From: Tom Lane <tgl@sss.pgh.pa.us>
To: "=?gb18030?B?wb249rqi19O1xLX5?=" <1726002692@qq.com>
cc: "=?gb18030?B?cGdzcWwtYnVncw==?=" <pgsql-bugs@lists.postgresql.org>
Subject: Re: BUG #17382: When vacuum full or vacuumdb - F is executed,
 a large number of empty files will be generated in the
In-reply-to: <tencent_455C88A54F4AFB73D3824A2558585F911209@qq.com>
References: <17382-9b238ae4286aef7d@postgresql.org>
 <2441394.1643210985@sss.pgh.pa.us>
 <tencent_455C88A54F4AFB73D3824A2558585F911209@qq.com>
Comments: In-reply-to "=?gb18030?B?wb249rqi19O1xLX5?=" <1726002692@qq.com>
	message dated "Fri, 28 Jan 2022 21:42:47 +0800"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <2772240.1643386532.1@sss.pgh.pa.us>
Content-Transfer-Encoding: quoted-printable
Date: Fri, 28 Jan 2022 11:15:32 -0500
Message-ID: <2772241.1643386532@sss.pgh.pa.us>
Archived-At: 
 <https://www.postgresql.org/message-id/2772241.1643386532%40sss.pgh.pa.us>
Precedence: bulk

"=3D?gb18030?B?wb249rqi19O1xLX5?=3D" <1726002692@qq.com> writes:
> Sorry, it's my fault. The empty files generated by "vacuumdb - f " is lo=
cated under the $pgdata / base / 99522 directory

Oh, if they are under the database's subdirectory then it's not so
surprising.  VACUUM FULL (or, actually, any table deletion or rewrite)
leaves behind an empty "tombstone" file that is just there to prevent
re-use of that table number for the moment.  It'll go away at the next
checkpoint.

			regards, tom lane