[ntp:questions] Re: drift/offset

[Richard B. Gilbert]

Folkert van Heusden wrote:

>>>If I look at the grap of the drift/offset of my ntp-server, everything
>>>seems to be rather jumpy:
>>>http://keetweej.vanheusden.com/ntp_graph.png
>>>The drift can be explained by a bad crystal on the motherboard, but the
>>>offset puzzles me: that server is connected to a DCF77 receiver! It is
>>>also syncing to some time-servers on the net.
>>>Might it be that I'm doing something wrong?
>>>
>>>      
>>>
>>Your system is obviously performing poorly!  Whether you are doing 
>>something wrong or not is a question we can't answer without knowing 
>>exactly what you are doing.
>>Can you post your ntp.conf file?   Does your DCF77 receiver give you a 
>>PPS signal?  If so, are you using it?
>>    
>>
>
>No, just a time-indication every minute.
>It's this one: http://www.gude.info/english/frame-mouseclock.htm
>
>my ntp.conf:
>-----------
># DCF77 receiver
># "Der DCF77 Sender steht in Mainflingen (50 0 Nord, 09 00 Ost), etwa 25 km südöstlich von Frankfurt/Main."
># http://www.maporama.com/
># Mainflingen: 50 1' 54" , 9 1' 33" 50.0318 , 9.026
># Gouda/2804 RR: 52 1' 37" , 4 42' 15" 52.0272 , 4.7044
># http://www.fcc.gov/mb/audio/bickel/distance.html
># -> 374.766km
># speed of light: 299 782.458 km/s
># 1.25012652ms vertraging
># time1 ... ?
># fudging time1 overrides the default delay in the PARSE driver, which for
># mode 14, RAW DCF at 50baud, is 258ms (RAWDCF_BASEDELAY in refclock_parse.c)
>server 127.127.8.0 prefer mode 14
>fudge  127.127.8.0 time1 0.25925012652
># keetweej
>peer 192.168.64.99
>
>#server ntp.xs4all.nl iburst prefer
>server 194.109.22.18
>#server ntp2.nl.uu.net
>server 193.79.237.14
>#server ntp3.fau.de
>server 131.188.3.223
>#server ntps1-1.cs.tu-berlin.de
>server 130.149.17.8
>#
>server time.windows.com
>server time.apple.com
>server clock.redhat.com
>
>driftfile /etc/ntp/drift
>
>authenticate no
>
>restrict default nomodify
>restrict 127.0.0.1
>restrict 192.168.64.99
>restrict 192.168.0.0 mask 255.255.0.0 nomodify
>restrict 192.168.64.1
>restrict 194.109.22.18
>restrict 130.149.17.8
>restrict 131.188.3.223
>restrict 193.79.237.14
>
>
>  
>
>>Looking at the graph of your drift and offset measurements, I think I 
>>see correlation between the drift and the offset measurements.   If 
>>there is correlation, one could be cause and the other effect or both 
>>could be effects due to some other cause.
>>Since your drift measurements are being made with respect to your DCF77 
>>receiver and network servers the problem could be either the crystal 
>>oscillator or the time sources!    The offsets could be explained by 
>>"clock hopping"!
>>    
>>
>
>Hopping between servers you mean? Because (unscientific approach: ) every
>time I look (ntpq -c pe) it always seems to use my dcf receiver.
>
>     remote           refid      st t when poll reach   delay   offset  jitter
>==============================================================================
>*GENERIC(0)      .DCFa.           0 l   47   64  367    0.000   10.298  12.457
>+keetweej.intran 131.188.3.223    2 u    7   64  377    0.467    6.608  10.577
>-auth1.xs4all.nl 193.79.237.14    2 u   81   64  376    7.967   -6.370   7.870
>+ntp1.NL.net     .GPS.            1 u   44   64  377   18.415   -3.600   9.333
>+ntp3-rz.rrze.un .DCFp.           1 u   33   64  377   26.334    2.004   7.096
>+sombrero.cs.tu- .PPS.            1 u  108   64  356   25.123    7.413   8.219
>-time.windows.co 192.43.244.18    2 u   29   64  377  175.538  -15.622  14.507
>-time0.apple.com 17.254.0.49      2 u    9   64  377  165.564   -6.926   8.140
>+clock3.redhat.c .CDMA.           1 u   26   64  377  188.041    6.197   9.578
>
>  
>
>>Was your system synchronized with the same source for the entire 
>>period?   What is your poll interval for that source?
>>    
>>
>
>I have not changed those settings
>
>  
>
>>Can you post the output of ntpq -p?
>>    
>>
>
>You'll find it above.
>
>
>Thanks!
>
>
>Folkert van Heusden
>
>Op zoek naar een IT of Finance baan? Mail me voor de mogelijkheden!
>+------------------------------------------------------------------+
>|UNIX admin? Then give MultiTail (http://vanheusden.com/multitail/)|
>|a try, it brings monitoring logfiles to a different level! See    |
>|http://vanheusden.com/multitail/features.html for a feature list. |
>+------------------------------------------= www.unixsoftware.nl =-+
>Phone: +31-6-41278122, PGP-key: 1F28D8AE
>  
>
The graph at http://keetweej.vanheusden.com/ntp_graph.png
appears to show points taken at intervals of  much less than your poll 
interval of 64 seconds!  How are you computing the offsets?   An offset 
is presumably  t2 - t1;  what is the source of t2 and t1 for each point 
on the graph?