From: Tom Lane <tgl@sss.pgh.pa.us>
To: chjischj@163.com
cc: pgsql-bugs@lists.postgresql.org
Subject: Re: BUG #17548: Aggregate queries on partitioned tables can cause
 OOM.
In-reply-to: <17548-4d206a94ee58f3a5@postgresql.org>
References: <17548-4d206a94ee58f3a5@postgresql.org>
Comments: In-reply-to PG Bug reporting form <noreply@postgresql.org>
	message dated "Tue, 12 Jul 2022 16:25:58 -0000"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <1942039.1657664154.1@sss.pgh.pa.us>
Date: Tue, 12 Jul 2022 18:15:54 -0400
Message-ID: <1942040.1657664154@sss.pgh.pa.us>
Archived-At: 
 <https://www.postgresql.org/message-id/1942040.1657664154%40sss.pgh.pa.us>
Precedence: bulk

PG Bug reporting form <noreply@postgresql.org> writes:
> Aggregate queries on partitioned tables can cause out of memory error.

This is resolved in PG v13 and later.  In older versions, it's
advisable to do an explicit ANALYZE on a large table, so that the
planner knows that the number of groups is large and will avoid
trying to do a hash aggregation.

(I don't think it's terribly specific to partitioned tables,
for what that's worth, except that stats for a partitioned
table might be more likely to be out-of-date.)

			regards, tom lane