Tim Lundström schreef:
> I know that the monitoring system can't be used as an absolute measure
> of the server performance. However for many server admins there are no
> other way to at least have an indication of how your server is
> performing.
All ntp servers, even those with a locally connected reference clock,
should have about three external reference servers via the network.
This assures that even when your local clock is broken, ntpd can
determine that fault and reject the time it provides.
It also gives you info about the average RTT to those reference servers,
so you can be alerted when it is too high.
It is possible to write this data to logfiles for analysis (e.g.
plotting with gnuplot). This allows you to judge the quality of your
server without relying on the ntp pool monitoring system.
(of course you still rely on the reference servers, but they can be much
more local to your system)
Example for my system:
remote refid st t when poll reach delay offset
jitter
==============================================================================
LOCAL(0) LOCAL(0) 10 l 3 64 377 0.000 0.000
0.004
SHM(0) .GPS. 0 l - 16 0 0.000 0.000
4000.00
SHM(1) .PPS. 0 l - 16 0 0.000 0.000
4000.00
-auth1.xs4all.nl 193.79.237.14 2 u 844 1024 377 5.605 3.522
0.222
-auth2.xs4all.nl 193.79.237.14 2 u 898 1024 377 7.467 0.693
0.276
*ntp0.nl.uu.net .GPS. 1 u 935 1024 177 21.470 3.375
6.199
+doei.cs.uu.nl 131.211.81.10 2 u 885 1024 377 8.618 0.366
0.475
+ntp1-rz.rrze.un .DCFp. 1 u 696 1024 377 27.372 -4.997
0.572
-meow.febo.com 192.168.1.234 2 u 822 1024 377 120.469 0.614
11.148
Note my SHM-connected GPS receiver is broken due to hardware failure.
But I still have the two NTP servers of my provider xs4all.nl (which are
not very reliable) and some good other servers as a reference.
Rob
_______________________________________________
timekeepers mailing list
[email protected]
https://fortytwo.ch/mailman/cgi-bin/listinfo/timekeepers
