On Thu, 2011-01-06 at 21:31 -0600, Dale wrote: > William Kenworthy wrote: > > > > Dale, can you post (a sanitised) version of what 'ntpq -p' gives after > > ntpd has been running for some time, and the sanitised result of > > 'ntptrace. Also include your full (sanitised) ntp.conf > > and /etc/conf.d/ntpd. > > > > This might help us see more detail of what is happening. > > > > BillK > > > > * sanitised - obfuscate public IP's only. > > > > > > > I should have posted this a while back. Here we go: > > r...@fireball / # ntpq -p > remote refid st t when poll reach delay offset > jitter > ============================================================================== > +triangle.kansas 128.252.19.1 2 u 4 64 127 56.270 673.749 > 238.194 > +B1-66ER.matrix. 192.43.244.18 2 u 14 64 377 75.505 672.846 > 165.087 > +kallisti.us 208.90.144.52 3 u 22 64 377 62.917 663.831 > 168.738 > *kazilik.haqr.ne 209.51.161.238 2 u 42 64 377 66.483 653.962 > 166.119 > r...@fireball / # ntptrace > localhost: stratum 16, offset 0.000000, synch distance 0.000240 > r...@fireball / # >
Notice the difference in ntptrace between mine below and yours? The asterisk in your ntpq table indicates that is the chosen server - but not that it is actually locked to it - ntptrace is showing it is not locked. rattus ~ # ntptrace localhost: stratum 6, offset 0.001309, synch distance 0.160683 moriah: stratum 5, offset -0.004895, synch distance 0.155622 dns.iinet.net.au: stratum 4, offset 0.001463, synch distance 0.135967 ***Association ID 60244 unknown to server rattus ~ # Did you try the tinker panic 0 or -g arguments I suggested earlier? A third option is to increase the slew window rather than stepping - the default slew is less than what you are trying to force it to do so it fails and gives up. rattus ~ # ntptrace localhost: stratum 6, offset 0.001309, synch distance 0.160683 moriah: stratum 5, offset -0.004895, synch distance 0.155622 dns.iinet.net.au: stratum 4, offset 0.001463, synch distance 0.135967 ***Association ID 60244 unknown to server rattus ~ # I would suggest stopping ntpd and modifying the ntp.conf file as suggested, deleting /etc/adjtime and the drift file then rebooting. Run for a few hours and see if it still jumps time. ntpd will correct some extreem time issues but it needs to be properly configured. If it is actually jumping time rather than incorrect drift/slew/adjustime issues then you need to track down why - good luck with that! So far all I am seeing is a clock thats off too far to correct and isnt being allowed to correct by the config BillK
