Penned by Stuart Henderson on 20170209 18:57.59, we have:
| On 2017-02-09, Eric Brown <[email protected]> wrote:
| > Dear List,
| >
| > I've recently learned (and discovered) that time in VM's is tricky
| > business. I'm looking for the least stupid way to keep any semblance of
| > time in vmd instances while I hungrily await a "correct solution" to
| > descend from the heavens.
| >
| > I've disabled openntpd, installed ntp package (but not its daemon). Now
| > I am running ntpdate every minute from cron. It seems to keep the
| > clock, well, within a minute.
| >
| > Can anyone think of a better solution to this problem?
|
| Not a hugely better solution, but rdate(8) is in base, so at least you
| don't need the ntp package..
I could be wrong, but seeing this in my guest:
sysctl hw.sensors
hw.sensors.vmmci0.timedelta0=-7127.806752 secs, OK, Mon Feb 27 11:02:53.434
and this in ntpctl output:
sensor
wt gd st next poll offset correction
vmmci0
1 1 0 8s 15s 81357.122ms 0.000ms
suggests to me that the time passed to the guest is used as a timedelta sensor
using the native ntpd, no need for network traffic!
--
Todd T. Fries . http://todd.fries.net/pgp.txt . @unix2mars . github:toddfries
