Our NTP servers are set up as a peer group looking out at us.pool.ntp.org. Our internal servers are set up to look at the local group. These internal servers are also peered to each other (Oracle RAC servers need this I am told. . .). Three servers do not look at the nodes specified in the ntp.conf under the 'server' directive. They look to a nameserver that had previously been used as a time server. This one server doesnt look anywhere.
<p>
The /var/log/messages/ entries:
Feb 9 13:06:22 myserver-01 ntpd: ntpd shutdown succeeded
Feb 9 13:06:27 myserver-01 ntpdate[4417]: no server suitable for synchronization found
Feb 9 13:06:27 myserver-01 ntpd: failed
Feb 9 13:06:27 myserver-01 ntpd[4774]: ntpd 4.2.0a@1.1190-r Thu Oct 5 04:11:33 EDT 2006 (1)
Feb 9 13:06:27 myserver-01 ntpd[4774]: precision = 1.000 usec
Feb 9 13:06:27 myserver-01 ntpd[4774]: Listening on interface wildcard, 0.0.0.0#123
Feb 9 13:06:27 myserver-01 ntpd[4774]: Listening on interface wildcard, ::#123
Feb 9 13:06:27 myserver-01 ntpd[4774]: Listening on interface lo, 127.0.0.1#123
Feb 9 13:06:27 myserver-01 ntpd[4774]: Listening on interface eth0, 10.200.23.31#123
Feb 9 13:06:27 myserver-01 ntpd[4774]: Listening on interface eth0:1, 10.200.23.41#123
Feb 9 13:06:27 myserver-01 ntpd[4774]: Listening on interface eth1, 192.168.23.31#123
Feb 9 13:06:27 myserver-01 ntpd[4774]: kernel time sync status 0040
<p>
Why would none of these servers look to the time server specified in the ntp.conf? It is attached (for syntax review) with identifiers falsified.

Attached Files

ntp.conf.txt (1.0 KB, 27 views)

well nothing there says that it's not looking at those servers, merely that nothing useful was found, which is different.

run "ntpdate -d my.ser.ver.ip" to do a one shot debugging query of a server, and see what that says. Additionally on your local servers run "ntpp -pn" to check the status of them. If they are flapping around without any synchonrization then nothing is going to use them - ntp is cleverer than that, and more annoying as a result.

Have you checked your init scripts to see if someone has manually specified a server as the argument to ntpdate?

Attached the most recent result of ntpq -p. Do these results indicate that the ntp service is running properly?

Attached Files

ntpq.txt (857 Bytes, 26 views)

you can't tell? well yes it shows in the second line (with the *) that you're happily synced to a strata 2 server online, so in turn will be providing a strata 3 service internally. So if you can't do a one shot ntpdate command from the client to that server, i'd be expecting network issues in the way. local firewall maybe?