Marc Fromm wrote:
If I restart the ntpd service the time is updated to the correct time: /sbin/service ntpd restart
I.E. /etc/init.d/ntpd restart which, although a deprecated procedure, will probably run ntpdate.
If I run the command below the time is updated to the correct time: ntpdate -u time-nw.nist.gov If I don't do any of the above my time is never updated and the server becomes hours behind. How or when does ntpd automatically update the time?
When the measured error is between 128ms and 1,000 seconds and has persisted for about 15 minutes. If it is less, ntpd tweaks the effective clock frequency every second, or asks the kernel to do so every tick, so as to slowly (order of hours) converge the clock onto the correct time. If it is greater, ntpd aborts, as its idea of the time is unsafe.
Note, selective failures of ntpd relative to some options of ntpdate can be due to firewalling of UDP port 123.
The standard ntpq diagnostics (peers, rv 0, assoc, rv for every association id) and the syslog entries would be helpful, together with any local changes to the configuration file. You should leave the sytem running for at least 30 minutes before running ntpq (less is possible, but 30 minutes should cover all likely settings).
As service is part of RedHat, not ntpd, and /etc/init.d/ntpd will also be part of RedHat, I assume you are using some version of RedHat, but you really should have told us which, and which version of ntpd.
_______________________________________________ questions mailing list [email protected] http://lists.ntp.org/listinfo/questions
