Ntp -- getting it to work

[Brandon Allbery]

On Sun, Nov 23, 2014 at 10:41 PM, Brandon Allbery <allbery.b at gmail.com>
wrote:

> FWIW I disabled pacemaker here and it solved the problem of ntpd not being
> able to adjust the clock (pacemaker was apparently undoing it). But it is
> still not *syncing* the clock after the initial sync, it is building up
> delays over time and then fixing them all at once, like it thinks it's an
> ntpdate cron job. Apple's config doesn't seem to do anything to make this
> happen....
>

Ok, the discussion linked by Lawrence suggests that this is expected;
Apple's ntpd syncs once, creates ntp.drift if it does not exist, and
effectlvely does nothing afterward, with pacemaker adjusting the time based
on that initial ntp.drift. Which does not work: you really need ntp.drift
to be calibrated over a period of at least several days before it is
reliable, *and* there is always the possibility that hardware behavior
changes over time especially as new hardware is "broken in".

So I will kill off Apple's foo and use ntpd from ports, I think. It may not
"save battery", but last I checked iMacs did not have batteries. :)

-- 
brandon s allbery kf8nh                               sine nomine associates
allbery.b at gmail.com                                  ballbery at sinenomine.net
unix, openafs, kerberos, infrastructure, xmonad        http://sinenomine.net
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.macosforge.org/pipermail/macports-users/attachments/20141123/1c3584e7/attachment.html>