[ntp:questions] Re: Nagios NTP Warnings - All My Servers

Richard B. Gilbert rgilbert88 at comcast.net
Sat Jun 10 21:15:15 UTC 2006


karl at klxsystems.net wrote:
> Hi,
> 
> All machines talking to my main NTP node are giving errors back to the
> Nagios check_ntp command.
> 
> What's the best way to troubleshoot this?  Any useful commands?
> 
> --
> Here are some of the warnings from the pool of servers I have running:
> 
> NTP WARNING: Offset 5.549596 sec > +/- 2 sec, jitter 9.111 msec
> NTP WARNING: Offset 5.560099 sec > +/- 2 sec, jitter 9.534 msec
> NTP WARNING: Offset 5.596052 sec > +/- 2 sec, jitter 10.068 msec
> 
> When I run ntpq -p I get:
> /usr/sbin/ntpq -p
> 
> remote           refid      st t when poll reach   delay   offset  jitter
> ==============================================================================
> *toshi.keneli.or .GPS.            1 u   63   64  377   87.424    7.085  
> 1.970
> +dankulp.com     ntp2-ext.schlun  3 u    2   64  377   81.263    9.698  
> 0.630
> +mx11.wtfo-guru. ntp-1.cns.vt.ed  3 u   80   64  376   77.466    8.761  
> 0.163
> 
> 
> And when I run /usr/sbin/ntpdc -c kerninfo:  I get:
> pll offset:           0.007944 s
> pll frequency:        4.733 ppm
> maximum error:        0.074316 s
> estimated error:      0.001374 s
> status:               0001  pll
> pll time constant:    2
> precision:            1e-06 s
> frequency tolerance:  512 ppm
> pps frequency:        0.000 ppm
> pps stability:        512.000 ppm
> pps jitter:           0.0002 s
> calibration interval: 4 s
> calibration cycles:   0
> jitter exceeded:      0
> stability exceeded:   0
> calibration errors:   0
> 
> When I view my /var/log/messages I see:
> Jun 9 11:37:53 myhost ntpd[17964]:kernel time discipline status change 41
> 
> Jun 9 11:40:01 myhost ntpd[17964]:kernel time discipline status change 1
> 
> Any suggestions?
> 
> -karlski
> 
> 
> 
> 
> _______________________________________________
> questions mailing list
> questions at lists.ntp.isc.org
> https://lists.ntp.isc.org/mailman/listinfo/questions
> 

Try to find some servers closer to you in netspace.  Delays of 77.4 to 
87.4 milliseconds are rather large.  The error in transmitting time from 
server to client is bounded by 1/2 the round trip delay.  That means 
that the time you are getting from each server could be off by 1/2 the 
delay figure for that server.  It probably is nowhere near that bad but 
"probably" is the best you can say.

A little research suggests that toshi.keneli.org is located in Seattle, 
Washington, while dankiulp.com is located in Karlesruhe, Germany.  The 
name of your third server is not sufficiently complete to allow me to 
locate it.

Geographic diversity is a property far less desirable in NTP servers 
than closeness in net space!!!!!!

I'd also suggest adding a fourth server.  With only three servers, the 
failure of any one of those servers causes your configuration to 
degenerate to the worst possible case: two servers.   It is written that 
a man with two clocks can never be certain what time it is!!  When ntpd 
has only two working servers, it has no basis for determining which one 
is more nearly correct.




More information about the questions mailing list