Date: Wed, 8 Apr 2026 16:48:31 -0500
From: Nathan Bossart <nathandbossart@gmail.com>
To: Sami Imseih <samimseih@gmail.com>
Cc: Andres Freund <andres@anarazel.de>, Tom Lane <tgl@sss.pgh.pa.us>,
	Alexander Lakhin <exclusion@gmail.com>,
	Bharath Rupireddy <bharath.rupireddyforpostgres@gmail.com>,
	Robert Treat <rob@xzilla.net>, satyanarlapuram@gmail.com,
	pgsql-hackers <pgsql-hackers@postgresql.org>, tndrwang@gmail.com
Subject: Re: Add pg_stat_autovacuum_priority
Message-ID: <adbNL1puSbBi8nMo@nathan>
References: <adQsdvPPNviWMCXb@nathan>
 <5684f479-858e-4c5d-b8f5-bcf05de1f909@gmail.com>
 <3077290.1775668665@sss.pgh.pa.us>
 <adaWuTR7oCKodH7k@nathan>
 <3097984.1775674400@sss.pgh.pa.us>
 <adan_E0o69p81lIj@nathan>
 <3101163.1775676098@sss.pgh.pa.us>
 <u3akkfw3vaocviudt6f7ft4kxjc2273rh3tfhz5rfwg6jrcc2t@wurwws6wgjjz>
 <adbHYQMZ0pHQttWa@nathan>
 <CAA5RZ0uL1ukKvV0Gwgok9ut_203q_E_tzBrwhDAuJCKVNwWZVw@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: 
 <CAA5RZ0uL1ukKvV0Gwgok9ut_203q_E_tzBrwhDAuJCKVNwWZVw@mail.gmail.com>
Archived-At: <https://www.postgresql.org/message-id/adbNL1puSbBi8nMo%40nathan>
Precedence: bulk

On Wed, Apr 08, 2026 at 04:33:00PM -0500, Sami Imseih wrote:
>> On Wed, Apr 08, 2026 at 04:40:03PM -0400, Andres Freund wrote:
>> > Note that the whole cached state does automatically get reset at the end of
>> > the transaction (AtEOXact_PgStat()->pgstat_clear_snapshot()), just like it did
>> > before the shmem stats stuff.
>>
>> I see a lot of memory used for the pgStatEntryRefHash table, too (e.g., ~16
>> MB for 100K tables).  What's interesting is that I cannot reproduce similar
>> usage with views like pg_stat_all_tables.  If memory was not a concern, I
>> think the "bool *may_free" idea would be fine.
> 
> Instead of may_free, which is invasive, what about pgstat_fetch_entry_nocache
> which can be called by 2 new APIs pgstat_fetch_stat_tabentry_nocache() and
> pgstat_fetch_stat_tabentry_nocache_ext(). This way a caller that uses
> these will be required to pfree?

This might help avoid memory usage within a snapshot, but as Andres notes,
this gets released automatically at the end of the transaction.

For a database with 100K tables, here's what I see before calling the view:

          name          | used_bytes
------------------------+------------
 PgStat Shared Ref Hash |       8576
 PgStat Shared Ref      |       2960
 PgStat Pending         |       4712
(3 rows)

After calling the view but before committing the transaction, I see this:

          name          | used_bytes
------------------------+------------
 PgStat Snapshot        |    4194688
 PgStat Shared Ref Hash |    4194688
 PgStat Shared Ref      |    8048240
 PgStat Pending         |       3064
(4 rows)

And after committing, I see:

          name          | used_bytes
------------------------+------------
 PgStat Shared Ref Hash |    4194688
 PgStat Shared Ref      |    8048240
 PgStat Pending         |        360
(3 rows)

-- 
nathan