I've been trying to get NTP to synchronize our workstations and servers on an isolated network (not connected to anything in the outside world). The two Linux servers on the network will not synchronize with the NTP server and keep their clocks accurate. I have managed to sync Windows XP workstations to the NTP server.

I've run ntpd -q to do an initial synchronization, but that fails. Running ntptime 192.168.40.40 successfully updates the clock, however.

Running ntpd in debug mode shows requests being sent and responses received (sorry, I can't paste it here because I'd have to type it all out by hand). When I run ntpd -q, 16 send/receive statements are shown, but then ntpd exits saying "no reply; clock not set."

My ntp.conf file has:

server 192.168.40.40
driftfile /var/lib/ntp/drift

In ntpq I get:

ntpq> pe
remote refid st t when poll reach delay offset jitter
=====================================================================
192.168.40.40 192.168.40.1 2 u 32 64 377 0.174 -5.062 14.726

ntpq> as
ind assID status conf reach auth condition last_event cnt
==========================================================
1 15804 9014 yes yes none reject reachable 1

ntpq> rv 15804
assID=15804 status=9014 reach, conf, 1 event, event_reach,
srcadr=192.168.40.40, srcport=123, dstadr=192.168.40.72, dstport=123, leap=00, stratum=2, precision=-6, rootdelay=31.250, rootdispersion=11072.174, refid=192.168.40.1, reach=377, unreach=0, hmode=3, pmode=4, hpoll=6, ppoll=6, flash=00 ok, keyid=0, ttl=0, offest=-3.552, delay=0.174, dispersion=455.268, jitter=9.651

On the other server, running the rv command in ntpq results in similar output, but with flash=400 indicating peer distance exceeded, which I find odd since the NTP server is only separated by a switch, so I don't understand why ntpd would decide it's too far away.

So, if anyone can help me figure out how to get my setup to accept the NTP server I set up as valid, I would appreciate it. Thanks.

Could you run and post the results of the following:

I seen ntpd synchronise against the server running against the local host instead of the remote server. This is because the stratum assigned to the remote host is higher; thus, not it's trusted.

I seem to remember that ntp will stop accepting updates from an ntp server if that server hasn't been able to sync from a higher (smaller number) stratum server, it may be because your servers are isolated that you're having the issue.

hth

ntpq -pn has the same output as the interactive peer command (i.e. same output as the output below the line that says ntpq> pe above).

To lay out the whole setup, (and I know this is more complicated than it needs to be, but there's a good reason) we have a "stratum 1" NTP server (whose time source is its local clock) at 192.168.40.1 that the stratum 2 NTP server at 192.168.40.40 syncs to. Then all of our workstations and servers are supposed to sync to 192.168.40.40. I can't get the two Linux machines to sync to .40 or .1. I have left them running overnight just to make sure it wasn't that I was being impatient, and they still hadn't synced when I came in in the morning.

I configured one of my Linux servers to synchronize with its local clock, then set the other server to synchronize to it. Works perfectly, even with the server set at stratum 11. The main difference I can see is that rootdispersion (from ntpq command peers) is >10000 when I sync to the .40 or .1 NTP server, but is <15 when I sync to the Linux server. Maybe my .40 or .1 servers have a bad configuration that is causing the problem. Sound reasonable?

hi csmith7383,
i have the same issue, how did you solve this?


thanks,
gchibas