Horton

Re: pg_waldump: support decoding of WAL inside tarfile

amul sul <sulamul@gmail.com>

View thread
From: Amul Sul <sulamul@gmail.com>
To: Jakub Wartak <jakub.wartak@enterprisedb.com>
Cc: Robert Haas <robertmhaas@gmail.com>, PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Date: 2025-11-21T11:44:26Z
Lists: pgsql-hackers

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

  1. Move tar detection and compression logic to common.

  2. pg_verifybackup: Enable WAL parsing for tar-format backups

  3. pg_waldump: Add support for reading WAL from tar archives

  4. pg_waldump: Preparatory refactoring for tar archive WAL decoding.

  5. pg_verifybackup: Verify tar-format backups.

On Wed, Nov 19, 2025 at 1:50 PM Jakub Wartak
<jakub.wartak@enterprisedb.com> wrote:
>
> On Mon, Nov 17, 2025 at 5:51 AM Amul Sul <sulamul@gmail.com> wrote:
> >
> > On Thu, Nov 6, 2025 at 2:33 PM Amul Sul <sulamul@gmail.com> wrote:
> > >
> > > On Mon, Oct 20, 2025 at 8:05 PM Robert Haas <robertmhaas@gmail.com> wrote:
> > > >
> > > > On Thu, Oct 16, 2025 at 7:49 AM Amul Sul <sulamul@gmail.com> wrote:
> > > > [....]
> > > Kindly have a look at the attached version. Thank you !
> > >
> >
> > Attached is the rebased version against the latest master head (e76defbcf09).
>
> Hi Amul, thanks for working on this. I haven't really looked at the
> source code deeply (I trust Robert eyes much more than mine on this
> one), just skimmed a little bit:
>
> 1. As stated earlier, get_tmp_walseg_path() is still vulnerable (it
> uses predictable path that could be used by attacker in $TMPDIR)
>

Yeah, I haven't done anything regarding this since I am unsure of what
should be done and what the risks involved are. I am thinking of
taking Robert's opinion on this.

> 2. On the usability front:
>
> a. If you do `pg_waldump --path pg_wal.tar -s 0/31000000` it will dump
> a lot of WAL records and then print final:
> pg_waldump: error: could not find file "000000010000000000000034" in archive
>
> However, with `pg_waldump --path pg_wal.tar -s 0/31000000
> --stats=record` (not passing '-e') it will simply bailout without
> printing stats and with error:
> pg_waldump: error: could not find file "000000010000000000000034" in archive
>
> IMHO, it could print stats if it was capable of getting at least 1 WAL record.
>

The similar behavior in the current pg_waldump when using the --path
option with a WAL directory and a starting LSN. E.g:

$ pg_waldump -s 0/04FE36E0 --path=/tmp/backup/tmp/ --stats=record

pg_waldump: first record is after 0/04FE36E0, at 0/04FE3F90, skipping
over 2224 bytes
pg_waldump: error: could not find file "000000010000000000000009": No
such file or directory


> 3. The most critical issue for me was the initial lack of error
> pass-through from pg_waldump (when used with WALs in tar) to the
> pg_verifybackup. Now it works fine, so thanks for this:
>

Thanks, that was exactly the intention -- to complete pg_verifybackup
for tar-formatted backup verification.

Regards,
Amul