an 
post-boot OS console, another for the remote-frontpanel stuff.

PC-weasel setup is a no-brainer; the other stuff isn't.

For space-challenged racks -- anything smaller than 4U-- I'd probably
with an Intel server motherboard with ServerWorks, rather than pc-weasel.
The Weasel is an easier solution, if there's space inside the server.