> a second PC (directly connected to the switch): short after reboot I got a
> large file via ftp; the performance was ok (~950kB/s) but the collision
> count was *higher* than the TX count (>100% ;-)!
>
> I thought this may happen because the other machine transmit (data) packets
> very fast and my transmitter isn't able to get the wire for the ACKs?
> Is it possible to saturate a 10BaseT/half duplex wire with one ftp transfer?
> Can someone with deeper network knowledge comment on this?
It's easy to saturate even 100BaseTx with a single TCP/IP stream.
Note that some drivers count 1-15 collisions per packet, others hardware
reports only "at least one collision occured" and thus only increments the
collision count by one.
One of my planned changes was to put in full network statistics, with
seperate fields for
Tx deferred
Tx single collision events.
Tx multiple collision (>1) events
[[ Insert flames here about incompatibly changing the format of
/proc/net/dev, and another flame about failing to adding all of the relevant
fields a change was made to the statistics structure. I designed the format
of /proc/net/dev for a reason. Once again it's too easy to add or change
stuff to the official, and almost impossible to keep a design clean. ]]
Donald Becker becker@cesdis.gsfc.nasa.gov
USRA-CESDIS, Center of Excellence in Space Data and Information Sciences.
Code 930.5, Goddard Space Flight Center, Greenbelt, MD. 20771
301-286-0882 http://cesdis.gsfc.nasa.gov/pub/people/becker/whoiam.html
-
To unsubscribe from this list: send the line "unsubscribe linux-net" in
the body of a message to majordomo@vger.rutgers.edu