Re: kern/53016: Clock not stable

[Robert Elz <kre%munnari.OZ.AU@localhost>]

The following reply was made to PR kern/53016; it has been noted by GNATS.

From: Robert Elz <kre%munnari.OZ.AU@localhost>
To: gnats-bugs%NetBSD.org@localhost
Cc: 
Subject: Re: kern/53016: Clock not stable 
Date: Wed, 04 Apr 2018 09:35:41 +0700

     Date:        Wed,  4 Apr 2018 02:10:01 +0000 (UTC)
     From:        mlelstv%serpens.de@localhost (Michael van Elst)
     Message-ID:  <20180404021001.05AE67A1E9%mollari.NetBSD.org@localhost>
 
   |  That's not how ntpd works. The clock is adjusted all the time by
   |  running it slower or faster.
 
 That's how NTP is supposed to work, but there is a maximum value to the
 "slower or faster" that it will apply.   If the underlying clock is off by more
 than what the adjustment can handle, then time will still slide, just slower.
 When ntpd detects the difference between the correct time and the clock
 time has become too big, it causes time to jump, and the whole process
 repeats.
 
   |  Please check with 'ntpq -p' whether ntpd actually synchronizes to
   |  a time server.
 
 I doubt that is the issue.   Rather attempt to see why ...
 
 kivinen%iki.fi@localhost said:
   | It is actually worse than that. When I stopped ntpd to check whether this
   | issue was fixed by latest kernel and forgot to restart it, the clock was
   | around 20 seconds off after about 10 hours or so.  
 
 That's 2 secs/hour, or 1/1800 which is more than ntp can fix by adjusting.
 
 I suspect that the system needs to switch to using some other clock source
 than the one it is picking by default.
 
 kre