I think I noticed that error before but did not pay close attention to it since my NTP just worked.

The fact is that something changed a while back (the new PPS API was accepted in 2000) and the fudge command was not updated accordingly. Either that, or the setup was always wrong?

The old /etc/conf/ntpd setup file included a line like this:

fudge   127.0.0.1     stratum 10

which is wrong because the local clock is not consider to be part of 127.0.0.1. I've never changed that IP address because I rarely look at the ntpd logs and since I did today, I noticed that error.

The correct settings (which I had in Ubuntu 16.04 but I never applied them) is to use a different IP address:

#Changes recquired to use pps synchonisation as explained in documentation:
#http://www.ntp.org/ntpfaq/NTP-s-config-adv/

#server 127.127.8.1 mode 135 prefer    # Meinberg GPS167 with PPS
#fudge 127.127.8.1 time1 0.0042        # relative to PPS for my hardware

#server 127.127.22.1                   # ATOM(PPS)
#fudge 127.127.22.1 flag3 1            # enable PPS API

As we can see, they offer a couple of examples, although they do not explain here what to use to access the standard computer clock.

According to this email, you should use address 127.127.1.0. For sure it eliminates the error message in the log file. So there is that. But I am not too sure how to test that it actually works. I guess I'll have to look into that from one of my VPS and see how it reacts.

So right now I have this in my file:

fudge   127.127.1.0     stratum 10

But again, since my other NTP URLs generally work as expected, I probably won't notice any difference either way.