xntp time reset in syslog continuously

[Ho_5]

Hi all,

About DNS:
I found the next messages appear in the syslog.log (almost every 3 hours).
My question is, is that normal?? If not what I have to do??

Thanks.

Regards,

John

Apr 2 18:41:31 dns xntpd[907]: time reset (step) -0.201763 s
Apr 2 18:41:31 dns xntpd[907]: synchronisation lost
Apr 2 18:46:51 dns xntpd[907]: synchronized to 192.168.20.17, stratum=3
Apr 2 18:46:51 dns xntpd[907]: synchronized to 192.168.20.18, stratum=2

[Michael Steele_2]

When your client gets more than +128 seconds out of sync with the NTP server it can't reset. You probably have some network congestion problems. I've pasted the definition below:

Synchronization lost.

This message indicates that NTP has cleared all of the statistics registers and has started evaluating all available time servers and choosing the "best" one. This message appears whenever a step adjustment (greater than 128 milliseconds) is made ...it probably means that you have network congestion problems. To review this, do the following:

Run ntpq -p

Examine the dispersion statistics.

See link:

http://docs.hp.com/cgi-bin/fsearch/framedisplay?top=/hpux/onlinedocs/B2355-90685/B2355-90685_top.html&con=/hpux/onlinedocs/B2355-90685/00/00/66-con.html&toc=/hpux/onlinedocs/B2355-90685/00/00/66-toc.html&searchterms=syslog%7clost%7cxntpd&queryid=20030420-084925

[Steven E. Protter]

You are connected to a Microsoft Time Source that does not comply with standards that HP-UX and any other form of Unix needs to work properly.

The test above(reply 1) may adjust your Unix clock, but will generate and error message when you run it.

The NT time source needs to be upgraded to a standarized version, or perhaps you can go straight to the Internet for time(not likely).

Here is a link to forward to the NT admin.

Public Server List:
http://www.eecis.udel.edu/~mills/ntp/servers.html

Dr. David Mills, the Guru of ntp's home page.

http://www.eecis.udel.edu/~mills/

Somewhere buried in there is a version of software for NT Workstation/Server/Windows 2K.

It is my understanding that Microsoft began complying with standards with Windows 2000. I get my 5 system times from an W2K Pro server

SEP

[John Bolene]

You say DNS and then talk about xntpd.

I assume DNS is a server name.

Xntpd resets the time every 3 hours and this is normal.

What may not be normal but is certainly acceptable is you have 20.17 (stratum 3) pointing to 20.18 (stratum 2) which I assume points to the Internet.

[Bill Douglass]

What is happening is xntpd is finding the system clock off by more than 128ms. When that happens, it steps the clock, or simply changes the time to what the time source claims is accurate.

Normally, xntpd tried to keep the clock accurate by slewing, whereby it slows or speeds up the system clock gradually until it come into compliance. This is less disruptive, but take longer than stepping the clock. So the daemon slews the clock if the difference is under 128ms, and steps it otherwise.

One cause is a poor hardware clock, that drifts out of sync rapidly. Another is a faulty time source. It also appears that high latency can affect xntpd's ability to keep the clock sync'd, although I've never run into a situation like that, and I've had servers around the globe syncing to a central source.

Since you have two time servers listed, I'm guessing that you have a bad system clock on your host. Depending on how your system is used, this may or may not be a problem. The clock will be kept reasonably close, but there will me minute disruptions like the kind you indicated.

You could also verify with the admin of the time servers that they are indeed locked to a stratum 1 source, as well as ping the link from your host to the servers to check for unusually high latency.