Eskimo North


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Strange Network Problem fixed




     We had some strange network problem where the router kept taking the
100-base-T connection to the switch out of service for a couple of seconds then
putting it back in, sometimes thousands of times a day for the last couple of
weeks.

     This was accompanied by indications of carrier loss, CRC errors, and
framing errors.

     I replaced the 100-base-T card, the cable between the router and the
switch, and tried different ports on the switch all to no avail.

     This problem turned out to be caused by a new machine we were getting
setup for news and for virtual servers for handling mail for wholesale
customers.

     The machine was installed with a 2.4.20 kernel, there are nasty bugs in
the HME ethernet driver for the 2.4.x kernels which Aaron thought a one-line
patch fixed.  Although the patch kept the machine from going down, it didn't
stop it from screwing up the local LAN.

     I had experienced this problem earlier when I tried 2.4.x on another
machine but did not suspect it this time because I thought this machine had
been installed with a 2.2.x kernel.

     We have not had a single instance of the router bouncing the 100-base-T
port since we shut the troublesome server down.

     At any rate, sorry for the random pauses, random servers being
unreachable, etc, but the troublesome machine is now shut down until it can be
resinstalled.