[CentOS] script to monitor interface errors

2009/7/23 Tom Brown <tom at ng23.net>:
> Has anyone knocked up a script to monitor for interface errors before?
>
> i know i can look at the interface statistics with ethtool -S but i
> wonder if anything has something that looks at the last value for this
> statistic and then can alert if they increase?
>
> any thought how to solve this? it has to be host based and cant be on
> the switch
>

surprise, surprise

I have a similar situation. On a very slow Pentium I machine, runing
2.2.XXX kernel, sometimes I have overruns and frame errors, which
choke up the interface and I have to restart it using
The overruns are due to torrent traffic generated by one of the geeks here.

ifconfig eth1 down
ifconfig eth1 up

So I was thinking of using a cron job and a simple script that
restarts the interface, provided the errors have increased since last
check.