Re: Flush some statistics within running transactions

Hi,

On Thu, Feb 19, 2026 at 12:58:12PM +0900, Michael Paquier wrote:
> On Wed, Feb 18, 2026 at 05:40:46AM +0000, Bertrand Drouvot wrote:
> > PFA a mandatory rebase (nothing that needs review) due to a92b809f9da1.
>
> I don't find the design of this patch appealing, and my mind points
> towards two pieces of it:
> 1) The new requirement related to pgstat_schedule_anytime_update()
> that a stats kind needs to call to enable a timeout. This partially
> doubles with pgstat_report_fixed. And I suspect that this extra set
> of requirements, introducing a new level of complexity for in-core
> stats kinds as well as extension developers, would be the source of
> more bugs.

Yeah, maybe we should re-think the way we report that we have something to flush,
but I think that's more a general discussion that should also take care of
pgstat_report_fixed.

> 2) The timeout requirement itself, relying on a timeout threshold
> controlled by a backend-side configuration.

What are you concerns with this?

> With that in mind, wouldn't it be simpler if we introduced an API that
> could be used from client applications instead, in a model similar
> what we do for procsignal.c/h?

That's another angle to look at it but I think that giving this responsability to
the clients would not solve the concerns we had in [1] (that led to 039549d70f6
and to this thread). It seems to me that a solution/design that does not allow
us to "revert" 039549d70f6 does not suit our needs. Thoughts?

> Not a fan of the hardcoded sleeps in the tests, either.

Yeah, after our off-list discussion yesterday, I tried to implement the same
trick that f1e251be80a has done with injection points (nice trick by the way!),
but that led to:

TRAP: failed Assert("CritSectionCount == 0 || (context)->allowInCritSection"), File: "mcxt.c", Line: 1237
postgres: main: walwriter (ExceptionalCondition+0x9e)[0xc4ce4d]
postgres: main: walwriter (MemoryContextAlloc+0x8c)[0xc8f0ec]
postgres: main: walwriter (MemoryContextStrdup+0x37)[0xc8fea1]
postgres: main: walwriter (pstrdup+0x22)[0xc8fee4]
postgres: main: walwriter (substitute_path_macro+0x65)[0xc56068]
postgres: main: walwriter [0xc55e90]
postgres: main: walwriter (load_external_function+0x59)[0xc553db]
postgres: main: walwriter [0xc7b858]
postgres: main: walwriter [0xc7c125]
postgres: main: walwriter (IsInjectionPointAttached+0x18)[0xc7c20c]
postgres: main: walwriter (pgstat_count_backend_io_op+0x12f)[0xa9a116]
postgres: main: walwriter (pgstat_count_io_op+0x169)[0xa9cb57]
postgres: main: walwriter (pgstat_count_io_op_time+0x1cc)[0xa9cda7]

So, I did not spend that much time on it. I could if we strongly think that those
sleeps have to be discarded though.

[1]: https://postgr.es/m/erpzwxoptqhuptdrtehqydzjapvroumkhh7lc6poclbhe7jk7l%40l3yfsq5q4pw7

Regards,

--
Bertrand Drouvot
PostgreSQL Contributors Team
RDS Open Source Databases
Amazon Web Services: https://aws.amazon.com