Solution: ntpdate[4933]: no server suitable for synchronization found

Our local setup consists of a single ntp server which synchronizes time from asia pool of ntp.org ([0-2].asia.pool.ntp.org) and several ntp servers in the intranet which synchronizes from the above server. This setup ensures identical clock on all the machines on the intranet as well as reduces load on ntp.org servers. However we faced with an infuriating problem in getting the clients to sysnchronize with the local server. In some machines we would get this error message in log ( /var/log/messages )during startup:
“ntpdate[4933]: no server suitable for synchronization found”

Restarting ntpd would give this error message:

[root@jaguar ntp]# /sbin/service ntpd start
ntpd: Synchronizing with time server:            [FAILED]
Starting ntpd:                                   [  OK  ]

The second problem was that ntp server was taking a long time to come up (with failure message as described above) thereby slowing down booting time. After much investigation and tinkering (not to mention futile searches with Google), I accidentally came across a simple solution.

Open /etc/ntp/step-tickers and replace the list with the name of your local ntp server (from which you want to synchronize).

Note: You can also do the same with /etc/ntp/ntpservers for your satisfaction. However I didn’t see the file being actually used in Fedora Core 6. ntp server also worked without this file being present.

Changing /etc/ntp/step-tickers did the trick. Now ntp server comes up in a fraction of a second and without any error messages. Problem solved.

The end result is:

[root@jaguar ntp]# /sbin/service ntpd restart
Shutting down ntpd:                         [  OK  ]
ntpd: Synchronizing with time server:       [  OK  ]
Starting ntpd:                              [  OK  ]