spec/lib/gitlab/database/migration_helpers_spec.rb (2,722 lines of code) (raw):
# frozen_string_literal: true
require 'spec_helper'
RSpec.describe Gitlab::Database::MigrationHelpers, feature_category: :database do
include Database::TableSchemaHelpers
include Database::TriggerHelpers
let(:model) do
ActiveRecord::Migration.new.extend(described_class)
end
before do
allow(model).to receive(:puts)
end
it { expect(model.singleton_class.ancestors).to include(described_class::WraparoundVacuumHelpers) }
describe 'overridden dynamic model helpers' do
let(:test_table) { :_test_batching_table }
before do
model.connection.execute(<<~SQL)
CREATE TABLE #{test_table} (
id integer NOT NULL PRIMARY KEY,
name text NOT NULL
);
INSERT INTO #{test_table} (id, name)
VALUES (1, 'bob'), (2, 'mary'), (3, 'amy');
SQL
end
describe '#define_batchable_model' do
it 'defines a batchable model with the migration connection' do
expect(model.define_batchable_model(test_table).count).to eq(3)
end
end
describe '#each_batch' do
before do
allow(model).to receive(:transaction_open?).and_return(false)
end
it 'calls each_batch with the migration connection' do
each_batch_name = ->(&block) do
model.each_batch(test_table, of: 2) do |batch|
block.call(batch.pluck(:name))
end
end
expect { |b| each_batch_name.call(&b) }.to yield_successive_args(%w[bob mary], %w[amy])
end
end
describe '#each_batch_range' do
before do
allow(model).to receive(:transaction_open?).and_return(false)
end
it 'calls each_batch with the migration connection' do
expect { |b| model.each_batch_range(test_table, of: 2, &b) }.to yield_successive_args([1, 2], [3, 3])
end
end
end
describe '#remove_timestamps' do
it 'can remove the default timestamps' do
Gitlab::Database::MigrationHelpers::DEFAULT_TIMESTAMP_COLUMNS.each do |column_name|
expect(model).to receive(:remove_column).with(:foo, column_name)
end
model.remove_timestamps(:foo)
end
it 'can remove custom timestamps' do
expect(model).to receive(:remove_column).with(:foo, :bar)
model.remove_timestamps(:foo, columns: [:bar])
end
end
describe '#add_timestamps_with_timezone' do
it 'adds "created_at" and "updated_at" fields with the "datetime_with_timezone" data type' do
Gitlab::Database::MigrationHelpers::DEFAULT_TIMESTAMP_COLUMNS.each do |column_name|
expect(model).to receive(:add_column)
.with(:foo, column_name, :datetime_with_timezone, { default: nil, null: false })
end
model.add_timestamps_with_timezone(:foo)
end
it 'can disable the NOT NULL constraint' do
Gitlab::Database::MigrationHelpers::DEFAULT_TIMESTAMP_COLUMNS.each do |column_name|
expect(model).to receive(:add_column)
.with(:foo, column_name, :datetime_with_timezone, { default: nil, null: true })
end
model.add_timestamps_with_timezone(:foo, null: true)
end
it 'can add just one column' do
expect(model).to receive(:add_column).with(:foo, :created_at, :datetime_with_timezone, anything)
expect(model).not_to receive(:add_column).with(:foo, :updated_at, :datetime_with_timezone, anything)
model.add_timestamps_with_timezone(:foo, columns: [:created_at])
end
it 'can add choice of acceptable columns' do
expect(model).to receive(:add_column).with(:foo, :created_at, :datetime_with_timezone, anything)
expect(model).to receive(:add_column).with(:foo, :deleted_at, :datetime_with_timezone, anything)
expect(model).to receive(:add_column).with(:foo, :processed_at, :datetime_with_timezone, anything)
expect(model).not_to receive(:add_column).with(:foo, :updated_at, :datetime_with_timezone, anything)
model.add_timestamps_with_timezone(:foo, columns: [:created_at, :deleted_at, :processed_at])
end
it 'cannot add unacceptable column names' do
expect do
model.add_timestamps_with_timezone(:foo, columns: [:bar])
end.to raise_error %r{Illegal timestamp column name}
end
end
describe '#add_concurrent_index' do
context 'outside a transaction' do
before do
allow(model).to receive(:transaction_open?).and_return(false)
allow(model).to receive(:disable_statement_timeout).and_call_original
end
it 'creates the index concurrently' do
expect(model).to receive(:add_index)
.with(:users, :foo, algorithm: :concurrently)
model.add_concurrent_index(:users, :foo)
end
it 'creates unique index concurrently' do
expect(model).to receive(:add_index)
.with(:users, :foo, { algorithm: :concurrently, unique: true })
model.add_concurrent_index(:users, :foo, unique: true)
end
context 'when the index exists and is valid' do
before do
model.add_index :users, :id, unique: true
end
it 'does leaves the existing index' do
expect(model).to receive(:index_exists?)
