[prev in list] [next in list] [prev in thread] [next in thread] 

List:       ntp-hackers
Subject:    [ntp:hackers] More on time.windows.com and time-nw.nist.gov
From:       "David L. Mills" <mills () udel ! edu>
Date:       2001-12-26 19:06:34
[Download RAW message or body]

FYI,

I've been watching these critters for a couple of days. Looking out from
net 128.4 here, we see delay and jitter varying over half a second and
generally unusable time, at least for the usual Unix NTP daemon
operations. The route spans 26-29 hops via the usual ISP suspects,
including our off-campus provider Voicenet. In fact, I suspect Voicenet
is the suspect of choice.

Now, looking out the side door from net 140.173, which is the CAIRN
research net, The delay is 115 ms with jitter about 2 ms. A trace shows
11 hops and 81 ms before apparently hitting some router that doesn't do
the ICMP Time Exceeded message required for traceroute to work.

At the moment time.windows.com does not respond to ping or ntp. When
last heard it was 600 ms from UTC, which would definitely label it as
falseticker.

Here's the lesson. Be advised do NOT toss time.windows.com in a ntpd
configuration file. Not only will it be useless as a time source, but as
it may wander to and fro over the second, it will cause the intersection
algorithm to toss out first one peer then another and probably clockhop
more than you would like. If you do want to watch the thing, I suggest
you have at least three other stalwart sources and/or mark your best one
as prefer.

Dave
[prev in list] [next in list] [prev in thread] [next in thread]