ntp drift too high
The Network Time Protocol (NTP) daemon (
ntpd) is responsible for keeping system
time synchronised with standard time servers on the Internet.
ntpd.drift file records the latest estimate of clock frequency error. If this value gets too high, this can be
an indicator of large error between the system clock and the real time.
If the system clock is out of sync on a Vmware host, first make sure the synchronisation between guest and host has been disabled. This is enabled by default in the image configuration and could cause issues when ntpd is running on the same box.
To check the status, enable or disable Vmware periodic time synchronization, we can use the
$ sudo vmware-toolbox-cmd timesync status Enabled $ sudo vmware-toolbox-cmd timesync disable Disabled $ sudo vmware-toolbox-cmd timesync status Disabled
If the Vmware timesync is disabled, we can use fabric-scripts tasks to help resynchronise the system clock:
fab $environment -H jumpbox-1.management ntp.status fab $environment -H jumpbox-1.management ntp.resync
The fab script will try to slew the time offset, which means continually adding/subtracting little bits of time until the clock is in sync. This is in contrast to a step change, where the clock’s time is just changed. Step changes can cause - for example - log timestamp inconsistencies.
According to the
ntpdateman page, the slew forced by the
-Bflag can take hours to gradually take effect.
ntpdatefunctionality has been made available in the
ntpdprogram. To resync an offset bigger than 1000, you can run
sudo service ntp stop; sudo ntpd -gq; sudo service ntp start.