🚀 go-pugleaf

I've got a small problem running xntpd on a small subnet here.  There
problem is that occasionally, xntpd (through ntpq or xntpdc) will say
that a particular host hasn't been reached in some long period of time,
several hours or even days.  However, I know that the machines are both
up and running, and xntpd is otherwise healthy.  Right now, I've got
two machines which have been peering with each other successfully for
quite a while (about a year), and although they're both up and running
fine, each machine shows the other as unreachable.  On both of the
machines, "reach" is 0.  On one of them, "when" is 25h, but on the
other it's less than 1000.  I know there's nothing wrong with the
network, these machines are physically only about 10 feet apart and
everything else about them works fine (rlogin, NFS, etc).  Usually, if
I unconfigure the other peer from each machine, wait a few minutes,
then reconfigure them, it comes back to life.  Sometimes it seems to
come back to life on it's own after a while.

As an additional oddity, sometimes I see one host thinking the other
is dead, but the other one sees the first one OK.  Strange.

I've seen this both between internal hosts, and between internal and external
hosts, at times when everything else about NTP seemed to be working
just fine on both ends.  I can ping them manually (and ntpq them), but the
xntpd's aren't cooperating.

Could it be something wrong in my "restrict" statements in the config file?
If that was the problem, why would it only happen sometimes, and why
would it sometimes just go away again after a while?

Any clues would be appreciated.  This is mostly a curiosity,
it's not causing any other problems that I know of.

-steve