Re: COPY FREEZE and setting PD_ALL_VISIBLE/visibility map bits

From: Tatsuo Ishii <ishii(at)sraoss(dot)co(dot)jp>
To: lubennikovaav(at)gmail(dot)com
Cc: pgsql-hackers(at)lists(dot)postgresql(dot)org, pavan(dot)deolasee(at)gmail(dot)com, ibrar(dot)ahmad(at)gmail(dot)com
Subject: Re: COPY FREEZE and setting PD_ALL_VISIBLE/visibility map bits
Date: 2020-10-28 05:46:53
Message-ID: 20201028.144653.966121790856478893.t-ishii@sraoss.co.jp
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

> Status update for a commitfest entry.
>
> This patch is ReadyForCommitter. It applies and passes the CI. There are no unanswered questions in the discussion.
>
> The discussion started in 2015 with a patch by Jeff Janes. Later it was revived by Pavan Deolasee. After it was picked up by Ibrar Ahmed and finally, it was rewritten by me, so I moved myself from reviewers to authors as well.
>
> The latest version was reviewed and tested by Ibrar Ahmed. The patch doesn't affect COPY FREEZE performance and significantly decreases the time of the following VACUUM.

I have tested the patch on my laptop (mem 16GB, SSD 512GB) using the
data introduced in up thread and saw that VACCUM after COPY FREEZE is
nearly 60 times faster than current master branch. Quite impressive.

By the way, I noticed following comment:
+ /*
+ * vmbuffer should be already pinned by RelationGetBufferForTuple,
+ * Though, it's fine if is not. all_frozen is just an optimization.
+ */

could be enhanced like below. What do you think?
+ /*
+ * vmbuffer should be already pinned by RelationGetBufferForTuple.
+ * Though, it's fine if it is not. all_frozen is just an optimization.
+ */

Best regards,
--
Tatsuo Ishii
SRA OSS, Inc. Japan
English: http://www.sraoss.co.jp/index_en.php
Japanese:http://www.sraoss.co.jp

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Tatsuro Yamada 2020-10-28 06:07:56 Re: list of extended statistics on psql
Previous Message Tatsuro Yamada 2020-10-28 05:41:40 Re: [spam] Re: list of extended statistics on psql