[ntp:questions] how to have offset < 1ms

Rob nomail at example.com
Wed Apr 14 17:54:50 UTC 2010


lhommedumatch <ludocluba at yahoo.com> wrote:
> Hi,
>
> Some news:
>      remote           refid      st t when poll reach   delay
> offset  jitter
> ==============================================================================
> *ntpgmtaceb      .1PPS.           1 u  734 1024  377   29.779
> 1.182   0.228
>
> So still this problem in my LAN => because delay is too high
>
> Don't know if it helps but a ping to ntpgmtaceb:
> PING ntpgmtaceb (192.168.21.5) 56(84) bytes of data.
> 64 bytes from ntpgmtaceb (192.168.21.5): icmp_seq=1 ttl=63 time=0.128
> ms
> 64 bytes from ntpgmtaceb (192.168.21.5): icmp_seq=2 ttl=63 time=0.137
> ms
> 64 bytes from ntpgmtaceb (192.168.21.5): icmp_seq=3 ttl=63 time=0.155
> ms
> 64 bytes from ntpgmtaceb (192.168.21.5): icmp_seq=4 ttl=63 time=0.139
> ms
>
> My PC is connected to a CISCO Catalyst 4510R as well as the ntp server
> clock.

Can you try to setup another NTP client system, a completely different
computer connected at another point in the network, and check what the
delay is there?  Maybe such a system already exists elsewhere on the
ship.  It seems that there is a problem somewhere but it is unclear where.

The fact that ping works ok probably indicates the problem is not in
the network, or maybe it only occurs when the system is idle.  When you
start pinging, something wakes up and stays alive while you ping, but
when it is sleeping (e.g. some "powersave" mode) the delay is longer.

Is the system this NTP is running on active on the network?
Maybe you can keep the ping running in the background and check if things
work better when you do that.

What is this "ntpgmtaceb"?   A special NTP device, a general purpose
system that may be heavily loaded?  Is it running Windows or some Unix
variant?




More information about the questions mailing list