On Tue, Jul 27, 1999 at 09:24:03AM +0200, Patrick Kormann wrote:
> Hi!
>
> I guess this doesn't strictly belong to the squid mailinglist. On the other
> hand, this only happens on each of our 4 linux proxies, not on any other
> linux machine.
> Regurarly, out of blue sky, we get the following problem:
> Jul 27 04:09:29 proxy1 kernel: eth0: Transmit timed out: status 0050 0080
> at 175049833/175049848 command 000c0000.
> Jul 27 04:09:29 proxy1 kernel: eth0: Trying to restart the transmitter...
> Jul 27 04:09:34 proxy1 kernel: eth0: Transmit timed out: status 0050 0090
> at 175049833/175049848 command 000c0000.
> Jul 27 04:09:34 proxy1 kernel: eth0: Trying to restart the transmitter...
> Jul 27 04:09:39 proxy1 kernel: eth0: Transmit timed out: status 0050 0090
> at 175049833/175049848 command 000c0000.
> Jul 27 04:09:39 proxy1 kernel: eth0: Trying to restart the transmitter...
> Jul 27 04:09:44 proxy1 kernel: eth0: Transmit timed out: status 0050 0090
> at 175049833/175049848 command 000c0000.
> Jul 27 04:09:44 proxy1 kernel: eth0: Trying to restart the transmitter...
>
You have intel eepro100s? Their driver/hardware is buggy under Linux and
high load (eg lots of UDP), I found that the latest kernel 2.2.10 and
sticking this in conf.modules fixed it most of the time:
alias eth0 eepro100
options eepro100 multicast_filter_limit=3
The only other thing left to you is to have a logger that detects that
kind of error in the logs and reboots. Or replace the intel cards with
something else.
tom@interact.net.au
Received on Tue Jul 27 1999 - 02:25:46 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:47:34 MST