Horton
Menu

Re: BUG #18146: Rows reappearing in Tables after Auto-Vacuum Failure in PostgreSQL on Windows

Thomas Munro <thomas.munro@gmail.com>

View thread
From: Thomas Munro <thomas.munro@gmail.com>
To: Michael Paquier <michael@paquier.xyz>
Cc: Robert Haas <robertmhaas@gmail.com>, Heikki Linnakangas <hlinnaka@iki.fi>, Alexander Lakhin <exclusion@gmail.com>, Tom Lane <tgl@sss.pgh.pa.us>, Laurenz Albe <laurenz.albe@cybertec.at>, rootcause000@gmail.com, pgsql-bugs@lists.postgresql.org
Date: 2024-12-20T22:18:23Z
Lists: pgsql-bugs 
On Fri, Dec 20, 2024 at 3:53 PM Michael Paquier <michael@paquier.xyz> wrote:
> On Thu, Dec 19, 2024 at 10:44:05AM -0500, Robert Haas wrote:
> > I think mostly my thought was that if we could remove smgrtruncate()
> > entirely in favor of smgrtruncatefrom(), or just keep it called
> > smgrtruncate() but add a mandatory additional argument, that would be
> > less error-prone than having two versions between which future hackers
> > must pick.

Yeah.

> Hmm.  Indeed.  As a HEAD change, keeping only a smgrtruncate() is
> tempting as it creates a parallel with md.c.  I am not completely sure
> how to make all that leaner with the smgrnblocks() calls that save the
> old number of blocks for each fork.  But perhaps Thomas has a fancy
> idea if it comes down to that, and it could always be done later.

I did that, but also back-patched it like that.  I realise now that
that was an ABI mistake, and I plan to change it back to the v5
arrangement (smgrtruncate() unchanged, smgrtruncatefrom() with the new
argument) in the back-branches only, just in case someone is using raw
smgrtruncate() in compiled code in the wild.  Will post a patch soon.

Commits

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

  1. Fix C error reported by Oracle compiler.

  2. Restore smgrtruncate() prototype in back-branches.

  3. Fix corruption when relation truncation fails.

  4. RelationTruncate() must set DELAY_CHKPT_START.

  5. WAL-log inplace update before revealing it to other sessions.

  6. Fix bugs in MultiXact truncation

  7. Fix possible recovery trouble if TRUNCATE overlaps a checkpoint.