On Sep 25, 2013, at 9:44, AlbyVA <[email protected]> wrote: > > "...A couple of times an hour the pool system checks the time from your > server and compares it to the local time..."
The code for the monitor is at https://github.com/abh/ntppool/blob/master/monitor -- it's been working essentially like that since Adrian setup the scoring about a decade ago. "Local time" is usually being synced to a Meinberg sponsored GPS-driven time server hosted locally (by Phyber). Someone cut the antenna cable recently so the last weeks, so it's been synced to another older Meinberg Lantime hosted by sonic.net or to the (stratum 1) time servers at UCLA). And yes, it'd make sense to monitor this host more closely than I do. On the todo! Typical peers output is below. remote refid st t when poll reach delay offset jitter ============================================================================== -207.171.... 164.67.62.194 2 u 21 1024 377 0.386 0.226 0.164 -207.171.7.151 69.25.96.13 2 u 1001 1024 377 2.404 -0.640 0.005 2607:f238:3::12 64.142.122.38 2 u 989 1024 377 0.981 -1.193 0.254 -207.171.7.153 69.25.96.13 2 u 997 1024 377 2.550 -0.667 0.102 *64.142.122.38 .PPS. 1 u 6 1024 377 14.496 0.028 0.120 +2607:f010:3fe:1 .GPS. 1 u 464 1024 377 1.658 -0.067 0.030 +164.67.62.212 .GPS. 1 u 1016 1024 377 1.511 -0.093 0.657 -69.25.96.13 .ACTS. 1 u 181 1024 377 12.722 0.463 0.416 > I was wondering, where is "local time" being obtained? Is it an average of > all the other servers in a Country? All I see is a monitoring station being > in Los Angeles. Yes, currently. Making more monitoring stations is on the todo. There's a bunch of infrastructure changes needed to make more work in a sane way, so I'm going to roll it out slowly. As several in the thread pointed out, the graphs are not intended for that sort of sub-ms precision anyway. I believe most of the inaccuracy in the graphs comes from network "fuss", and that'll be the case for a regular client, too. (Generally/on average/...) I haven't looked for a while, but last I looked I didn't see a significant negative or positive bias in the offsets looking across the monitoring data. It'd be fun if someone would have a look again... Ask _______________________________________________ pool mailing list [email protected] http://lists.ntp.org/listinfo/pool
