Re: git: b1c95af45488 - main - rc.conf: correct $ntp_leapfile_sources
Date: Thu, 07 Dec 2023 23:56:37 UTC
On Thu, Dec 7, 2023 at 3:27 PM Steffen Nurpmeso <steffen@sdaoden.eu> wrote: > Xin Li wrote in > <d75b041f-05f8-44c1-8de6-1fef89b7e537@delphij.net>: > |On 2023-12-06 22:34, Philip Paeps wrote: > |> On 2023-12-07 14:26:05 (+0800), Warner Losh wrote: > |>> We should point to bipm > |>> https://hpiers.obspm.fr/iers/bul/bulc/ntp/leap-seconds.list since > they > |>> are > |>> the source of truth, no? > |> > |> I went for the IANA copy because data.iana.org is a much shorter and > |> trustworthy looking URL. And it's also where other operating systems > |> get their copies. > | > |My understanding is that IANA's copy is part of tzdata and it's only > |updated when a new set of zone data is released, so it's sometimes > |outdated. It is actually going to be outdated really soon by the way. > > But nothing will change. > It is only about the included end-of-life tag why there is > discussion at all. > The IANA TZ data is always updated as necessary, "early enough". > Yes. TZ data updates multiple times a year. The lead time on NIST/BIPM updating the file usually is within days or weeks after the new leap is announced. But ntpd can't possibly use it for about 5 months. TZ updates are plenty fast. The bigger problem is that we have to do a EN to get a new set of zone files. If we had a way to fetch them, we could just copy this file from the updated zone files. > Also the beasts are about to get rid of leap seconds until 2035 > (and they have beaten onto the Russians which' GLONASS is capable > to deal properly, as far as the discussion was, the fact it is not > earlier). Bets can be placed whether it will happen before > a possible occurring leap second, or not. (My bet is that they > run everything against the wall, and then run away yelling about > the evil leap second, after having missed to create an appropriate > environment to deal with the reached scientific level to keep us > truly within one second of our home planet, and his star. > O tempora, o mores. But that is off-topic.) > Yea, we're years away from the next leap second. And there's still a good chance we'll have at least one more. And there's also rumblings that leaps will stop before 2035 (that's the current absolute last date, and there's several folks that want to pull that in). What will happen for sure isn't well known... ntpv5 discussions have, at times, assumed there will be no more leap seconds and so ntpv5 needn't have anything to accommodate them. > |The IERS one is more up-to-date because they publish the bulletin. > > In general i think distribution of load is a good thing, and > i find it very unfriendly to put all the load onto some jealous > institute (if it is one) and its single server. > The FreeBSD project has an established set of mirrors, and, the > way i see the excessive use of installations on clowds and such, > for example for github actions which spawn dozens of OS > installations to test a commit (doh!). > We can skew the load in time by spreading all our users out over the few months we have if there's a load issue. > Btw PHK had a thrilling idea of DNS distributing leap ticks some > years ago, and he even started to host it. As it unfortunately > did not fly i did not track it further. > Would also be an idea for the FreeBSD project: simply download the > file ones, then place a DNS record that FreeBSD installations then > can query. DNSSEC is in place i think. > Yea, that's not a thing that's happening. It was an interesting idea, but hasn't been standardized and there's little to apetite to distribute this way. > |The bundled version was from NIST ftp, but fetching from ftp for every > |FreeBSD system out there was too scary for me. > | > |There may be some security / privacy concerns if we direct users to a > |place that we do not have control, by the way. > > Interesting aspect! > There might be, but this sounds somewhat speculative. What's the anticipated concerns? Warner