Message-ID: <8038b7ca-d4ba-40fb-824b-5f984fb80612@dalibo.com>
Date: Fri, 20 Dec 2024 10:57:40 +0100
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
Subject: Re: Why a bitmap scan in this case?
From: =?UTF-8?Q?Fr=C3=A9d=C3=A9ric_Yhuel?= <frederic.yhuel@dalibo.com>
To: Jon Zeppieri <zeppieri@gmail.com>,
 Greg Sabino Mullane <htamfids@gmail.com>
Cc: pgsql-performance@lists.postgresql.org
References: 
 <CAKfDxxzqxm0gS44+4aUDejLq_iszqLJwR+EaM4x2pojrkV58eg@mail.gmail.com>
 <CAKAnmmL327t5MRJ2uonQ3JB5bLegRbfa-WtpaTNBnbJ1iVef5Q@mail.gmail.com>
 <CAKfDxxw-=_naXS-ypU9qWj8eNbAEJvGCy5ZxHA5HW=+M-baDcw@mail.gmail.com>
 <43c7d623-1805-4db2-a8fc-9a99d3633320@dalibo.com>
Content-Language: fr, en-US
Autocrypt: addr=frederic.yhuel@dalibo.com; keydata=
 xjMEXn3bgxYJKwYBBAHaRw8BAQdA6tX5FT/n5ztMWIoBdl6k5avvu65fv6ryfVzIx/aH3V3N
 JEYuIFlodWVsIDxmcmVkZXJpYy55aHVlbEBkYWxpYm8uY29tPsKWBBMWCAA+FiEEi7OTyf6H
 ePS+C6n11XGWhN8zGNUFAmZDVJ8CGwMFCQtfi90FCwkIBwIGFQoJCAsCBBYCAwECHgECF4AA
 CgkQ1XGWhN8zGNV65gEA6o+UrebFjn2CIAOYwP3CiiJugiM5TMH9yJyehQua5ZoA/2ChxiVn
 8OUaAw/ErxJmQi1mSN0tNZBL/KhkUPBDUDQNzjgEXn3bgxIKKwYBBAGXVQEFAQEHQCuh5cLQ
 AtEBXHa2Fmtp0kduBu7msM7qO/gaEcdjFUdiAwEIB8J+BBgWCAAmAhsMFiEEi7OTyf6HePS+
 C6n11XGWhN8zGNUFAmZDVJ8FCQtfi90ACgkQ1XGWhN8zGNWOOAEArSqePBqfd4Kx4ulACaWO
 6fM+XSDfUlBAQXx9rU6DR4MBALwozl9g91tRysim6lKxggvBfp/PvbpTZNrxYLWB2fYP
In-Reply-To: <43c7d623-1805-4db2-a8fc-9a99d3633320@dalibo.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 8bit
Archived-At: 
 <https://www.postgresql.org/message-id/8038b7ca-d4ba-40fb-824b-5f984fb80612%40dalibo.com>
Precedence: bulk


On 12/20/24 09:16, Frédéric Yhuel wrote:
> 
> 
> On 12/19/24 20:09, Jon Zeppieri wrote:
>> The table is freshly vacuumed. If I disable bitmap scans, it will do
>> an index only scan, which performs better. For the bitmap heap scan,
>> it says "Heap Blocks: exact=27393," whereas for the index only scan,
>> it's "Heap Fetches: 27701."
> 
> So you have 100% heap fetches. Are you sure that your table is freshly 
> vacuumed? Please note that VACUUM FULL doesn't create the visibility 
> map, so you still have to run a plain VACUUM for this.
> 
> 

And if you still have heap fetches, you could try VACUUM (INDEX_CLEANUP 
TRUE), so that it doesn't bypass index vacuuming.

(In case the pages that contain theses 27701 tuples amount to less than 
2% of the total number of pages of the table).

Nice explanation here: 
https://www.cybertec-postgresql.com/en/making-the-postgresql-visibility-map-visible/#vacuum-doesnt-update-the-visibility-map-as-it-should