On 4/11/06, Randy B <[EMAIL PROTECTED]> wrote: > > On 4/5/06, Vivek Khera <[EMAIL PROTECTED]> wrote: > > ISC's ntp is well known and understood and considered very accurate. > > I see no other choice. > > > > After Running OpenNTP for a while now, I feel less uncomfortable with it - > after the first 12 hours or so, the clock swings (+/-12ms) evened out, and > it's staying quite comfortably within +/- 2-3ms with very little jitter. In > the following output of 'ntpq -c peers', the system in question is > 'balrog-priv'; note the odd reference clock - I think that's an artifact of > the minimal implementation that doesn't allow that level of querying. In > fact, for the most part it seems to stay well within 1ms (it refers to > no-such-system, dies-irae, and the local system I'm querying from).
I might have to give it a try on my boxes (running OpenBSD) at work. ISC ntpd can't keep the clock sync'd when you have lots of jitter (which we do - due to traffic loads on the box trying to be sync'd). It eventually gives up attempting to sync the clock. A ntp daemon "that works" it better than an ntp daemon that when it works, is millisecond precise, but "doesn't work". FWIW, when a carp pair gets it's dates out of sync by more than a second or two, hilarity ensues and it's _not_ a pretty sight (that was my joy first thing yesterday morning). --Bill --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]