NTP error -synchronisation lost

[Chapaya]

Hi ,
my linux machine losts connection with ntp server :
Oct 14 15:19:50 hp1 ntpd: ntpd startup succeeded
Oct 14 15:20:24 hp1 ntpd[23579]: kernel time discipline status change 41
Oct 14 15:21:26 hp1 ntpd[23579]: synchronisation lost
Oct 14 15:23:16 hp1 ntpd[23579]: kernel time discipline status change 1
[root@hp1 log]# ntpq -p
remote refid st t when poll reach delay offset jitter
==============================================================================
*LOCAL(0) LOCAL(0) 5 l 16 64 377 0.000 0.000 0.004
ntp1 10.203.205.55 2 u 53 64 377 0.393 7807.14 1031.00
[root@hp1 log]#

[Steven E. Protter]

Shalom,

Make a manual adjustment in the system to get it within an hour or so of the time server. ntp won't work if the time is too far off.

SEP

[Matti_Kurkela · Accepted solution]

Your "jitter" value is 1031 milliseconds, i.e. 1.031 seconds. The value measures how widely the network round-trip time seems to vary.

NTP tries to get your machine's clock synchronized to within 128 milliseconds of the NTP server's idea of "real time". With a round-trip delay jitter as high as 1031 milliseconds, that's quite hard to do.
Even a long round-trip delay is not a problem for ntpd, if the delay is stable. A varying delay causes problems.

Your network might be so congested it's causing the large jitter. If other hosts in the same network segment can use the same NTP server successfully, this is *not* the problem.

The other possibility is that your machine's system clock has problems. What's the model of your machine and what's your Linux kernel version number? Sometimes the power-saving modes in the latest processors will confuse Linux's timekeeping functions, if the kernel is not up to date.

MK

[Chapaya]

Hi Matti ,

Is "jitter" value is configurable or value is been measured by ntp ?

BR,
Eran.

[Matti_Kurkela]

It is measured, like the "delay" and "offset" values.

Your "delay" value (the actual time taken for a message to reach the NTP server and a response to come back) looks good, but the "jitter" value indicates the delay seems to be varying rather widely.

Either there's congestion in your network, causing significant variance in the packet travel times, or there's something wrong with your machine's local clock, causing the delay measurements to have widely different results each time.

MK

[Chapaya]

other hosts in same segment are synchronized well , i will check bios setup .

[Matti_Kurkela]

You might wish to check you're up to date with kernel patches, so that any power-saving capabilities and hardware peculiarities your system might have are fully supported.

If the server used to synchronize well, but now has frequent problems, there might be a hardware fault.

MK