NTP time not updated periodically
Posted: Mon Jan 15, 2018 9:24 pm
Eric, folks upgrading to newest firmware often will likely not see this as the switch sets its time on each reboot. I've just noticed that some of my 1.3.9 switches are off by 10 minutes after an uptime of more than a year. Switches with 1.4.7 haven't run that long obviously and the deviation is less, but still some 2 minutes in some cases. Obviously, with 1.4.8 and newer, you could probably not see the effect as this firmware is too new for drift to add up and become noticeable.
So I successfully tried
ntpclient -s -D -h <ntp.server.name.or.ip>
which tells me that the connection to our NTP server is okay and that the switches (tested up to 1.4.7) do not use ntpclient periodically. /usr/bin/monitor initiates ntpclient only if the year is 1970 (after boot) and there is no crond around to run it periodically. A quick fix would be to start ntpclient in the background from /usr/bin/monitor once a day, i.e. every 8640th cycle.
So I successfully tried
ntpclient -s -D -h <ntp.server.name.or.ip>
which tells me that the connection to our NTP server is okay and that the switches (tested up to 1.4.7) do not use ntpclient periodically. /usr/bin/monitor initiates ntpclient only if the year is 1970 (after boot) and there is no crond around to run it periodically. A quick fix would be to start ntpclient in the background from /usr/bin/monitor once a day, i.e. every 8640th cycle.