In message <20040122142317.GA14974@zoo.bernard.org>, Jim Bernard writes:
>On Thu, Jan 15, 2004 at 03:53:02PM -0600, Frederick Bruckman wrote:
>> On Wed, 14 Jan 2004, Steve Bellovin wrote:
>> > server          raptor.research.att.com
>> > server          135.207.59.17
>> 
>> Also, two servers is not a good number to have, as
>> there's no way for the daemon to know which one is correct when they
>> disagree.
>
>  (Still trying to catch up on mail after the holidays.)
>
>  Actually, the recommended configuration is to use at least 3.  See, e.g.,
>
>    http://www.eecis.udel.edu/~mills/ntp/html/notes.html
>
>from which I quote:
>
>    It is vital to carefully consider the issues of robustness and reliability
>    when selecting the sources of synchronization. Normally, not less than thr
>ee
>    sources should be available, preferably selected to avoid common points of
>    failure. It is usually better to choose sources which are likely to be "cl
>ose"
>    to you in terms of network topology,...
>
>The daemon goes to a lot of trouble to determine which server currently
>has the best time.


Understood.  But if the daemon is going to misbehave that badly with 
two servers configured, what happens if one of three configured servers 
becomes unavailable?


		--Steve Bellovin, http://www.research.att.com/~smb