On Dec 27, 2010 7:54 AM, "Joe" <j...@jretrading.com> wrote: > > On Sun, 26 Dec 2010 23:57:42 +0000 > Chris Davies <chris-use...@roaima.co.uk> wrote: > > > S Mathias <smathias1...@yahoo.com> wrote: > > > ping -W 1 -c 4 google.com >& /dev/null | grep "100% packet loss" > > > ping -W 1 -c 4 www.yahoo.com >& /dev/null | grep "100% packet loss" > > > > > both sides "false", because they have no output, because google.com > > > and www.yahoo.com is reachable. > > > > Or because your local system is so offline that it's got no way of > > resolving those names to IP addresses. > > > > I'd recommend you turn your pattern match around to look for any sort > > of success, rather than one specific instance of failure. > > > > I would recommend going further than that: I've often seen a router > pass pings and replies by DNS but not web pages or some other protocols. > My current scripts check for specific strings from up to six websites, > rotating the order each time so one site doesn't get too much traffic. > There is a log kept which I check every month or two to confirm the > sites still contain the strings. > > It may seem a bit overengineered, but I made progressively more and more > complex scripts until they worked reliably and without false positives, > and I'm afraid this was the point I reached. They haven't changed for > some years now, nor has the number increased beyond six. > > Here's a fragment from script rw1: > > logfile="/var/log/routerboot" > echo -e -n `date` >> $logfile > echo -n " -rw1- " >> $logfile > count=0 > count=$((count+`curl -s www.google.com | grep -c content-type`)) > echo -n $count >> $logfile > if [ $count -eq 0 ] > then > count=$((count+`curl -s www.google.co.uk | grep -c content-type`)) > echo -n $count >> $logfile > if [ $count -eq 0 ] > then > .. etc, culminating in router reboot code ...
First, your correct that ping tests aren't the end all be all and if you can't access the web, you've got issues one way or the other. However, if you're going to go as far as to start testing different protocols on a line, you might as well base your script on scapy framework and call it a day (complete low level control of the build up and breakdown of a protocol). Also, you can ask nmap to do a full handshake.
