Re: race condition when writing pg_control - Mailing list pgsql-hackers

From Thomas Munro
Subject Re: race condition when writing pg_control
Date
Msg-id CA+hUKGJ2Mfp1s+nKk1rxQytB0p2OVxUUiYTpi4S3a7UX862K5Q@mail.gmail.com
Whole thread Raw
In response to Re: race condition when writing pg_control  (Noah Misch <noah@leadboat.com>)
Responses Re: race condition when writing pg_control
Re: race condition when writing pg_control
List pgsql-hackers
On Fri, Jul 12, 2024 at 11:43 PM Noah Misch <noah@leadboat.com> wrote:
> On Sat, May 18, 2024 at 05:29:12PM +1200, Thomas Munro wrote:
> > On Fri, May 17, 2024 at 4:46 PM Thomas Munro <thomas.munro@gmail.com> wrote:
> > > The specific problem here is that LocalProcessControlFile() runs in
> > > every launched child for EXEC_BACKEND builds.  Windows uses
> > > EXEC_BACKEND, and Windows' NTFS file system is one of the two file
> > > systems known to this list to have the concurrent read/write data
> > > mashing problem (the other being ext4).
>
> > First idea idea I've come up with to avoid all of that: pass a copy of
> > the "proto-controlfile", to coin a term for the one read early in
> > postmaster startup by LocalProcessControlFile().  As far as I know,
> > the only reason we need it is to suck some settings out of it that
> > don't change while a cluster is running (mostly can't change after
> > initdb, and checksums can only be {en,dis}abled while down).  Right?
> > Children can just "import" that sucker instead of calling
> > LocalProcessControlFile() to figure out the size of WAL segments yada
> > yada, I think?  Later they will attach to the real one in shared
> > memory for all future purposes, once normal interlocking is allowed.
>
> I like that strategy, particularly because it recreates what !EXEC_BACKEND
> backends inherit from the postmaster.  It might prevent future bugs that would
> have been specific to EXEC_BACKEND.

Thanks for looking!  Yeah, that is a good way to put it.

The only other idea I can think of is that the Postmaster could take
all of the things that LocalProcessControlFile() wants to extract from
the file, and transfer them via that struct used for EXEC_BACKEND as
individual variables, instead of this new proto-controlfile copy.  I
think it would be a bigger change with no obvious-to-me additional
benefit, so I didn't try it.

> > I dunno.  Draft patch attached.  Better plans welcome.  This passes CI
> > on Linux systems afflicted by EXEC_BACKEND, and Windows.  Thoughts?
>
> Looks reasonable.  I didn't check over every detail, given the draft status.

I'm going to upgrade this to a proposal:

https://commitfest.postgresql.org/49/5124/

I wonder how often this happens in the wild.



pgsql-hackers by date:

Previous
From: Thomas Munro
Date:
Subject: Re: Confine vacuum skip logic to lazy_scan_skip
Next
From: Fujii Masao
Date:
Subject: Re: MERGE/SPLIT partition commands should create new partitions in the parent's tablespace?