hebo@HP850.MBARI.ORG (Bob Herlien) (03/19/91)
Dave,
Well, perhaps ntp v3 is the answer to our internet desolation.
Unfortunately, HP-UX (HP's incarnation of Unix) is rather a special
case, and I've been quite happy to let the experts at HP Labs
(aka Ken Stone) do the hard part.
I can't afford the time to port the new version. But perhaps I'll just
take a peek. Where can I find v3? I've found ntp.3.4/ntp-test.tar.Z
on louie.udel.edu, but somehow I'd gained the impression that that was
the "old" version of ntp. Am I wrong?
-----------------------------------------------
Bob Herlien
MBARI (Monterey Bay Aquarium Research Institute)
hebo@hp850.mbari.org
P.S. Stanford is busy again this morning. I just watched one of our
stratum 2's drop 6 in a row, and 7 out of 8, to an external stratum 1.
That generated a "reach" vector with one bit on, a new record.
Funny thing is, it had already wobbled previously, and the new wobble
brought them *closer* together. Random walk.ken@HPSDLZ.SDD.HP.COM (Ken Stone) (03/19/91)
> Well, perhaps ntp v3 is the answer to our internet desolation. > Unfortunately, HP-UX (HP's incarnation of Unix) is rather a special > case, and I've been quite happy to let the experts at HP Labs > (aka Ken Stone) do the hard part. Flattery will get you everywhere but I must admit that I'm part of the San Diego Division of HP, not HPLabs :-) > I can't afford the time to port the new version. But perhaps I'll just > take a peek. Where can I find v3? I've found ntp.3.4/ntp-test.tar.Z > on louie.udel.edu, but somehow I'd gained the impression that that was > the "old" version of ntp. Am I wrong? It was my impression that there was NOT yet a unix incarnation of the V3 spec (only the fuzzy code) ? If there is one I certainly would like to know about it. -- Ken
Mills@udel.edu (03/19/91)
Bob, The ntp.3.4 distribution is basically a hopped-up version 1 implementation, while the xntp distribution is version 2. I don't remember offhand who is working on it, but I have had several enquiries on details of the spec, some indicating keen scrutiny. I watch most of the stratum-1 critters from here and observe generally stable behavior with a couple of tens of milliseconds. Sometimes one or more of them drop to stratum 2 and sometimes they disappear for periods to maybe an hour, but in general the behavior is not nearly as bad as you report. But, then, I am not watching BARRNET hosts specifically. Here on SURANET we experience typically a 3-percent drop rate, but that does not materially affect NTP performance. Dave