eth0 dies at high cpu load

[ross8653]

i formated my crappy server that i had running for about a year with no problems and was giving it to a friend to use. Installed everything i needed physically (monitor/keyboard attached) and put it at the location, turned it on, recompiled the kernel once there with a keyboard and monitor and we left.

We went home and it was up and working, i tried an emerge sync and about 3 pages in we lost the connection, figured it was a hickup in my network but the server wouldnt respond. I tried pinging it from other locations same thing.
I would get maybe 30% of the pings back all at 3000+ ms times. Sometimes i would get back 15 in a row, sometimes i wouldnt get back any for a minute. It remains unresponsive for about 3-5 minutes if you loose a connection doing something (not in screen)

The server does not respond on any port (installed apache to test) until the cpu is done with it's task. I tried putting a lot of stuff in screen (emerge -puD world) and it would remain unresponsive until it finished, but would always finish it's job. Also at first i thought it was resetting but it's not by checking uptime.

so i asked some friends to try to determine where the problem is, I've used the same hardware in the same box before, all i did was rip out a sound card (before format). i'm not sure what to post because i dont know where the problem is, any help/idea what to post would be nice

oh yeah "dmesg" shows nothing when this happens

[ross8653]

looked harder at ifconfig and noticed i had a couple thousand errors

so the lesson is to not use nic's that you find in the dumpster