.with(:users, :id, { algorithm: :concurrently, unique: true }).and_call_original
expect(model).not_to receive(:remove_index)
expect(model).not_to receive(:add_index)
model.add_concurrent_index(:users, :id, unique: true)
end
end
context 'when an invalid copy of the index exists' do
before do
model.add_index :users, :id, unique: true, name: index_name
model.connection.execute(<<~SQL)
UPDATE pg_index
SET indisvalid = false
WHERE indexrelid = '#{index_name}'::regclass
SQL
end
context 'when the default name is used' do
let(:index_name) { model.index_name(:users, :id) }
it 'drops and recreates the index' do
expect(model).to receive(:index_exists?)
.with(:users, :id, { algorithm: :concurrently, unique: true }).and_call_original
expect(model).to receive(:index_invalid?).with(index_name, schema: nil).and_call_original
expect(model).to receive(:remove_concurrent_index_by_name).with(:users, index_name)
expect(model).to receive(:add_index)
.with(:users, :id, { algorithm: :concurrently, unique: true })
model.add_concurrent_index(:users, :id, unique: true)
end
end
context 'when a custom name is used' do
let(:index_name) { 'my_test_index' }
it 'drops and recreates the index' do
expect(model).to receive(:index_exists?)
.with(:users, :id, { algorithm: :concurrently, unique: true, name: index_name }).and_call_original
expect(model).to receive(:index_invalid?).with(index_name, schema: nil).and_call_original
expect(model).to receive(:remove_concurrent_index_by_name).with(:users, index_name)
expect(model).to receive(:add_index)
.with(:users, :id, { algorithm: :concurrently, unique: true, name: index_name })
model.add_concurrent_index(:users, :id, unique: true, name: index_name)
end
end
context 'when a qualified table name is used' do
let(:other_schema) { 'foo_schema' }
let(:index_name) { 'my_test_index' }
let(:table_name) { "#{other_schema}.users" }
before do
model.connection.execute(<<~SQL)
CREATE SCHEMA #{other_schema};
ALTER TABLE users SET SCHEMA #{other_schema};
SQL
end
it 'drops and recreates the index' do
expect(model).to receive(:index_exists?)
.with(table_name, :id, { algorithm: :concurrently, unique: true, name: index_name }).and_call_original
expect(model).to receive(:index_invalid?).with(index_name, schema: other_schema).and_call_original
expect(model).to receive(:remove_concurrent_index_by_name).with(table_name, index_name)
expect(model).to receive(:add_index)
.with(table_name, :id, { algorithm: :concurrently, unique: true, name: index_name })
model.add_concurrent_index(table_name, :id, unique: true, name: index_name)
end
end
end
it 'unprepares the async index creation' do
expect(model).to receive(:add_index)
.with(:users, :foo, algorithm: :concurrently)
expect(model).to receive(:unprepare_async_index)
.with(:users, :foo, algorithm: :concurrently)
model.add_concurrent_index(:users, :foo)
end
context 'when targeting a partition table' do
let(:schema) { 'public' }
let(:name) { :_test_partition_01 }
let(:identifier) { "#{schema}.#{name}" }
before do
model.execute(<<~SQL)
CREATE TABLE public._test_partitioned_table (
id serial NOT NULL,
partition_id serial NOT NULL,
PRIMARY KEY (id, partition_id)
) PARTITION BY LIST(partition_id);
CREATE TABLE #{identifier} PARTITION OF public._test_partitioned_table
FOR VALUES IN (1);
SQL
end
context 'when allow_partition is true' do
it 'creates the index concurrently' do
expect(model).to receive(:add_index).with(:_test_partition_01, :foo, algorithm: :concurrently)
model.add_concurrent_index(:_test_partition_01, :foo, allow_partition: true)
end
end
context 'when allow_partition is not provided' do
it 'raises ArgumentError' do
expect { model.add_concurrent_index(:_test_partition_01, :foo) }
.to raise_error(ArgumentError, /use add_concurrent_partitioned_index/)
end
end
end
end
context 'inside a transaction' do
it 'raises RuntimeError' do
expect(model).to receive(:transaction_open?).and_return(true)
expect { model.add_concurrent_index(:users, :foo) }
.to raise_error(RuntimeError)
end
end
end
describe '#remove_concurrent_index' do
context 'outside a transaction' do
before do
allow(model).to receive(:transaction_open?).and_return(false)
allow(model).to receive(:index_exists?).and_return(true)
allow(model).to receive(:disable_statement_timeout).and_call_original
end
describe 'by column name' do
it 'removes the index concurrently' do
expect(model).to receive(:remove_index)
.with(:users, { algorithm: :concurrently, column: :foo })
model.remove_concurrent_index(:users, :foo)
end
it 'does nothing if the index does not exist' do
expect(model).to receive(:index_exists?)
