My old white light NAS drives have an ntp daemon running:
/usr/sbin/ntpdated
This keeps the time updated from a timeserver I have on our internal network
(it is itself syncing to various sources, and can get through the firewall…it
has dual network cards, too much information…)
My new WD Mybook Live 3TB does not even have this program. I’ve just
spent a few hours trolling through the system trying figure out where it is.
Interestingly, there is an hourly cron job set up to clean up old old ntp log
files based on the ntp.conf file log directory, even though the ntp.conf
file is not used, and no log files are generated because there is not ntpdated
being run. What fun.
Looks like it syncs on boot, which is nice, but we leave these drives up and
running all the time, and they are on UPS. I believe one time they were up
for 270 days straight. So syncing on reboot really doesn’t do it.
Wondered if anyone knows what happened here?
Apparently the checkbox beside “NTP service” is misleading.
I agree – and I filed a bug report over that.
The clock is horribly inaccurate:
MyBookLive:~# uptime
09:00:10 up 17 days, 7:58, 1 user, load average: 1.56, 1.44, 1.38
MyBookLive:~# date
Sun Mar 24 09:00:21 CDT 2013
MyBookLive:~# ntpdate pool.ntp.org
25 Mar 18:25:13 ntpdate[30509]: step time server 96.44.142.5 offset 120242.434048 sec
Only 17 days up, and the clock is more than 30 hours wrong!! :laughing:
…about all you can do is set up a crontab job to run ntpdate ever hour or two…
WD support directed me to the GUI interface : Support and button ‘Create and save system report’.
Basically dumps out all the log files etc.
So, yes there is a cron job that runs something named ntp daily at 3 a.m.
However if you look at the script ntp in /etc/cron.daily it appears to be an ntpdate log cleanup script.
Which is ironic since ntpdate never runs that I can see.
I ended up creating a script in /etc/cron.hourly/ntp that looks like this:
#!/bin/sh
/usr/sbin/ntpdate-debian -s
Seems ok for the moment. Though updating 0.3 seconds every hour, so yes
the clock seems pretty inaccurate.
I have a ticket open with support, maybe I can convince them they missed
something here…
I see that ntpdate is deprecated and ntpdated is the newer thing to use,
but ntpdated does not exist on the system.
A bit more info…
/etc/init.d/ntpdate runs at initialization
In this script the following happens:
start snippet
invoke-rc.d --quiet ntp stop || true
/usr/sbin/ntpdate-debian -s $OPTS || :
invoke-rc.d --quiet ntp start || true
end snippet
It is trying to stop then start non-existant service ‘ntp’
In between it does a call to ntpdate-debian.
ntpdate-debian works to set the time (it called ntpdate) but
is not a service, one time deal
ntp service does not exist. If you run the line above without the --quiet parm:
MYBOOK:/etc/init.d# invoke-rc.d ntp start
invoke-rc.d: unknown initscript, /etc/init.d/ntp not found.
And indeed there is not ntp script in /etc/init.d