[Pool] Score is drop in cycles
ntp-list at mallos.nl
Wed Feb 19 13:59:23 UTC 2014
Hi Mingda Wang,
On 02/19/2014 10:11 AM, lst_hoe02 at 79365-rhs.de wrote:
> Zitat von 王明达 <mingdawang at 163.com>:
>> I try add my ntp server to pool, but I found it's score is drop in cycles , I
>> think it's related to traffic,but the system's load is very low, I didn't find
>> any error in log, configure file pls see the attachment. How to find the
>> bottleneck of the system? Looking forward to your reply.
> You have a very long round trip time to your peers you get time from. Delay values
> should be below 100ms in today networks, you should maybe look for "closer" NTP peers.
I saw one of your peers is located in the Netherlands, Europe. That explains the
round-trip times. Having stratum 1 peers is nice, but nearby stratum 2 or even a
few stratum 3 servers might be a better choice (if I remember right, the pool
monitoring system accepts up to stratum 5 servers in the pool). One way to obtain
them is via 'host 0.asia.pool.ntp.org' (and '1...', '2...', '3...') multiple times
at different times (to allow for DNS-rotation). Then check with 'ntpdate -qp 1
<<ip>>' the stratum and with traceroute (or traceroute6 for IPv6) the number of
hops. You can also check their scores and see if their score is stable.
> Furthermore the Jitter values are bad which could lead to the strange score plot.
I just looked at the CSV data for the server. The time is reasonable accurate, so
the high round-trip time and jitter are not really a problem. But, you have
frequently no response at all. This causes the drops in the score (at this time).
You need to figure out why the NTP-requests from the monitoring server do not reach
your server, or why the answers don't get back. You may have bad internet
connectivity or packets may be dropped due to high traffic somewhere in between (or
something else...) Perhaps you can check packets from/to the monitoring server with
something like tcpdump.
Hope this helps!
More information about the pool