I normally check my Internet Connection is live by running a wee cron job every 30 minutes or so which will ping an external address and dump the output to a file, I then search the file for the expected response, in this case "2 received", as I ping an address twice and then if this is the case, do nothing more, if not, then I log it or whatever.

I'm looking for something a little more robust as I've noticed that there are times when I'm logging downtime but the connection is actually up, possibly missing one ping reply or so.

All ideas, comments appreciated...

O this is on an old Slackware server.

Thanks

You could for example create dependencies like first check a few repo mirrors for a file ('curl -s'?), if that doesn't work tcptraceroute or ping your next hop router and if that doesn't work arping your first hop router or CPE? And why not start by making the script more robust? Just save debug output and see what condition you're missing?..

Just some ideas out of the box.
There usually is a lot of chatter on a nic. Some of that may be monitored maybe also. Arp may be an idea. Some of the fail over concepts may work also.

You can test for a connection to a port with netcat

That simple line might not be your ultimate solution but I hope that helps you down a successful path.