what was up with ntp again?

[Daniel J. Luke]

> On Mar 23, 2015, at 2:36 PM, Brandon Allbery <allbery.b at gmail.com> wrote:
> On Mon, Mar 23, 2015 at 2:27 PM, Christopher Jones <jonesc at hep.phy.cam.ac.uk> wrote:
> yes, I did reboot around then yesterday. Perhaps I mis understood Brendan’s mail though but that seemed to suggest even a reboot wouldn’t update it. But if it does then fine, (non)mystery solved…
> 
> I was under the impression a reboot wasn't enough. Maybe Apple has changed this so it's regenerated at each reboot... but that would leave us back at the question of why the clock drifts so badly. Does everyone who sees that *really* not reboot often?

I think there was a darwin-dev post about this a while back and it's something like the way ntpd updates the drift file, the mod time doesn't get set/updated (maybe until it exits?) with the exception of ntpd startup (so effectively when you reboot or stop/start ntpd it gets updated). pacemaker thus doesn't see updates (since it looks for the mod time).

I believe xnu doesn't expose as much clock api as other *bsds do, too (which is why ntimed doesn't work on Mac OS X yet, and why apple's version of ntpd is pretty heavily patched).

--
Daniel J. Luke                                                                   
+========================================================+                        
| *---------------- dluke at geeklair.net ----------------* |                          
| *-------------- http://www.geeklair.net -------------* |                          
+========================================================+                        
|   Opinions expressed are mine and do not necessarily   |                          
|          reflect the opinions of my employer.          |                          
+========================================================+