[Gllug] NETDEV WATCHDOG: eth0: transmit timed out

[Steven]

Hi everybody,

I'd like your opinion/advise on following error (2 occurences retrieved
from syslog):


Feb 11 18:18:55 stsriscpc kernel: NETDEV WATCHDOG: eth0: transmit timed
out
Feb 11 18:18:55 stsriscpc kernel: eth0: transmit timed out, network
cable problem?
Feb 11 18:18:55 stsriscpc kernel: eth0: state: { status=BA86 cfg1=C008
cfg2=1208
}
Feb 11 18:18:55 stsriscpc kernel: eth0: { rpr=BE99 rea=BEBE tpr=0000 }
Feb 11 18:18:55 stsriscpc kernel: eth0: tx head=C tx tail=D
Feb 11 18:18:55 stsriscpc kernel: eth0: packet status = 08804704
                                                                                                                                                       
Feb 11 19:45:26 stsriscpc kernel: NETDEV WATCHDOG: eth0: transmit timed
out
Feb 11 19:45:26 stsriscpc kernel: eth0: transmit timed out, network
cable problem?
Feb 11 19:45:26 stsriscpc kernel: eth0: state: { status=BA86 cfg1=C008
cfg2=1208
}
Feb 11 19:45:26 stsriscpc kernel: eth0: { rpr=B467 rea=B4B4 tpr=0000 }
Feb 11 19:45:26 stsriscpc kernel: eth0: tx head=C tx tail=D
Feb 11 19:45:26 stsriscpc kernel: eth0: packet status = 08804704


This errors comes every now and then, AFAIK not linked to a specific
event, on my home server serving DNS, Apache, monit, cacti, ntop, ...

Hardware platform is Acorn RiscPC running Debian 3.1 for ARM.

I understand that the network connection is temporarily lost and rebuilt
soon afterwards.
The advise I got is to disable TCP segmentation offloading (TSO) through
'ethtool -K eth0 tso off'

What is your opinion on the cause and advise

Thx, Steven

-- 
Gllug mailing list  -  Gllug at gllug.org.uk
http://lists.gllug.org.uk/mailman/listinfo/gllug