.with(:users, :foo, { algorithm: :concurrently, unique: true }).and_return(false)
expect(model).not_to receive(:remove_index)
model.remove_concurrent_index(:users, :foo, unique: true)
end
it 'unprepares the async index creation' do
expect(model).to receive(:remove_index)
.with(:users, { algorithm: :concurrently, column: :foo })
expect(model).to receive(:unprepare_async_index)
.with(:users, :foo, { algorithm: :concurrently })
model.remove_concurrent_index(:users, :foo)
end
context 'when targeting a partition table' do
let(:schema) { 'public' }
let(:partition_table_name) { :_test_partition_01 }
let(:identifier) { "#{schema}.#{partition_table_name}" }
let(:index_name) { :_test_partitioned_index }
let(:partition_index_name) { :_test_partition_01_partition_id_idx }
let(:column_name) { 'partition_id' }
before do
model.execute(<<~SQL)
CREATE TABLE public._test_partitioned_table (
id serial NOT NULL,
partition_id serial NOT NULL,
PRIMARY KEY (id, partition_id)
) PARTITION BY LIST(partition_id);
CREATE INDEX #{index_name} ON public._test_partitioned_table(#{column_name});
CREATE TABLE #{identifier} PARTITION OF public._test_partitioned_table
FOR VALUES IN (1);
SQL
end
context 'when dropping an index on the partition table' do
it 'raises ArgumentError' do
expect { model.remove_concurrent_index(partition_table_name, column_name) }
.to raise_error(ArgumentError, /use remove_concurrent_partitioned_index_by_name/)
end
end
end
describe 'by index name' do
before do
allow(model).to receive(:index_exists_by_name?).with(:users, "index_x_by_y").and_return(true)
end
it 'removes the index concurrently by index name' do
expect(model).to receive(:remove_index)
.with(:users, { algorithm: :concurrently, name: "index_x_by_y" })
model.remove_concurrent_index_by_name(:users, "index_x_by_y")
end
it 'does nothing if the index does not exist' do
expect(model).to receive(:index_exists_by_name?).with(:users, "index_x_by_y").and_return(false)
expect(model).not_to receive(:remove_index)
model.remove_concurrent_index_by_name(:users, "index_x_by_y")
end
it 'removes the index with keyword arguments' do
expect(model).to receive(:remove_index)
.with(:users, { algorithm: :concurrently, name: "index_x_by_y" })
model.remove_concurrent_index_by_name(:users, name: "index_x_by_y")
end
it 'raises an error if the index is blank' do
expect do
model.remove_concurrent_index_by_name(:users, wrong_key: "index_x_by_y")
end.to raise_error 'remove_concurrent_index_by_name must get an index name as the second argument'
end
it 'unprepares the async index creation' do
expect(model).to receive(:remove_index)
.with(:users, { algorithm: :concurrently, name: "index_x_by_y" })
expect(model).to receive(:unprepare_async_index_by_name)
.with(:users, "index_x_by_y", { algorithm: :concurrently })
model.remove_concurrent_index_by_name(:users, "index_x_by_y")
end
context 'when targeting a partition table' do
let(:schema) { 'public' }
let(:partition_table_name) { :_test_partition_01 }
let(:identifier) { "#{schema}.#{partition_table_name}" }
let(:index_name) { :_test_partitioned_index }
let(:partition_index_name) { :_test_partition_01_partition_id_idx }
before do
model.execute(<<~SQL)
CREATE TABLE public._test_partitioned_table (
id serial NOT NULL,
partition_id serial NOT NULL,
PRIMARY KEY (id, partition_id)
) PARTITION BY LIST(partition_id);
CREATE INDEX #{index_name} ON public._test_partitioned_table(partition_id);
CREATE TABLE #{identifier} PARTITION OF public._test_partitioned_table
FOR VALUES IN (1);
SQL
end
context 'when dropping an index on the partition table' do
it 'raises ArgumentError' do
expect { model.remove_concurrent_index_by_name(partition_table_name, partition_index_name) }
.to raise_error(ArgumentError, /use remove_concurrent_partitioned_index_by_name/)
end
end
end
end
end
end
context 'inside a transaction' do
it 'raises RuntimeError' do
expect(model).to receive(:transaction_open?).and_return(true)
expect { model.remove_concurrent_index(:users, :foo) }
.to raise_error(RuntimeError)
end
end
end
describe '#remove_foreign_key_if_exists' do
context 'when the foreign key does not exist' do
before do
