[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: 'The' timestamp of a snapshot of deb.debian.org

On Sun, Sep 18, 2022 at 10:58:37AM +0200, Roland Clobus wrote:
> All of these timestamps (for sid) are close to each other, but not
> identical. I would guess that the earliest timestamp is the 'real'
> timestamp, but it is accessible (on snapshot.d.o) only with a later
> timestamp.

Consider that the archive "freezes" when it starts working on an update.
Once it finishes its job it generates the Release files, and puts the
current timestamp *inside* the file.  But then that file gets copied
around, signed, etc and that will change the timestamp.

Also, for example, I just spotted a line on IRC saying that this time
around copying the archive from the live one to archive.debian.org also
reset the mtimes...

So I really wouldn't trust mtimes in any way, only what's inside the
Release file.

What's in the trace file is IIRC the time when the archive kicks the
mirrors off, which also happens after the generation of the indexes, and
really has no business in this discussion as it's only a tool mirrors
use to coordinate and see if they are too much out of sync.
As you noticed snapshots indexes by time of the archive run, which is…
honestly I don't particularly consider it a good idea myself but that's
how the design decision went I guess.
(does it make sense to archive twice if there were two mirror pushes for
the same identical set of index files?)

                        Mattia Rizzolo

GPG Key: 66AE 2B4A FCCF 3F52 DA18  4D18 4B04 3FCD B944 4540      .''`.
More about me:  https://mapreri.org                             : :'  :
Launchpad user: https://launchpad.net/~mapreri                  `. `'`
Debian QA page: https://qa.debian.org/developer.php?login=mattia  `-

Attachment: signature.asc
Description: PGP signature

Reply to: