Re: snapshot too old issues, first around wraparound and then more.

On Wed, Jun 16, 2021 at 7:17 AM Robert Haas <robertmhaas(at)gmail(dot)com> wrote:
> Progress has been pretty limited, but not altogether nonexistent.
> 55b7e2f4d78d8aa7b4a5eae9a0a810601d03c563 fixed, or at least seemed to
> fix, the time->XID mapping, which is one of the main things that
> Andres said was broken originally. Also, there are patches on this
> thread from Thomas Munro to add some test coverage for that case,
> another problem Andres noted in his original email. I guess it
> wouldn't be too hard to get something committed there, and I'm willing
> to do it if Thomas doesn't want to and if there's any prospect of
> salvaging the feature.

FTR the latest patches are on a different thread[1]. I lost steam on
that stuff because I couldn't find a systematic way to deal with the
lack of checks all over the place, or really understand how the whole
system fits together with confidence. Those patches to fix an xid
wraparound bug and make the testing work better may be useful and I'll
be happy to rebase them, depending on how this discussion goes, but it
seems a bit like the proverbial deckchairs on the Titanic from what
I'm reading... I think the technique I showed for exercising some
basic STO mechanisms and scenarios is probably useful, but I currently
have no idea how to prove much of anything about the whole system and
am not personally in a position to dive into that rabbit hole in a
PG15 time scale.

[1] https://www.postgresql.org/message-id/CA%2BhUKGJyw%3DuJ4eL1x%3D%2BvKm16fLaxNPvKUYtnChnRkSKi024u_A%40mail.gmail.com