[ntp:questions] PPS timestamp comparsion fails
Nickolay Orekhov
nowhere at mail.ru
Mon Jul 2 11:50:34 UTC 2012
Hello!
Please, look at the following config and ntpq report ( ntpd 4.2.7p164 ):
tos mindist 0.128
# PPS reference clock
server 127.127.22.0 maxpoll 3
fudge 127.127.22.0 refid PPSI
# NMEA reference clock
server 127.127.20.0 mode 80 prefer maxpoll 3
fudge 127.127.20.0 refid NMEA stratum 8 time2 0.018
ntpq> pe
remote refid st t when poll reach delay offset
jitter
==============================================================================
xPPS(0) .PPSI. 0 l 2 8 377 0.000 8.664
0.267
xGPS_NMEA(0) .NMEA. 8 l 2 8 377 0.000 -1047.2
1.960
This two time sources became falsetickers, but the real difference between
them is only 55ms. So increased tos mindist
should help me to stay them both "on the island".
This situation can be a bit synthetic but sometimes we use for example NTP
source + PPS and that's very useful config.
Please, could you give me some hints according this situation.
PPS can't have offset more than 1 second, so any other source with big
offset + PPS will be falsetickers.
Is it a bug? Maybe ntpd should somehow differ "PPS offset" from general
offsets when determining falsetickers?
Or maybe using this scheme is only possible when i use PPS + 2 other
sources ? And when I don't have 2 sources should i use PPS + GPS
as one source?
More information about the questions
mailing list