[Sussex] NTP Weirdness

Steve Dobson steve.dobson at syscall.org.uk
Wed Apr 23 11:44:53 UTC 2008


Hi all.

I've come across a little problem I've not seen before and was wondering
if any of you have encountered the same.

The system is a duel Pentium 4 2.6GHz running Debian (Etch) 4.0.  I've
installed ntp to keep the clock up to date but it drifts badly, racing
the clock ahead of real time by tens of minutes per day.

Close to the system (in the DMZ has it happens) is the site's NTP
server.  It syncs via the ISP's time server and from some in the debian
pool from ntp.org.  It working just fine.  I know this because all the
other servers on that site sync to it, including a M$ Small Business
Server, and they are not having a problem.

On the same subnet, and therefore with the same round trip network costs
for the NTP packets is a 4 x Xeon 2.8GHz Debain system and a 2 x Xeon
2.8GHz system.  Both of which are configured to use NTP is the same way
and both of which keep good time - just like my systems at home do.

Any clues as it what's going on?

Steve





More information about the Sussex mailing list