I am using PostgreSQL 9.6. I have a query Like this:
SELECT anon_1.id AS anon_1_id, anon_1.is_valid AS anon_1_is_valid, anon_1.first_name AS anon_1_first_name, anon_1.last_name AS anon_1_last_name,
anon_1.patronymic_name AS anon_1_patronymic_name,
anon_1.experience AS anon_1_experience, anon_1.user_id AS anon_1_user_id, anon_1.rating_points as rating_points
FROM (SELECT DISTINCT ON (doctors.id) doctors.id AS id, doctors.created_at AS created_at, doctors.updated_at AS updated_at, doctors.is_valid AS is_valid, doctors.pretty_url AS pretty_url, doctors.first_name AS first_name, doctors.last_name AS last_name, doctors.patronymic_name AS patronymic_name, doctors.phone AS phone, doctors.birthday AS birthday, doctors.avatar AS avatar, doctors.experience AS experience, doctors.science_degree AS science_degree, doctors.sex_id AS sex_id, doctors.yclients_staff_id AS yclients_staff_id, doctors.user_id AS user_id, doctor_has_specialties.rating_points AS rating_points, clinic_branch_has_doctors.price AS price, clinic_branch_has_doctors.doctor_type AS doctor_type, clinic_branch_has_doctors.is_house_call AS is_house_call, clinic_branch_has_doctors.house_call_price AS house_call_price
FROM doctors
JOIN doctor_has_specialties ON doctors.id = doctor_has_specialties.doctor_id
JOIN clinic_branch_has_doctors ON doctor_has_specialties.id = clinic_branch_has_doctors.doctor_has_specialty_id
JOIN clinic_branches ON clinic_branches.id = clinic_branch_has_doctors.clinic_branch_id
JOIN city_areas ON city_areas.id = clinic_branches.city_area_id
JOIN cities ON cities.id = city_areas.city_id
WHERE doctors.is_valid = true
AND clinic_branch_has_doctors.is_valid = true
AND clinic_branches.is_valid = true
AND doctor_has_specialties.specialty_id = 1
AND cities.id = 1) AS anon_1 ORDER BY anon_1.rating_points DESC
LIMIT 100 OFFSET 0
Here the most important part of the query is the last one, LIMIT and OFFSET. When I run this query I get all my data up to 100 rows. Everything is fine not. Here is the screenshot from pgAdmin:
Here notice row with id 20. Now If I try to OFFSET 10, I get my data from 11th row as expected which is object with id 22. Everything is fine too. But If I try OFFSET 10 LIMIT 10, I get strange result. Mainly row with id 20 appears too, but it shouldn't. Here is the screenshot:
Have no idea what is wrong with that. Is it Postgres bug? Or I am doing something wrong.
There is no "bug" at work here. You specified the following ordering, which was used when LIMIT
and OFFSET
were being applied:
ORDER BY anon_1.rating_points DESC
However, in the case two or more records are tied with the same rating_points
values, Postgres makes no guarantee about what the ordering will be. This is why you are seeing a user with anon_id_1
of 20
apparently moving around. Postgres has done nothing wrong; it has honored your request to order by the rating_points
, but you never told it what to do in the case of tie.
To resolve this, you could add a second condition to the ORDER BY
:
ORDER BY
anon_1.rating_points DESC,
anon_id_1
This would break the tie with regard to ordering, and, assuming anon_id_1
is a primary key, the results would appear to be stable after making this change.
If you love us? You can donate to us via Paypal or buy me a coffee so we can maintain and grow! Thank you!
Donate Us With