dns Health Checkups

[Martin McCormick]

	Occasionally, I get a phone call that starts out "What's wrong
with dns?"  99.9 times out of 100, nothing is wrong with dns.  It's
usually everything from the individual P.C. to the occasional cable
cut or router crash that temporarily disconnects our campus from our
Internet service provider.

	I usually run a shell script that pings all the root name
servers plus a list of sites in far-flung places as well as a few in
other towns that are on our ISP and see how many answers come back.

	Is there any source code for tests I might be able to
incorporate in to a cron job that would occasionally do a lookup or
two and return a status that could be used to either notify me of
trouble or stay the course.  Nothing's wrong.

	In the past 6 or 8 years that we have been using bind, I can't
think of one real dns failure that wasn't the result of human error
here or hardware trouble with the host or its network connection.

	Are there better ways to get this kind of information?  I am
not locked in on anything yet, just trying to stay ahead of the game
so I can give a quick honest answer when the phone rings.

	Our network management systems ping every 15 minutes.  That's
not often enough and all it tells is that the Ethernet adaptor is
answering echo request packets.  I have seen a few times in which
that's exactly all it was doing.

Martin McCormick WB5AGZ  Stillwater, OK 
OSU Center for Computing and Information Services Data Communications Group