[clug] random ethernet shutdown

Darren Freeman daz111 at rsphysse.anu.edu.au
Thu Oct 16 13:56:27 EST 2003


Dear List,

I have a brand new Asus A7N8X Deluxe motherboard (N-Force 2 chipset).

There's two builtin ethernet ports, the N-Force one and a 3-com one. I'm
using only one, with the driver being 3c90x.

I'm testing my system stability because it kept locking up with the case
off. Uptime is about 3 days now, and the machine is a server so it needs
to be reasonably healthy.

Starting yesterday I was getting occasional failure of the network. The
number of tx errors was climbing steadily and the kernel message:

Oct 16 13:48:48 host kernel: NETDEV WATCHDOG: eth0: transmit timed out
Oct 16 13:48:53 host last message repeated 10 times

kept coming up at the same time. When this problem occurs the mouse is
unresponsive for about 100ms once every second.

The solution is to drop eth0 and bring it up again (also add the gateway
again to the routing tables).

But as of an hour ago I find that this is happening about once a minute.
Basically every time I go to access the network I have to run a little
script to do the workaround. It's obviously no good for a server
machine!

Anyone heard of anything similar? Is it because the motherboard is so
new? I found that this mobo has >15 interrupts and any card configured
on a higher IRQ number actually panics the machine when the kernel
activates interrupts at boot time. So I removed the offending card.

Maybe N-Force2 is just too new?

Darren




More information about the linux mailing list