gitlab-org--gitlab-foss/doc/development/database/keyset_pagination.md

10 KiB

stage group info
Enablement Database To determine the technical writer assigned to the Stage/Group associated with this page, see https://about.gitlab.com/handbook/engineering/ux/technical-writing/#assignments

Keyset pagination

The keyset pagination library can be used in HAML-based views and the REST API within the GitLab project.

You can read about keyset pagination and how it compares to the offset based pagination on our pagination guidelines page.

API overview

Synopsis

Keyset pagination with ActiveRecord in Rails controllers:

cursor = params[:cursor] # this is nil when the first page is requested
paginator = Project.order(:created_at).keyset_paginate(cursor: cursor, per_page: 20)

paginator.each do |project|
  puts project.name # prints maximum 20 projects
end

Usage

This library adds a single method to ActiveRecord relations: #keyset_paginate.

This is similar in spirit (but not in implementation) to Kaminari's paginate method.

Keyset pagination works without any configuration for simple ActiveRecord queries:

  • Order by one column.
  • Order by two columns, where the last column is the primary key.

The library detects nullable and non-distinct columns and based on these, adds extra ordering using the primary key. This is necessary because keyset pagination expects distinct order by values:

Project.order(:created_at).keyset_paginate.records # ORDER BY created_at, id

Project.order(:name).keyset_paginate.records # ORDER BY name, id

Project.order(:created_at, id: :desc).keyset_paginate.records # ORDER BY created_at, id

Project.order(created_at: :asc, id: :desc).keyset_paginate.records # ORDER BY created_at, id  DESC

The keyset_paginate method returns a special paginator object which contains the loaded records and additional information for requesting various pages.

The method accepts the following keyword arguments:

  • cursor - Encoded order by column values for requesting the next page (can be nil).
  • per_page - Number of records to load per page (default 20).
  • keyset_order_options - Extra options for building the keyset paginated database query, see an example for UNION queries in the performance section (optional).

The paginator object has the following methods:

  • records - Returns the records for the current page.
  • has_next_page? - Tells whether there is a next page.
  • has_previous_page? - Tells whether there is a previous page.
  • cursor_for_next_page - Encoded values as String for requesting the next page (can be nil).
  • cursor_for_previous_page - Encoded values as String for requesting the previous page (can be nil).
  • cursor_for_first_page - Encoded values as String for requesting the first page.
  • cursor_for_last_page - Encoded values as String for requesting the last page.
  • The paginator objects includes the Enumerable module and delegates the enumerable functionality to the records method/array.

Example for getting the first and the second page:

paginator = Project.order(:name).keyset_paginate

paginator.to_a # same as .records

cursor = paginator.cursor_for_next_page # encoded column attributes for the next page

paginator = Project.order(:name).keyset_paginate(cursor: cursor).records # loading the next page

Because keyset pagination does not support page numbers, we are restricted to go to the following pages:

  • Next page
  • Previous page
  • Last page
  • First page

Usage in Rails with HAML views

Consider the following controller action, where we list the projects ordered by name:

def index
  @projects = Project.order(:name).keyset_paginate(cursor: params[:cursor])
end

In the HAML file, we can render the records:

- if @projects.any?
  - @projects.each do |project|
    .project-container
      = project.name

  = keyset_paginate @projects

Performance

The performance of the keyset pagination depends on the database index configuration and the number of columns we use in the ORDER BY clause.

In case we order by the primary key (id), then the generated queries are efficient because the primary key is covered by a database index.

When two or more columns are used in the ORDER BY clause, it's advised to check the generated database query and make sure that the correct index configuration is used. More information can be found on the pagination guideline page.

NOTE: While the query performance of the first page might look good, the second page (where the cursor attributes are used in the query) might yield poor performance. It's advised to always verify the performance of both queries: first page and second page.

Example database query with tie-breaker (id) column:

SELECT "issues".*
FROM "issues"
WHERE (("issues"."id" > 99
      AND "issues"."created_at" = '2021-02-16 11:26:17.408466')
    OR ("issues"."created_at" > '2021-02-16 11:26:17.408466')
    OR ("issues"."created_at" IS NULL))
ORDER BY "issues"."created_at" DESC NULLS LAST, "issues"."id" DESC
LIMIT 20

OR queries are difficult to optimize in PostgreSQL, we generally advise using UNION queries instead. The keyset pagination library can generate efficient UNION when multiple columns are present in the ORDER BY clause. This is triggered when we specify the use_union_optimization: true option in the options passed to Relation#keyset_paginate.

