Aug 10 2007: For reference, here is a typical clean (boring) day:
Offset,
Round Trip Time
May 3 2008: Here is another boring day.
Note that both clocks have a significant offset.
That offset is stable and easily big enough to measure.
Offset,
Round Trip Time
Aug 14 to Sep 17, 2007: There was an extended glitch
somewhere on the path from my house to USNO-PA.
The round trip time increased by 80 ms.
Since the offset time was about half of that,
I assume the extra routing time was all on one direction.
Start: Offset,
Round Trip Time
Finish: Offset,
Round Trip Time
Dec 21, 2007 to Mar 4, 2008: There was another extended glitch
on the path to USNO-PA.
Start: Offset,
Round Trip Time
Finish: Offset,
Round Trip Time
For extra credit, there is a one hour glitch on a different server.
Aug 20, 2007: A 2 hour glitch that hit both clocks,
and another 1 hour glitch that only hit one clock.
Note that this took place during the month long 40 ms offset
described above.
Offset,
Round Trip Time
Jan 31, 2008: There are two glitches in this
graph,
The first is a shift of about 1.5 ms in the RTT.
It doesn't show on this scale, but there is an offset
jump of roughly half of that.
Twelve hours later, the offset shifts by 7 or 8 ms
without any change in the RTT.
I don't know what happened.
