[ntp:questions] ntpq displays faulty offset

Lena Kleyner lenak at qwilt.com
Sun Jul 6 11:19:40 UTC 2014


Hi,

  Our system uses NTP package version 4.2.6 (Linux CentOS 6.5). Due to
system constraints, we have configured the ntpd to update the clock only
using slewing.

  At some point, one of our system's clock's was ahead by 20 minutes, and
we noticed what appears to be a faulty output from the ntpq utility. In the
following, please note the offset value:
<output>

ntpq> peers
     remote           refid      st t when poll reach   delay   offset
 jitter
==============================================================================
*172.246.0.10    10.9.8.233       4 u    3   64  377    0.138  -172017
153.311
</output>


As you can see, the offset (-172017) displayed is 10 times smaller than the
real one.
When trying to read the association status, the offset displayed seems
correct (offset=-1720176.867). Note, however, that the last eight
measurement offsets have the same problem as the previous command:
<output>

ntpq> associations

ind assid status  conf reach auth condition  last_event cnt
===========================================================
  1 61621  963a   yes   yes  none  sys.peer    sys_peer  3

ntpq> pstatus 61621
associd=61621 status=963a conf, reach, sel_sys.peer, 3 events, sys_peer,
srcadr=172.246.0.10, srcport=123, dstadr=172.246.0.3, dstport=123,
leap=00, stratum=4, precision=-23, rootdelay=208.099, rootdisp=218.979,
refid=10.9.8.233,
reftime=d7415953.839cf905  Tue, Jun 10 2014 10:15:47.514,
rec=d74163f1.fe65931d  Tue, Jun 10 2014 11:01:05.993, reach=377,
unreach=0, hmode=3, pmode=4, hpoll=6, ppoll=6, headway=37, flash=00 ok,
keyid=0, offset=-1720176.867, delay=0.138, dispersion=1.444,
jitter=153.311, xleave=0.016,
filtdelay=     0.14    0.15    0.15    0.14    0.15    0.15    0.15    0.16,
filtoffset= -172017 -172021 -172024 -172028 -172031 -172034 -172038 -172041,
filtdisp=      0.00    1.01    1.98    2.96    3.92    4.91    5.90    6.86
</output>


  Is it a known issue of have I missed anything?

  Thank you in advance,
  Lena


More information about the questions mailing list