Is it unusual for a nic that was working 100baseT for 4 months to suddenly
develop packet loss? Nic is identified by lspci -v as "3Com Corporation
3c905B 100BaseTX [Cyclone] (rev 30)" using 3c59x module (kernel
2.4.20-4GB). It was Linux2/eth0 in following scenario 1 since May
(EFSP42 is switch/printserver):
pppoe-eth1/Linux2/eth0----EFSP42----eth0/Linux1/eth1----WAP wlan0
I noticed occasional internet lag on wireless laptop (wlan0), but thought
it was just internet worms (I stopped logging dropped packets due to
rapidly growing logs). But when things recently choked, I discovered
20-100% packet loss between Linux1 and Linux2 (and frame errors on Linux2
eth0). Linux1 could ping EFSP42 without problems, but sometimes could not
even get arp response from Linux2. Linux2 could ping Linux1, but with
varying packet loss.
However, with following scenario 2, there was NO packet loss between
wlan0 and Linux2 (maybe due to lower 11mbps max wireless speed,
although, WAP ethernet is 100baseT-FD):
pppoe-eth1/Linux2/eth0----WAP wlan0
Back to scenario 1, I swapped cables and modules on so the 3Com was
connected to 10baseT-HD DSL modem as eth1/Linux2 (both nics connected to
EFSP42 are now Dlink DFE530TX+ 8138too), and everything is now working
fine. There were 3 initial carrier errors on 3Com connected to DSL modem,
but no errors since, no packet loss between wlan0 and ISP's gateway, and
no lag.
PS: Linksys WAP11 v2.2 ethernet port is incompatible with Linksys EFSP42
ethernet ports (WAP11 Link LED would blink constantly). I did use proper
cables for WAP11 (patch cable to switch, or xover to PC nic) I wonder if
EFSP42 was also a problem for 3Com nic. The Dlink DFE530TX+ nics using
8138too have no problems with the EFSP42 switch/printserver.
--
David Efflandt - All spam ignored
http://www.de-srv.com/
http://www.autox.chicago.il.us/ http://www.berniesfloral.net/
http://cgi-help.virtualave.net/ http://hammer.prohosting.com/~cgi-wiz/