| Re: [chrony-dev] rtc under chrony has large fluctuations on some machines. |
[ Thread Index |
Date Index
| More chrony.tuxfamily.org/chrony-dev Archives
]
On Sat, Nov 28, 2009 at 11:45:56AM -0800, Bill Unruh wrote:
> have been looking at the offsets that chrony is reporting for the rtc
> measurements. On both of the machines on which I have the new git chrony (
> with the rtc patch) and one of the old machines ( without the patch) chrony is
> reporting large fluctuations ( of order 10ms) in the offset readings of the
> rtc clock. On other of the machines, the fluctuations are more like
> 100usec. I am not sure if this is due to the recent change in the
> way chrony is reading
> the rtc ( ie doing another select wait and then a read, instead of two reads
> with the second on a blocking read). I will do some more
> investigation to try to see what is going on.
Could it be that RTC emulated by HPET is not that good?
I did a quick check on two machines, according to "grep hpet /proc/timer_list"
one has HPET and the other doesn't. RTC on the machine with HPET is as
bad as you describe and much better on the one without.
Maybe adding "hpet=disable" to kernel parameters will fix it. I can't
reboot the machine right now to try it.
--
Miroslav Lichvar
---
To unsubscribe email chrony-dev-request@xxxxxxxxxxxxxxxxxxxx with "unsubscribe" in the subject.
For help email chrony-dev-request@xxxxxxxxxxxxxxxxxxxx with "help" in the subject.
Trouble? Email listmaster@xxxxxxxxxxxxxxxxxxxx.