Re: More BCP: revenge of RS232 and CLIs

["Joel N. Weber II" <ietf-opsec@joelweber.com>]

> > I'm a little skeptical that being able to change the baud rate is
> > really beneficial.  It seems like if the console port is always at a
> > fixed speed, there's less worrying about whether I'm seeing nothing on
> > the console port because this sparcstation 1 my friend gave me is
> > actually dead, or if it's just at a weird baud rate for some strange
> > reason.
>
> You've had that problem too ?

yep.  (along with the ``do I have the wrong number of null modems in
the chain?'' question.  and the annoyance of netbsd apparently using
different parity settings than the firmware.)

> >  And I've generally found in-band management to be a good
> > option for day to day work when the extra speed can be nice.
>
> Sure, but we're taking about this because there are times
> when you positivly, absolutly have to manage the device
> and in-band ain't working for some reason.

Right.  And for such cases, you can probably tolerate a little bit of
slowness in the line speed.  (And you might even appreciate saving
time playing the baud rate guessing game.)

Most vendors seem to let you change the baud rate, and most users seem
to have the good sense to leave it alone.  I don't think opsec is
going to set new standards here; I don't expect Cisco to rip a feature
out because opsec writes a draft saying that it's bad.

We might consider listing 9600/8/n/1 in the examples of the
appropriate section as an example of a good default that several
vendors support.

> > I think we should consider mandating that it MUST be possible to
> > configure a network device by plugging its serial port into a terminal
> > whose display understands the 95 printing ASCII characters, carriage
> > return, newline, bell, and backspace.
>
> Ugg.  In one sense you're right.  In another, I'm not sure people
> with non-latin alphabets (pictographs, ideaographs) would like
> it if we did that.   I really don't want to get into that one.

What do real world Cisco customers in Japan currently do about this?

> >  A device MAY send terminal
> > control commands that assume a more advanced terminal, such as a
> > vt100, in response to the user sending commands to the device, such as
> > control characters to edit the command line, as long as it is still
> > possible to completely configure the device from a terminal which does
> > not support these features.
>
> Come on, say it with me "ASR-33"...."Tel-e-TYpe".

I'm afraid I'm too young.  8-)

(Yeah, yeah, ``kids these days...'')