>   I would like to voice my agreement here.  I've been running NTP for
> years on all of my machines, at home and at work, and it's always done
> a very good job.
> 
>                   -Dave McGuire

Yep, me too.  I've been running ntp since netbsd1.0 with no problems.  Be 
sure to check the ntp page for the latest info and source code.

http://www.eecis.udel.edu/~ntp

A few hints I can relate are:

+) use the driftfile configuration.  Having a driftfile will allow the 
daemon to be more stable in the long run, and stabilize more quickly 
after a reboot or crash.  I've seen it help even on my PC-based server.  :)

+) use good netiquette and don't slave from Stratum 1 servers unless 
you're running a large network of hosts.  There's a good list of public 
servers available at the ntp page.  Set up logical trees, too.  If 
you run a network, have hosts on each network ntpdate their time from a 
server running xntpd higher up the "food chain."  And have servers like it 
peer with each other and slave from a local clock source or a Stratum 1 
server.

+) In your /etc/rc.local, use ntpdate to set the clock before you launch 
the ntp daemon itself.  This gives ntp a sane time to work from--very 
helpful if you're coming back from an extended outage.  If the time 
difference is too big between your "broken" time and the real time, ntp 
will bomb out and tell you to set it yourself.  :P

+) check the Usenet group comp.protocols.tcp-ip.ntp.  Pretty low on the 
traffic side, but occasionally some good info will pass through there.


+Matt
-- 
Matthew B. Wood                                              mbwood@netcom.com
    
      "Do you know how many time zones there are in the Soviet Union?"