> Now I am really confused.
> After more testing I have found that sending a file via scp or cat'ing through
>sendmail works like a champ if I go to a machine outside of the network. But
>seems to be a problem for the same machine when trying to go to a server
>connected to the same switch.
> OK.. maybe its the switch you say. Me too. Until I just came from putting the
>machines on a different switch and still having the same problem. It's also a
>completly different make and manufacture of switch.
Duplex problems often show up as weird performance problems related to the
speed of data transfer, size of the packets, or round-trip time to the
destination. This happens because during a duplex mismatch, the full duplex
side doesn't retransmit on a collision and if the half duplex side sees
traffic on the receive side while it is transmitting it thinks it's a
collision and stops sending the packet. So for situations (small packets or
larger round-trip latency) where the ACK is unlikely to collide with the
transmit, things appear to work okay. In other situations where the ACK
collides, performance crawls.
> Also one other weird question. What is the real difference between a cable
>with 2 pairs and a cable with 4 pairs were 10/100 ethernet is concerned. On
>another server that was using a SMC/DEC card I found it would go nuts when it
>had a 2 pair cable, but worked Ok with a 4 pair cable. From everything I can
>tell, 10/100 ethernet should not care abt the extra 2 pairs.
I don't know about the affects of 2 pair vs 4 pair, but one cable related
thing I do know will have an affect is stranded vs solid core wire. I've
always had problems using solid core wire for any significant length (more
than 6 feet) with the Pro/100. Stranded wire on the other hand has never given
me a problem even for runs of >100 feet.
-DG
David Greenman
Co-founder, The FreeBSD Project - http://www.freebsd.org
President, TeraSolutions, Inc. - http://www.terasolutions.com
Pave the road of life with opportunities.
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-net" in the body of the message
