[ntp:questions] dispersion/offset cycling between low and high

Dorothea Muecke dorothea.muecke at lycos.co.uk
Fri Dec 5 09:35:04 UTC 2003


Hello,

I have a PC here (running QNX4.25 patch G) that is flatly refusing to
synchronise reliably with our internal time server (GPS stratum 1). 
We have several other PC running QNX4.25 which give us no trouble
whatsoever.
The O/S and software setup is the same for all of them.

For illustration, I have here the typical statistics from ntpdc after
a
re-start of ntpd (after an xntpdate). In the beginning the offset is
low and dispersion high. Then the offset increases and dispersion
decreases until node3 briefly synchronises, but soon the offset is too
high and the dispersion jumps back to 64000 and this carries on and on
and on.

node3:

   (rem)  Address   (lcl)      Strat Poll Reach    Delay   Offset   
Disp
==========================================================================
-tym.f1.com      wildcard          0   64  0        0.0    0.0    
64000.0
-tym.f1.com      wildcard          1   64  007      20.0   521.0  
31595.0
-tym.f1.com      wildcard          1   64  017      20.0   704.0  
15685.0
-tym.f1.com      wildcard          1   64  037      20.0   886.0  
64000.0
-tym.f1.com      wildcard          1   64  177      20.0   371.0  
31597.0
-tym.f1.com      wildcard          1   64  377      20.0   552.0  
15684.0
*tym.f1.com      wildcard  	     1   64  377      20.0   920.0  
3795.0
-tym.f1.com      wildcard          1   64  377      20.0   173.0  
63500.0
-tym.f1.com      wildcard          1   64  377      20.0   350.0  
31588.0
-tym.f1.com      wildcard          1   64  377      20.0   1436.0 
64000.0
-tym.f1.com      wildcard          1   64  377      20.0   357.0  
31588.0
-tym.f1.com      wildcard          1   64  377      20.0   548.0  
15687.0
*tym.f1.com      wildcard          1   64  377      20.0   901.0  
3786.0
*tym.f1.com      wildcard          1   64  377      20.0   1084.0 
1821.0
*tym.f1.com      wildcard          1   64  377      20.0   1283.0 
856.0
-tym.f1.com      wildcard          1   64  377      20.0   1464.0 
64000.0


For comparision, another node, which shows rock-solid ntp start-up
behaviour:

node1:

  (rem)  Address   (lcl)      Strat Poll Reach    Delay   Offset   
Disp
==========================================================================
-tym.f1.com      wildcard          0   64  0        0.0    0.0    
64000.0
-tym.f1.com      wildcard          1   64  003      20.0   0.0    
63500.0
-tym.f1.com      wildcard          1   64  007      20.0   0.0    
31500.0
-tym.f1.com      wildcard          1   64  017      20.0   0.0    
15500.0
*tym.f1.com      wildcard          1   64  037      20.0   0.0    
7500.0
*tym.f1.com      wildcard          1   64  077      20.0   0.0    
3500.0
*tym.f1.com      wildcard          1   64  377      20.0   0.0    
500.0
*tym.f1.com      wildcard  	     1   64  377      20.0   0.0     0.0


Here is part of the logfile:

Dec  4 16:03:20 node<<3>> ntpd[19108]: /usr/local/sbin/ntpd version
$Revision: 3.4.1.9 $
Dec  4 16:03:20 node<<3>> ntpd[19108]: patchlevel 13
Dec  4 16:03:20 node<<3>> ntpd[19108]: Drift compensation value
initialized to -1.457885
Dec  4 16:03:20 node<<3>> ntpd[19108]: sys.precision set to -13
Dec  4 16:08:39 node<<3>> ntpd[19108]: clock: select peer 192.168.1.11
stratum 1 was UNSYNCED
Dec  4 16:08:39 node<<3>> ntpd[19108]: adjust: STEP 192.168.1.11 st 1
off 0.886752 drft -1.457885 cmpl 0.000000
Dec  4 16:08:39 node<<3>> ntpd[19108]: Lost NTP peer 192.168.1.11
Dec  4 16:12:55 node<<3>> ntpd[19108]: clock: select peer 192.168.1.11
stratum 1 was UNSYNCED
Dec  4 16:17:13 node<<3>> ntpd[19108]: adjust: STEP 192.168.1.11 st 1
off 1.460229 drft -1.457885 cmpl 0.000000
Dec  4 16:17:13 node<<3>> ntpd[19108]: Lost NTP peer 192.168.1.11
Dec  4 16:21:28 node<<3>> ntpd[19108]: clock: select peer 192.168.1.11
stratum 1 was UNSYNCED
Dec  4 16:25:46 node<<3>> ntpd[19108]: adjust: STEP 192.168.1.11 st 1
off 1.436342 drft -1.457885 cmpl 0.000000
Dec  4 16:25:46 node<<3>> ntpd[19108]: Lost NTP peer 192.168.1.11
Dec  4 16:30:01 node<<3>> ntpd[19108]: clock: select peer 192.168.1.11
stratum 1 was UNSYNCED


Has anyone an idea what could cause this behaviour or what we could
try to
get NTP to work stable on node3?

Any help is appreciated

Dorothea



More information about the questions mailing list