allow(model).to receive(:foreign_key_exists?).and_return(false)
end
it 'does nothing' do
expect(model).not_to receive(:remove_foreign_key)
model.remove_foreign_key_if_exists(:projects, :users, column: :user_id)
end
end
context 'when the foreign key exists' do
before do
allow(model).to receive(:foreign_key_exists?).and_return(true)
end
it 'removes the foreign key' do
expect(model).to receive(:remove_foreign_key).with(:projects, :users, { column: :user_id })
model.remove_foreign_key_if_exists(:projects, :users, column: :user_id)
end
context 'when the target table is not given' do
it 'passes the options as the second parameter' do
expect(model).to receive(:remove_foreign_key).with(:projects, { column: :user_id })
model.remove_foreign_key_if_exists(:projects, column: :user_id)
end
end
context 'when the reverse_lock_order option is given' do
it 'requests for lock before removing the foreign key' do
expect(model).to receive(:transaction_open?).and_return(true)
expect(model).to receive(:execute).with(/LOCK TABLE users, projects/)
expect(model).not_to receive(:remove_foreign_key).with(:projects, :users)
model.remove_foreign_key_if_exists(:projects, :users, column: :user_id, reverse_lock_order: true)
end
context 'when not inside a transaction' do
it 'does not lock' do
expect(model).to receive(:transaction_open?).and_return(false)
expect(model).not_to receive(:execute).with(/LOCK TABLE users, projects/)
expect(model).to receive(:remove_foreign_key).with(:projects, :users, { column: :user_id })
model.remove_foreign_key_if_exists(:projects, :users, column: :user_id, reverse_lock_order: true)
end
end
end
end
end
describe '#add_concurrent_foreign_key' do
before do
allow(model).to receive(:foreign_key_exists?).and_return(false)
end
context 'inside a transaction' do
it 'raises an error' do
expect(model).to receive(:transaction_open?).and_return(true)
expect do
model.add_concurrent_foreign_key(:projects, :users, column: :user_id)
end.to raise_error(RuntimeError)
end
end
context 'outside a transaction' do
before do
allow(model).to receive(:transaction_open?).and_return(false)
end
context 'target column' do
it 'defaults to (id) when no custom target column is provided' do
expect(model).to receive(:with_lock_retries).and_call_original
expect(model).to receive(:disable_statement_timeout).and_call_original
expect(model).to receive(:statement_timeout_disabled?).and_return(false)
expect(model).to receive(:execute).with(/SET statement_timeout TO/)
expect(model).to receive(:execute).ordered.with(/VALIDATE CONSTRAINT/)
expect(model).to receive(:execute).ordered.with(/RESET statement_timeout/)
expect(model).to receive(:execute).with(/REFERENCES users \(id\)/)
model.add_concurrent_foreign_key(:projects, :users,
column: :user_id)
end
it 'references the custom taget column when provided' do
expect(model).to receive(:with_lock_retries).and_call_original
expect(model).to receive(:disable_statement_timeout).and_call_original
expect(model).to receive(:statement_timeout_disabled?).and_return(false)
expect(model).to receive(:execute).with(/SET statement_timeout TO/)
expect(model).to receive(:execute).ordered.with(/VALIDATE CONSTRAINT/)
expect(model).to receive(:execute).ordered.with(/RESET statement_timeout/)
expect(model).to receive(:execute).with(/REFERENCES users \(id_convert_to_bigint\)/)
model.add_concurrent_foreign_key(:projects, :users,
column: :user_id,
target_column: :id_convert_to_bigint)
end
end
context 'ON DELETE statements' do
context 'on_delete: :nullify' do
it 'appends ON DELETE SET NULL statement' do
expect(model).to receive(:with_lock_retries).and_call_original
expect(model).to receive(:disable_statement_timeout).and_call_original
expect(model).to receive(:statement_timeout_disabled?).and_return(false)
expect(model).to receive(:execute).with(/SET statement_timeout TO/)
expect(model).to receive(:execute).ordered.with(/VALIDATE CONSTRAINT/)
expect(model).to receive(:execute).ordered.with(/RESET statement_timeout/)
expect(model).to receive(:execute).with(/ON DELETE SET NULL/)
model.add_concurrent_foreign_key(:projects, :users,
column: :user_id,
on_delete: :nullify)
