On 2010-02-04, David Lord <[email protected]> wrote: > unruh wrote: >> On 2010-02-04, David Lord <[email protected]> wrote: >>> unruh wrote: >>>> On 2010-02-03, David Lord <[email protected]> wrote: >>>>> Hal Murray wrote: >> ... >>>> Three possibilities. >>>> a) if you run linux/bsd, run chrony. It corrects for temp drifts much >>>> much faster. >>> That's out at moment as chrony isn't sufficiently stable and >>> backwards compatible. From my tests, I'd say offsets are lower >> >> Which version of chrony? It has received extensive alteration in the >> past few months, and chrony 1.24 is just coming out. > > 1.23, but my brief checks with 1.24pre indicated a further improvement. > >>> than ntpd by at least a factor of 3. Problem is monitoring >>> with chronyc requiring same version on remote as chronyd on >> >> Yes, because of all the changes the chronyc and chronyd must be the same >> version. >> > > That's a bit too much of a complication for me. I'd need to > setup build systems for older versions of NetBSD I'm > phasing out but might still be in use for a while.
?? The above was a comment on the versions of chronyd and chronyc, not of NetBSD. You can run the new chrony on the older versions of BSD, or are you having trouble compiling them? > >>> server. I've also had complete lockups when attempting to peer >>> chrony with ntpd. >> >> That is much more serious. Note that I have been running 10 chrony >> machines from one ntp server for the past 2 years and have never seen a >> lockup. Can you give more information, at least to >> [email protected] > > When I'm getting mails from new list I'll resend my my > original report but I've moved desktops and notebooks back > to ntpd. I've found using ntpd -q and having iburst on one > or more sources, I can get to within a few ms within 5-10 > minutes of startup. On NetBSD chrony can't do any better at > startup as RTC features aren't available so far as I can > tell. No, but you can use the initslew directive in chrony.conf to tell the machine to step initially rather than slew initially if the time is out by more than x seconds as determined from the intial ntp packets. > > ..... > > David _______________________________________________ questions mailing list [email protected] http://lists.ntp.org/listinfo/questions
