Re: FreeBSD:1[56]:amd64/base_latest/ looks to again not be updating when others are

On 2025-11-04 02:14:43 (+0800), Mark Millard wrote:

> On Nov 3, 2025, at 08:23, Baptiste Daroussin <bapt@FreeBSD.org> wrote:
>
>> On Mon 03 Nov 08:15, Mark Millard wrote:
>>> It is after 2025-Nov-02 16:11:00 as I write this.
>>>
>>> https://pkg.freebsd.org/FreeBSD:15:amd64/base_latest/?C=M&O=D shows 2025-Nov-02 01:50
>>> https://pkg.freebsd.org/FreeBSD:16:amd64/base_latest/?C=M&O=D shows 2025-Nov-02 00:51
>>>
>>> By contrast FreeBSD:14:amd64 and FreeBSD:1[56]:aarch64 have been updating:
>>>
>>> https://pkg.freebsd.org/FreeBSD:14:amd64/base_latest/?C=M&O=D   shows 2025-Nov-03 10:31
>>>
>>> https://pkg.freebsd.org/FreeBSD:15:aarch64/base_latest/?C=M&O=D shows 2025-Nov-03 08:55
>>> https://pkg.freebsd.org/FreeBSD:16:aarch64/base_latest/?C=M&O=D shows 2025-Nov-03 06:30
>>
>> As usual you can follow the publication here: https://people.freebsd.org/~dbaio/pkg-master-report.html
>
> Thanks much for the reference to that. Good to know.
> And far better than how I had been looking at the status.
>
>> you can see they have been built properly and sent for publication.
>
> Mirror kul indicates missing consistently for what
> I reported (and more for FreeBSD:1[56]:x86:64), the
> only one that does so. So I guess that is where my
> attempts are going.
>
> I wonder if there is a way to avoid a mirror that is
> not updating.

Package sets only become (atomically) visible when they are on all mirrors.  That goes for third party packages as well as pkgbase.

This makes pkg immune to dependency hells where different subsets of packages exist on different mirrors and geoDNS points you at two different ones (for whatever reason).

(The cost is that packages publish at the rate of the slowest mirror -- bra, jinx, kul and syd can be slow to catch up depending on internet weather.)

Philip