Horton

Re: Add a GUC check hook to ensure summarize_wal cannot be enabled when wal_level is minimal

Jelte Fennema-Nio <postgres@jeltef.nl>

View thread View on postgr.es
From: Jelte Fennema-Nio <postgres@jeltef.nl>
To: Nathan Bossart <nathandbossart@gmail.com>
Cc: Fujii Masao <masao.fujii@oss.nttdata.com>, pgsql-hackers@lists.postgresql.org
Date: 2024-07-03T14:45:10Z
Lists: pgsql-hackers 
On Wed, 3 Jul 2024 at 16:30, Nathan Bossart <nathandbossart@gmail.com> wrote:
> IME these sorts of GUC hooks that depend on the value of other GUCs tend to
> be quite fragile.  This one might be okay because wal_level defaults to
> 'replica' and because there is an additional check in postmaster.c, but
> that at least deserves a comment.

Yeah, this hook only works because wal_level isn't PGC_SIGHUP and
indeed because there's a check in postmaster.c. It now depends on the
ordering of these values in your config which place causes the error
message on startup.

This hits the already existing check:
summarize_wal = 'true'
wal_sumarizer = 'minimal'

This hits the new check:
summarize_wal = 'true'
wal_sumarizer = 'minimal'

And actually this would throw an error from the new check even though
the config is fine:

wal_sumarizer = 'minimal'
summarize_wal = 'true'
wal_sumarizer = 'logical'

> This sort of thing comes up enough that perhaps we should add a
> better-supported way to deal with GUCs that depend on each other...

+1. Sounds like we need a global GUC consistency check

Commits

  1. Do not summarize WAL if generated with wal_level=minimal.

  2. Revert "Auto-tune effective_cache size to be 4x shared buffers"

  3. Un-break ecpg test suite under --disable-integer-datetimes.

  4. Again fix initialization of auto-tuned effective_cache_size.

  5. Code review for auto-tuned effective_cache_size.

  6. Auto-tune effective_cache size to be 4x shared buffers