On Mon, May 28, 2012 at 12:46:45PM +0100, Zi Loff wrote: > I have a mail server which syncs its clock to a local ntp server. This > time server in turn syncs itself with a number of public servers and is > working correctly, as far as I can tell (ntpd -dv shows offsets under > 1ms). > > However, the mail server (an _old_ Pentium III @ 500Mhz) started > drifting furiously, and the ntpd (client) doesn't appear to be actually > adjusting the local clock, although it sets it correctly (if ntpd -s). > > The time server is a Soekris net4801, the mail server is an _old_ > x86 box (Pentium III 500Mhz). Both boxes run 5.0. > > $ sudo ntpd -svd > ntp engine ready > reply from 10.17.16.2: offset -5.781745 delay 0.001901, next query 8s > set local clock to Mon May 28 12:22:39 WEST 2012 (offset -5.781745s) > reply from 10.17.16.2: offset -0.249842 delay 0.001821, next query 7s > reply from 10.17.16.2: offset -0.471290 delay 0.001870, next query 8s > peer 10.17.16.2 now valid > reply from 10.17.16.2: offset -0.724371 delay 0.001822, next query 8s > reply from 10.17.16.2: offset -0.977404 delay 0.001875, next query 9s > reply from 10.17.16.2: offset -1.262066 delay 0.001814, next query 5s > reply from 10.17.16.2: offset -1.420363 delay 0.001776, next query 31s > reply from 10.17.16.2: offset -2.400023 delay 0.001832, next query 33s > adjusting local clock by -1.420363s > reply from 10.17.16.2: offset -2.057283 delay 0.001924, next query 34s > reply from 10.17.16.2: offset -3.131575 delay 0.001821, next query 34s > reply from 10.17.16.2: offset -4.201259 delay 0.001737, next query 34s > reply from 10.17.16.2: offset -5.275638 delay 0.001835, next query 31s > reply from 10.17.16.2: offset -6.255265 delay 0.001849, next query 31s > reply from 10.17.16.2: offset -7.234918 delay 0.001811, next query 31s > reply from 10.17.16.2: offset -8.214601 delay 0.001750, next query 31s > adjusting local clock by -4.481622s > reply from 10.17.16.2: offset -5.024796 delay 0.001837, next query 31s > reply from 10.17.16.2: offset -6.004269 delay 0.001842, next query 30s > reply from 10.17.16.2: offset -6.947157 delay 0.002090, next query 32s > reply from 10.17.16.2: offset -7.958516 delay 0.001831, next query 32s > ... > > As you can notice, the drift is of about 2s/min which is huge and quicky > amounts to a few hours and the next thing I know I'm getting mails which > were sent tomorrow... > > Is this drift too large to be adjusted smoothly by ntpd? And if so, any > idea as to what might be causing this? > > Many thanks > Zi Loff > > -- >
Correction: I realize it is adjusting the lock (offset diminishes after adjustment). The output above does not show the behaviour I referred to. I'm looking further into this, and I'll post some more info later. Sorry about the noise. (that is a big drift, though...) --