Sounds like the same issue I had with one of my servers, everything ran fine, I moved it to a datacenter, and ntp couldn't keep the time in sync at all, moved it back, and it was fine again. It was a dual cpu computer, and the default timing method used in the kernel on freebsd that I was using, kept changing depending on what cpu that thread ran on at any given time, giving me the same log messages your getting:
> Jan 22 07:15:48 dobby ntpd[51979]: kernel time sync status change 6001 > Jan 22 07:50:10 dobby ntpd[51979]: kernel time sync status change 2001 > Jan 22 11:00:35 dobby ntpd[51979]: kernel time sync status change 6001 > Jan 22 11:49:04 dobby ntpd[51979]: kernel time sync status change 2001 I found that for freebsd atleast, there where 3 different clock source options you can configure, and when I changed it, everything behaved correctly. I added this into /etc/sysctl.conf kern.timecounter.hardware=TSC Quoting Jeffrey Goldberg <[email protected]>: > I got an auto notice of problems with my server, and the graphs > certainly do look dismal. > > On Jan 22, 2009, at 10:15 PM, Ask Bjoern Hansen wrote: > >> ---------------------------------------------------------------------- >> 72.64.118.118 (current score: -16.0) http://www.pool.ntp.org/s/1784 >> ---------------------------------------------------------------------- > >> If you have any questions, just reply to this email. Otherwise no >> response is necessary. > > I have not been able to diagnose the problem. From what I see, my > server is acting properly, but I don't really know what sort of > diagnostic tools are available to me. > > Here is the result of ntpq -c 'rv 0' > > $ ntpq -c 'rv 0' > assID=0 status=06f4 leap_none, sync_ntp, 15 events, event_peer/ > strat_chg, > version="ntpd 4.2.4p5-a Tue Jan 13 22:16:19 CST 2009 (1)", > processor="i386", system="FreeBSD/7.1-STABLE", leap=00, stratum=3, > precision=-19, rootdelay=566.555, rootdispersion=578.905, peer=12437, > refid=17.151.16.22, > reftime=cd24718e.09c98675 Fri, Jan 23 2009 10:40:46.038, poll=5, > clock=cd24721c.da46d092 Fri, Jan 23 2009 10:43:08.852, state=4, > offset=0.192, frequency=-41.686, jitter=145.186, noise=52.291, > stability=5.242, tai=0 > > Every time I've checked in the last few hours, my offset has been less > than 15ms. I don't know what are healthy ranges for jitter and > stability, so I don't know if those give me any clue. > > I am getting time resets in my logs. I assume that these indicate > that something is wrong. But again, I don't know where to look. (Log > times are -06:00). > > Jan 22 07:15:48 dobby ntpd[51979]: kernel time sync status change 6001 > Jan 22 07:50:10 dobby ntpd[51979]: kernel time sync status change 2001 > Jan 22 11:00:35 dobby ntpd[51979]: kernel time sync status change 6001 > Jan 22 11:49:04 dobby ntpd[51979]: kernel time sync status change 2001 > Jan 22 13:48:35 dobby ntpd[51979]: kernel time sync status change 6001 > Jan 22 14:56:56 dobby ntpd[51979]: time reset -0.141398 s > Jan 22 15:00:44 dobby ntpd[51979]: kernel time sync status change 2001 > Jan 22 18:53:31 dobby ntpd[51979]: time reset -0.154392 s > Jan 22 22:36:22 dobby ntpd[51979]: time reset -0.147822 s > Jan 23 04:30:08 dobby ntpd[51979]: time reset -0.129208 s > Jan 23 10:37:47 dobby ntpd[51979]: time reset +0.159964 s > > Any help or advice would be useful. Nothing really changed on my > system around the time of the wonkiness, but on another machine on the > network, I did start using bittorrent (Transmission 1.50b1). It's > setting are such that it isn't making more than a dent in my network > traffic. But it's only serving about 10 peers, so I don't see how it > could be confusing my router. > > Cheers, > > -j > > > -- > Jeffrey Goldberg http://www.goldmark.org/jeff/ > > _______________________________________________ > timekeepers mailing list > [email protected] > https://fortytwo.ch/mailman/cgi-bin/listinfo/timekeepers > _______________________________________________ timekeepers mailing list [email protected] https://fortytwo.ch/mailman/cgi-bin/listinfo/timekeepers
