[ntp:questions] shm and reach

Matthew Lee mjlee at mindless.com
Thu May 7 23:47:02 UTC 2009


Hi,

I'm trying to use ntpd with gpsd on a linux system, but am unable to get ntpd to update my system clock.

I'm using the first SHM segment only (127.127.80.0) and do not have a PPS signal available.  I know that gpsd is running, has a 3D fix and is writing into the SHM segment.

Running ntpq -p shows that the reach is never non-zero.

     remote           refid      st t when poll reach   delay   offset  jitter
==============================================================================
 SHM(0)          .GPS.            0 l    -   16    0    0.000    0.000   0.004


Interestingly, ntpq's associations command says "yes" for reach:
ind assID status  conf reach auth condition  last_event cnt
===========================================================
  1 44334  8025   yes   yes  none    reject  clock expt  2

I assume that "condition == reject" is bad?



>From looking through the ntpd source code, it appears that the reach is only ever set to non-zero in the receive() function.  This makes sense when using remote NTP servers, but I don't understand how this works when there is no remote server, just the (local) SHM segment.  As far as I can tell, no packets are ever received, the reach never gets "seeded".  refclock_transmit() does get called regularly, so peer->reach <<= 1 does happen.... but shifting zero still gives you zero.

shm_poll() also gets called regularly and the appropriate timestamps are read from the SHM segment, but they never get used (due, I believe, to the reason outlined above).

My ntp.conf file is:

restrict default ignore
restrict 127.0.0.1
server 127.127.28.0 prefer minpoll 4 maxpoll 4
fudge 127.127.28.0 time1 0.420 refid GPS
broadcastdelay 0.008
driftfile /etc/ntp.drift
logfile /var/log/ntp.log


I'm running ntpd version 4.2.2p2
(I've also tried with ntpd 4.2.4.p6 and got the same outcome).


Thanks for any help...


-- 
Be Yourself @ mail.com!
Choose From 200+ Email Addresses
Get a Free Account at www.mail.com




More information about the questions mailing list