Needless to say, this can render a remote box rather useless and force a call to the webhotel folks for the "... yes, I know it's Linux, and it never needs a reboot, but..." call!

I found this little script called from cron seems to take care of the problem of never being able to get the server responding.

You can of course adjust the cron interval to suit your needs. Be sure to set your $GWADDR to your gateway address.

If your gateway address changes- don't forget to change it in the script or you will get bounced very X number of minutes. ;-)

#------------SCRIPT-STARTS-UNDER-THIS-LINE-----------#
#!/usr/bin/perl

#netboot: tests to see if gateway is unreachable and
#if so, restart networking. Useful for tulip based
#cards. Call from crontab
#set $GWADDR to your gateway address. 

$GWADDR="192.192.192.192";

$result=`/bin/ping -c 2 -w 3 $GWADDR 2>/dev/null`;
chop $result;
$reslen=length($result);
if ($reslen < 10 || $result=~/100\% packet loss/i){
 system("/etc/init.d/network restart");
}

#------------SCRIPT-ENDS-HERE------------------------#

#----------------- put this line in /etc/crontab
#0,10,20,30,40,50 * * * * root /path_to/netboot