Re: trying again to get incremental backup

On Tue, Oct 24, 2023 at 8:29 AM Robert Haas <robertmhaas(at)gmail(dot)com> wrote:
> Yeah, maybe so. I'm not quite ready to commit to doing that split as
> of this writing but I will think about it and possibly do it.

I have done this. Here's v7.

This version also includes several new TAP tests for the main patch,
some of which were inspired by our discussion. It also includes SGML
documentation for pg_walsummary.

New tests:
003_timeline.pl tests the case where the prior backup for an
incremental backup was taken on an earlier timeline.
004_manifest.pl tests the manifest-related options for pg_combinebackup.
005_integrity.pl tests the sanity checks that prevent combining a
backup with the wrong prior backup.

Overview of the new organization of the patch set:
0001 - preparatory refactoring of basebackup.c, changing the algorithm
that we use to decide which files have checksums
0002 - code movement only. makes it possible to reuse parse_manifest.c
0003 - add the WAL summarizer process, but useless on its own
0004 - add incremental backup, making use of 0003
0005 - add pg_walsummary debugging tool

Notes:
- I suspect that 0003 is the most likely to have serious bugs, followed by 0004.
- See XXX comments in the commit messages for some known open issues.
- Still looking for more comments on
http://postgr.es/m/CA+TgmoYdPS7a4eiqAFCZ8dr4r3-O0zq1LvTO5drwWr+7wHQaSQ@mail.gmail.com
and other recent emails where design questions came up

--
Robert Haas
EDB: http://www.enterprisedb.com