[dns-operations] Pinging the root name servers to check my connectivity?

[Simon Munton]

As with all monitoring, you also run into the interesting issue of "what 
am I monitoring?". If the connectivity at your distant test point is 
less reliable than your own, you get "you're down", when in fact they 
are down (or somewhere in between). The result depends on the weakest link.

We tried "pingdom" a while back, but stopped using it because the 
connectivity at their monitoring stations was clearly less reliable than 
our nodes, so they kept saying we were down when in fact it was them - 
I'm sure they're loads better now.

We had the same thing with DNSMON in Russia - RIPE would take no 
responsibility for the reliability of the probes, as they don't own or 
host many of them - so when their Russian probe was saying we were down 
I complained and they said it wasn't anything to do with them.

The problem was caused by the fact that probe's subnet wasn't being 
announced into the Moscow IX and (as is common IX practice) we were 
blocking replies over our international bandwidth - I reported it to the 
people hosting the probe, their reply was "that's a private subnet so we 
don't announce it".



Previously, where we have a Nagios server at a server farm we configure 
a "near router" and "far router(s)" and make all the host entries depend 
on those. This helps reduce the "your site is down", when in fact its 
just a local connectivity issue.