[ntp:questions] Re: NTP: time not synchronized

Maarten Wiltink maarten at kittensandcats.net
Wed Sep 6 08:36:56 UTC 2006


"pdp" <pasdepoil at hotmail.fr> wrote in message
news:1157461828.336482.64560 at m79g2000cwm.googlegroups.com...
[...]
> So we are using ntpd 4.1.72 on every machine.

If it works, it works. If it doesn't, people are going to tell you
to upgrade just to make sure you're not running into bugs long since
fixed.


> I will give you the result of the ntpq -p command for the server and
> the 2 clients (one OK and a failed one)
>
>                failed client:         OK client:            server:
> remote         01S.ccadomain.com      01S.ccadomain.com     LOCAL(0)
> refid          0.0.0.0                LCL                   LCL
> stratum        16                     1                     0
> type           unicast server         unicast server        Local Clock
> when           -                      114                   25
> poll           512                    256                   64
> reach          0                      377                   377
> delay          0.000                  44.713                0.000
> offset         0.000                  -5.545                0.000
> jitter         4000.00                80.834                0.001

This is an original and actually rather nice way of presenting ntpq -p
output.

The failed client has a reach of zero, meaning that it never talked to
the server at all. It may have talked _at_ the server, but it never got
any replies.

Don't put your local clock at stratum zero. It isn't. If you ever
acquire an Internet connection or a real refclock, you don't want your
local clock to even have a chance of drowning it out. Make it ten now,
so you can't forget to later.

Groetjes,
Maarten Wiltink





More information about the questions mailing list