>
> When I try to ping local nodes, I often see ping times in excess of
> 30 seconds. Channel activity is very low, yet I send out a ping ICMP
> packet and don't get a response for 30-40 and sometime over 60 seconds.
>
> These are local nodes.. Direct routes.
>
<clip>
> Does TCP/IP packet *REALLY* work at 1200 baud? I have been struggling
> with this stuff for going on 2 years now and have yet to get anything
> that even remotely approaches what I would consider acceptable operation.
> --
> Pete Rossi - WA3NNA
> qglx@superusers.dk FLAGS (\Recent \Seen))
That seems excessively slow. TCP/IP does work at 1200. Although it is
slow, it should be much faster than what you are reporting. I'm assuming
a direct AX.25 connection is much faster? (If not ... there's the
problem).
Be sure to set the IRTT at a high enough value for packet. I use:
route add vhf.wb2psi irtt 12000 ax0
That value is probably excessively high, but after the route is used, the
value is adjusted downward.
Also, datagram mode is much faster than virtual circuit. The only reason
I can think of that you might want to use VC is if you use TCP/IP over a
huge NETROM net. In that case, imagine an IP frame travelling hundreds
of miles over NETROM, and then getting dropped on the last 5-mile AX.25
link because you're using datagram. In that particular and rare case,
you're better off sacrificing local tcp/ip speed by using VC.
If you carefully study the "listen -at" output, you might see clues as to
why things are so slow.
I get telnet connections over 1200 to a local node in 5 seconds. After
connection, turn-around time is more like 3 seconds. Ping time averages
about 2.5 seconds.
-- Dave | Email: terhi.victor@logonet.com Brown | Packet: N2RJT@WB2PSI.#WNY.NY.USA.NOAM