Jim Norton wrote: > I'm pretty new to running an NTP server and have been active in the pool > for a few weeks now.
Welcome, Jim. > How does my server appear to be performing? I'm not sure yet what all > the terminology means... offset, jitter etc etc etc... > > http://www.pool.ntp.org/scores/65.197.143.152 Seems resonable according to the graph. Looking at your ntpq however... (These are the same thing except the remote is in IP vs domain name...) > [arthurdent 1] ~ $> ntpq -pn 65.197.143.152 > remote refid st t when poll reach delay offset jitter > ============================================================================== > x128.118.25.3 128.118.25.5 2 u 944 1024 377 144.349 491.617 86.064 > -198.82.1.202 198.82.247.40 2 u 1018 1024 377 160.415 23.064 16.678 > *130.126.24.44 128.174.38.133 2 u 1019 1024 377 70.899 2.919 0.780 > +67.18.58.210 132.163.4.101 2 u 34 1024 377 77.250 0.426 1.706 > +208.201.242.2 204.123.2.5 2 u 955 1024 377 32.596 -3.185 2.331 > > [arthurdent 2] ~ $> ntpq -p 65.197.143.152 > remote refid st t when poll reach delay offset jitter > ============================================================================== > xotc2.psu.edu 128.118.25.5 2 u 959 1024 377 144.349 491.617 86.064 > -ntp-2.cns.vt.ed 198.82.247.40 2 u 6 1024 377 175.063 37.323 14.259 > *ntp-2.gw.uiuc.e 128.174.38.133 2 u 7 1024 377 86.392 -1.259 4.178 > +taos.dayww.net 132.163.4.101 2 u 48 1024 377 77.250 0.426 1.706 > +enigma.wiredgoa 204.123.2.5 2 u 970 1024 377 32.596 -3.185 2.331 Notice your first line(128.118.25.3, otc2.psu.edu) starts with an x (donotes a falseticker, i.e. NTP deems it bad). I've never known this time server to EVER be accurate, although I've only been using NTPd on and off for 7 years now. Offset (491.617) is the number we're looking at here and is the number representing clock accuracy. For a US server such as yours, you can certainly find some servers in the pool that have a much lower delay. Delay indirectly relates to network distance although there are many things that add to delay. I tend to monitor a group of servers and pick servers that delay < 50. Others (especially network-distant countries) may have to be more liberal (delay <100-150) to get a group of servers to sync to. It pays to periodically check your ISPs gateways for NTP services. I noticed this week the Comcast gateways have added NTP service making the delay half that of any other NTP server I monitor. (First two are Comcast NTPservers.) > [arthurdent 3] ~ $> ntpq -pn > remote refid st t when poll reach delay offset jitter > ============================================================================== > -68.86.214.193 68.86.7.72 4 u 657 1024 373 10.289 -7.443 2.014 > -68.87.96.5 18.26.4.105 2 u 641 1024 373 19.737 -4.411 3.368 > -207.106.24.162 192.5.41.41 2 u 766 1024 371 36.551 -7.135 43.198 > *168.75.65.20 209.51.161.238 2 u 641 1024 377 34.252 3.820 3.103 _______________________________________________ timekeepers mailing list [email protected] https://fortytwo.ch/mailman/cgi-bin/listinfo/timekeepers
