Michael G. Schabert wrote:

> >> BSD reports numerous "silo overflows" (between 5 and 40) every few
> >> minutes under network traffic.
> >>
> >> I've got no idea what these are!
> >
> >They are when the modem was sending information too quickly to the
> >computer, and some information got lost. David's suggestion of lowering
> >the speed, to say 38400, is a good one. 57600 is just on the edge of what
> >the '030's can do. :-(

This worked quite well - I reduced the port speed to 38400 and silo overflows
were markedly reduced, I was still seeing a couple though.  This is a pity as
I regularly get 50kbps connections.
The pppd manual pages suggest that netBSD should be able to handle any serial
speed, but I have found this to be incorrect, nothing in between 38400 and
57600 work.



>
>
> I'd like to amend that by saying that 57600 is on the edge of what MacBSD
> can do on an '030. The Mac itself has no trouble whatsoever keeping up. I
> used to use my Mac SE to download at 57600 with no trouble under MacOS. The
> hardware can keep up just fine, so it's something under MacBSD's interrupt
> scheme that's not keeping up. Remember, that Mac SE is an 8 Mhz 68000.

I was acheiving nice 56k d/ls under macos on the SE/30.  The reason I tried
BSD on this box is that IPnetRouter and Vicom internet gateway failed under
MacOS, d/ls to the SE/30 were fine but clients got the first few packets then
just stopped.


>
>
> >> The SE/30 has an Asante ethernet card - not sure of model etc.....
> >>
> >> Any ideas out there???
> >>
> >> One other question, Paul Goyette's ppp kit provides dial-on-demand for
> >> those that have static IPs, but he suggests that this is impossible for
> >> dynamically assigned IPs.   Is this still the case?
> >
> >Yes, and it will always be the case. The problem is that dial on demand
> >keeps network connections open even when the modem gets disconnected. It
> >works as when the modem re-dials, it gets the same IP address it had
> >before. With dynamic addresses, that's not so. You'll get a different
> >address when you redial. :-(
>
> Yes, but why can't it be used just like having FreePPP set to allow
> applications to connect? That works fine under MacOS. I understand that
> socket connections couldn't be maintained across sessions, but you should
> be able to rig it to just re-ppp-up everytime we want to send something out
> & then time-out the connection after "n" minutes of inactivity. As long as
> you ppp-down upon that timeout, I dunno why there'd be a problem.

This is really what I was after - I'm not too worried about broken
connections, I'm just too lazy to type the command!

--
+------------------------------------------------------+
Hugh Harris.
School of Chemistry
UNSW
h.harris@unsw.edu.au
+------------------------------------------------------+