[ntp:questions] Re: linux ntp problem

Maarten Wiltink maarten at kittensandcats.net
Mon Nov 10 22:08:54 UTC 2003

"Mike Zupan" <mzupan at meso.com> wrote in message
news:9d302332.0311101300.c0e4155 at posting.google.com...
> Here is the ntpq -p on the server

(Jitter column deleted)

> [root at monitor ntp]# ntpq -p
>      remote           refid      st t when poll reach   delay   offset
> ======================================================================
>  ntp.alaska.edu  tick.usnogps.na  2 u    6   64    1  137.891  177.946
>  clock.isc.org   gps.laguna.vix.  2 u   16   64    1   89.710  183.408
>  montpelier.ilan          1 u   17   64    1  125.177  167.961
>  nist1.aol-ca.tr .ACTS.           1 u   14   64    1   86.821  188.877
>  LOCAL(0)        LOCAL(0)        13 l   21   64    0    0.000    0.000

<And I can't sync to it.>

That's correct. This table was captured less than a minute after starting
ntpd. You can tell from the reach column that one packet has been received
from each server, and from the when column that this happened 6/16/17/14
seconds ago.

The daemon needs several (at least 5 coherent ones, I think) samples from
a server to even start believing it, and will not sync until that time.
It will not serve time until it considers itself synced.

That's important. It will not serve time until it considers itself synced.

When it does, punctuation will appear in the first character column, even
before the association (server) names. A minus sign means bad, a plus
sign means good, an asterisk means synced to that server.

Other than this, congratulations. You've managed to get things mostly
right, down to the local clock at high stratum. Note that your third
server appears to lack a time reference.

Maarten Wiltink

More information about the questions mailing list