Jim Bernard said:

>  * PK recently fixed a bug in the sparc port that caused processes to be
>    spawned at elevated spl levels.  This was causing lost closk interrupts
>    and made ntp frequently lose synchronization.  If your clients are
>    sparcs, that could be the problem, but it shouldn't be a new problem
>    associated with the new ntp server.  I think some other ports may
>    have/have had similar problems, but i386 evidently did not.

One of the 1.4 machines is a sparc, but the other two are i386.

>  * I was perusing some ntp docs the other day and noticed that they
>    mentioned that an ntp4 client will work with an ntp3 server, but
>    not an ntp3 client with an ntp4 server.  Now, this surprised me,
>    since I thought I had seen them work interchangeably in the past,
>    but if it's really true, and if your server was ntp3 before the
>    upgrade, that could be the problem.

I don't know about this. My two Solaris/sparc boxes are running the
stock xntpd 3.4y included with the Solaris distribution and they seem
to be working fine.

If some versions of ntp 3.x don't work with ntp 4.x, then there could
be wide-ranging problems as stratum 1 servers upgrade to 4.99k23.
-- 
Rick Kelly  rmk@rmkhome.com  www.rmkhome.com