[ntp:hackers] Is this normal?

clemens at dwf.com clemens at dwf.com
Tue May 6 03:27:32 UTC 2008


My memory tells me that I haven't seen this behavior before, but it won't 
be the first time Ive called wolf for a behavior in ntp that has been there
all the time.

The ONCORE driver prints the times of the timehacks (there's got to be
a better word) to the clockstats file.

>From that data you can watch the timehacks bounce around a central
value, or better yet plot that value.

When I restart ntpd, there is usually a fraction of a second error in the
clock, and I would EXPECT that difference to be reduced to ZERO in
a short period of time.  My recollection is that watching ntpdc, the reach
would climb from 0 -> 377, and there would be a delta in the offset, 
followed by the reach being zeroed, and the sequence repeating.
The offset was reduced to zero in no time at all.

What I am seeing at the moment is a several hour (3 or 4) more or less
exponential decay of the offset from some large value to zero.

Is this normal?
Either its been there all along and I don't remember it, or this is
a new feature.  Ive tried both the 4.2.4p4 and the 4.2.5p113 releases
of ntp and see basically the same thing.

Puzzled.

-- 
                                        Reg.Clemens
                                        reg at dwf.com




More information about the hackers mailing list