Example:

# Triggers a simple query for the first page.
paginator1 = Project.order(:created_at, id: :desc).keyset_paginate(per_page: 2, keyset_order_options: { use_union_optimization: true })

cursor = paginator1.cursor_for_next_page

# Triggers UNION query for the second page
paginator2 = Project.order(:created_at, id: :desc).keyset_paginate(per_page: 2, cursor: cursor, keyset_order_options: { use_union_optimization: true })

puts paginator2.records.to_a # UNION query

Complex order configuration

Common ORDER BY configurations are handled by the keyset_paginate method automatically so no manual configuration is needed. There are a few edge cases where order object configuration is necessary:

  • NULLS LAST ordering.
  • Function-based ordering.
  • Ordering with a custom tie-breaker column, like iid.

These order objects can be defined in the model classes as normal ActiveRecord scopes, there is no special behavior that prevents using these scopes elsewhere (kaminari, background jobs).

NULLS LAST ordering

Consider the following scope:

scope = Issue.where(project_id: 10).order(Gitlab::Database.nulls_last_order('relative_position', 'DESC'))
# SELECT "issues".* FROM "issues" WHERE "issues"."project_id" = 10 ORDER BY relative_position DESC NULLS LAST

scope.keyset_paginate # raises: Gitlab::Pagination::Keyset::UnsupportedScopeOrder: The order on the scope does not support keyset pagination

The keyset_paginate method raises an error because the order value on the query is a custom SQL string and not an Arel AST node. The keyset library cannot automatically infer configuration values from these kinds of queries.

To make keyset pagination work, we must configure custom order objects, to do so, we must collect information about the order columns:

  • relative_position can have duplicated values because no unique index is present.
  • relative_position can have null values because we don't have a not null constraint on the column. For this, we must determine where we see NULL values, at the beginning of the result set, or the end (NULLS LAST).
  • Keyset pagination requires distinct order columns, so we must add the primary key (id) to make the order distinct.
  • Jumping to the last page and paginating backwards actually reverses the ORDER BY clause. For this, we must provide the reversed ORDER BY clause.

Example:

order = Gitlab::Pagination::Keyset::Order.build([
  # The attributes are documented in the `lib/gitlab/pagination/keyset/column_order_definition.rb` file
  Gitlab::Pagination::Keyset::ColumnOrderDefinition.new(
    attribute_name: 'relative_position',
    column_expression: Issue.arel_table[:relative_position],
    order_expression: Gitlab::Database.nulls_last_order('relative_position', 'DESC'),
    reversed_order_expression: Gitlab::Database.nulls_first_order('relative_position', 'ASC'),
    nullable: :nulls_last,
    order_direction: :desc,
    distinct: false
  ),
  Gitlab::Pagination::Keyset::ColumnOrderDefinition.new(
    attribute_name: 'id',
    order_expression: Issue.arel_table[:id].asc,
    nullable: :not_nullable,
    distinct: true
  )
])

scope = Issue.where(project_id: 10).order(order) # or reorder()

scope.keyset_paginate.records # works

Function-based ordering

In the following example, we multiply the id by 10 and order by that value. Because the id column is unique, we define only one column:

order = Gitlab::Pagination::Keyset::Order.build([
  Gitlab::Pagination::Keyset::ColumnOrderDefinition.new(
    attribute_name: 'id_times_ten',
    order_expression: Arel.sql('id * 10').asc,
    nullable: :not_nullable,
    order_direction: :asc,
    distinct: true,
    add_to_projections: true
  )
])

paginator = Issue.where(project_id: 10).order(order).keyset_paginate(per_page: 5)
puts paginator.records.map(&:id_times_ten)

cursor = paginator.cursor_for_next_page

paginator = Issue.where(project_id: 10).order(order).keyset_paginate(cursor: cursor, per_page: 5)
puts paginator.records.map(&:id_times_ten)

The add_to_projections flag tells the paginator to expose the column expression in the SELECT clause. This is necessary because the keyset pagination needs to somehow extract the last value from the records to request the next page.

iid based ordering

When ordering issues, the database ensures that we have distinct iid values in a project. Ordering by one column is enough to make the pagination work if the project_id filter is present:

order = Gitlab::Pagination::Keyset::Order.build([
  Gitlab::Pagination::Keyset::ColumnOrderDefinition.new(
    attribute_name: 'iid',
    order_expression: Issue.arel_table[:iid].asc,
    nullable: :not_nullable,
    distinct: true
  )
])

scope = Issue.where(project_id: 10).order(order)

scope.keyset_paginate.records # works