[ntp:questions] Problems distibuting time from GPSD program to NTPD

Askar Ibragimov askar.ibragimov at gmail.com
Mon Feb 4 14:24:44 UTC 2008


I have a gpsd and ntpd in connection running on a Fedora 5 box.
Sometimes (once in a few days) strange thing happen:  NTP reports that
connection to GPSD is lost, and never tries to recover it, so I
receive status like this:

ind assID status  conf reach auth condition  last_event cnt
  1 58316  9624   yes   yes  none  sys.peer   reachable  2
  2 58317  9424   yes   yes  none  candidat   reachable  2
  3 58318  80e3   yes   yes  none    reject  lost reach 14
  4 58319  80f3   yes   yes  none    reject  lost reach 15

ntpq> peers
     remote           refid      st t when poll reach   delay
offset  jitter
*ousrvr.oulu.fi     2 u    2   16  377    0.446
-0.014   0.451
+ousrvr2.oulu.fi     2 u    -   16  377    0.460
-0.037   0.496
 SHM(1)          .GPS1.           0 l  50h   16    0    0.000    0.000
 SHM(0)          .GPS.            0 l  50h   16    0    0.000    0.000

Both GPSD and NTPD are last versions. So NTPD give up at some point. I
also have another machine, on Fedora 7, which so far did not show such
a behaviour, but I'm not quite sure if it will not be repeated there
too. When I disconnect GPS receiver manually, NTP also reports lost
reach, but regain "reach" when I plug cables back. And of course, on
the system in question wiring is OK. There is a question to GPSD
experts, under which circumstances GPSD can stop reporting the time to
NTPD? Can I adjust GPSD behaviour \code to fix this? I would
appreciate any help.

My ntpd.conf is like this:
# oulu ntp
server minpoll 4 maxpoll 4 burst
server minpoll 4 maxpoll 4 burst

server minpoll 4 maxpoll 16 prefer
fudge refid GPS1

# GPSD defaults
# just GPS
server minpoll 4 maxpoll 16
fudge time1 0.150 refid GPS

step 0.06

Will be thankful for an advice.
Chees, Askar

More information about the questions mailing list