In article <[EMAIL PROTECTED]>, [EMAIL PROTECTED] wrote: > We have a machine which is running NTP version 4 and it's in sync with > its time server (a '*' appears in the output of ntpq -p; as we know > the server marked * is the one which is chosen for synchronization).
A combination of those with * and + are used for synchronisation. > Now the problem > is that even if the machine is synchronizing with its time server all > the time, the time difference between the two doesn't decrease. The > difference is around 2 min and it's been like this for one month or > so. I expect the machines to be within a few seconds of each other. ntpd is in difficulties if the times differ by more than 128ms. Have you or your OS vendor used any unusual options? Have you been overzealous with restrict? What does syslog say? You should have observed time steps with a standard configuration. If the system does step and then immediately steps back to 2 minutes, you have another time synchronisation application running, with a different idea of the time. > The only reason I can think of is that the clock crystal is behaving > wildly. This would only be a factor if the system was configured not to step at all and the clock error was exactly 500ms. With no clock error, and slew only, the clock should recover 46 seconds per day. _______________________________________________ questions mailing list [email protected] https://lists.ntp.isc.org/mailman/listinfo/questions
