[ntp:questions] Time steping regardless of "-x" slew only option

vwvr6vw brian.l.hostetler at lmco.com
Wed Jul 4 21:37:10 UTC 2007


The trace data shows that the NTP server is taking 1 second after
receiving the query from the client to respond.  This then causes the
client to report a 1000ms delay and a 500ms offset and then step time
that 500ms offset.  It always seems to be a 500ms step like this one.

 4 Jul 09:42:45 xntpd[21466]: time reset (step) 0.499690 s
0 Wed Jul  4 09:42:46 CUT 2007 systime: maximum slew: 0.098999998,
remainder = 0.400606709
0 Wed Jul  4 09:42:47 CUT 2007 systime: maximum slew: 0.098999998,
remainder = 0.301600965
0 Wed Jul  4 09:42:48 CUT 2007 systime: maximum slew: 0.098999998,
remainder = 0.202595221
0 Wed Jul  4 09:42:49 CUT 2007 systime: maximum slew: 0.098999998,
remainder = 0.103589477
0 Wed Jul  4 09:42:51 CUT 2007 systime: maximum slew: 0.098999998,
remainder = 0.004583733

       remote              refid      st t when poll reach   delay
offset    disp
==============================================================================
 processor1      LOCAL(0)         2 -   24   16    0  1000.41  499.690
16000.0

My NTP server seems to wake up once every second and do some action.
I'm assuming this is to adjust for the drift factor.  My server is in
the middle of a dispatch to do this processing when the I/O interrupt
is triggered by the query from the my NTP client.  But a reply is not
sent out on the socket until one second later when we wake up again
for our periodic slew adjustment.







More information about the questions mailing list