On Thu, Oct 07, 2010 at 06:27:56PM +0200, Miroslav Lichvar wrote: > The most promising idea I came up so far is to adjust the interval > based on number of samples kept in sourcestats, so it remains close to > a fixed number. This would be configurable, a higher value would mean > chrony should try to keep the buffer more filled and thus have better > accuracy. The default would be set so chrony uses less or same > resources than ntpd with same minpoll and maxpoll, while keeping > better accuracy.
This is now in git. A very primitive control system is used, but it seems to work quite nicely. Here are some results from simulations: http://mlichvar.fedorapeople.org/clknetsim/test9_ntp_offset.png http://mlichvar.fedorapeople.org/clknetsim/test9_ntp_poll.png http://mlichvar.fedorapeople.org/clknetsim/test9_chrony_offset.png http://mlichvar.fedorapeople.org/clknetsim/test9_chrony_poll.png http://mlichvar.fedorapeople.org/clknetsim/test9_start.png Please test and report if chrony is using shorter poll interval than you think it should. -- Miroslav Lichvar --- To unsubscribe email chrony-dev-requ...@chrony.tuxfamily.org with "unsubscribe" in the subject. For help email chrony-dev-requ...@chrony.tuxfamily.org with "help" in the subject. Trouble? Email listmas...@chrony.tuxfamily.org.