Re: Unexpected behavior after OOM errors

On Tue, Jul 21, 2026 at 02:24:13PM +0200, Matthias van de Meent wrote:
> I think I found the source of these Asserts; the attached fixes it.
> The problem is that we're registering the type entry handling as 'in
> progress' before the cache is initialized, and the error handling hook
> for in-progress type entries initializes the TYPEOID cache if there's
> an in-progress entry and the cache wasn't already initialized. And,
> initializing this catcache in non-transactional contexts isn't
> allowed, so the error handling causes the assertion to fail.

I was wondering if this update should happen further down, but that
led to an inconsistent handling of the in-progress list within the
post-error callback. Looking at that a second time, just moving it
down the initial lookup seems fine enough. Applied after an edit of
the comment section.

> Next, I'll be looking at the last issue.

The pgstat one? That would help a lot, thanks! If you are not
familiar with the internals of this area in the backend, this problem
should give a pretty good introduction.
--
Michael