end
end
context 'on_delete: :cascade' do
it 'appends ON DELETE CASCADE statement' do
expect(model).to receive(:with_lock_retries).and_call_original
expect(model).to receive(:disable_statement_timeout).and_call_original
expect(model).to receive(:statement_timeout_disabled?).and_return(false)
expect(model).to receive(:execute).with(/SET statement_timeout TO/)
expect(model).to receive(:execute).ordered.with(/VALIDATE CONSTRAINT/)
expect(model).to receive(:execute).ordered.with(/RESET statement_timeout/)
expect(model).to receive(:execute).with(/ON DELETE CASCADE/)
model.add_concurrent_foreign_key(:projects, :users,
column: :user_id,
on_delete: :cascade)
end
end
context 'on_delete: nil' do
it 'appends no ON DELETE statement' do
expect(model).to receive(:with_lock_retries).and_call_original
expect(model).to receive(:disable_statement_timeout).and_call_original
expect(model).to receive(:statement_timeout_disabled?).and_return(false)
expect(model).to receive(:execute).with(/SET statement_timeout TO/)
expect(model).to receive(:execute).ordered.with(/VALIDATE CONSTRAINT/)
expect(model).to receive(:execute).ordered.with(/RESET statement_timeout/)
expect(model).not_to receive(:execute).with(/ON DELETE/)
model.add_concurrent_foreign_key(:projects, :users,
column: :user_id,
on_delete: nil)
end
end
end
context 'ON UPDATE statements' do
context 'on_update: :nullify' do
it 'appends ON UPDATE SET NULL statement' do
expect(model).to receive(:with_lock_retries).and_call_original
expect(model).to receive(:disable_statement_timeout).and_call_original
expect(model).to receive(:statement_timeout_disabled?).and_return(false)
expect(model).to receive(:execute).with(/SET statement_timeout TO/)
expect(model).to receive(:execute).ordered.with(/VALIDATE CONSTRAINT/)
expect(model).to receive(:execute).ordered.with(/RESET statement_timeout/)
expect(model).to receive(:execute).with(/ON UPDATE SET NULL/)
model.add_concurrent_foreign_key(:projects, :users,
column: :user_id,
on_update: :nullify)
end
end
context 'on_update: :cascade' do
it 'appends ON UPDATE CASCADE statement' do
expect(model).to receive(:with_lock_retries).and_call_original
expect(model).to receive(:disable_statement_timeout).and_call_original
expect(model).to receive(:statement_timeout_disabled?).and_return(false)
expect(model).to receive(:execute).with(/SET statement_timeout TO/)
expect(model).to receive(:execute).ordered.with(/VALIDATE CONSTRAINT/)
expect(model).to receive(:execute).ordered.with(/RESET statement_timeout/)
expect(model).to receive(:execute).with(/ON UPDATE CASCADE/)
model.add_concurrent_foreign_key(:projects, :users,
column: :user_id,
on_update: :cascade)
end
end
context 'on_update: nil' do
it 'appends no ON UPDATE statement' do
expect(model).to receive(:with_lock_retries).and_call_original
expect(model).to receive(:disable_statement_timeout).and_call_original
expect(model).to receive(:statement_timeout_disabled?).and_return(false)
expect(model).to receive(:execute).with(/SET statement_timeout TO/)
expect(model).to receive(:execute).ordered.with(/VALIDATE CONSTRAINT/)
expect(model).to receive(:execute).ordered.with(/RESET statement_timeout/)
expect(model).not_to receive(:execute).with(/ON UPDATE/)
model.add_concurrent_foreign_key(:projects, :users,
column: :user_id,
on_update: nil)
end
end
context 'when on_update is not provided' do
it 'appends no ON UPDATE statement' do
expect(model).to receive(:with_lock_retries).and_call_original
expect(model).to receive(:disable_statement_timeout).and_call_original
expect(model).to receive(:statement_timeout_disabled?).and_return(false)
expect(model).to receive(:execute).with(/SET statement_timeout TO/)
expect(model).to receive(:execute).ordered.with(/VALIDATE CONSTRAINT/)
expect(model).to receive(:execute).ordered.with(/RESET statement_timeout/)
expect(model).not_to receive(:execute).with(/ON UPDATE/)
model.add_concurrent_foreign_key(:projects, :users,
column: :user_id)
end
end
end
context 'when no custom key name is supplied' do
it 'creates a concurrent foreign key and validates it' do
expect(model).to receive(:with_lock_retries).and_call_original
expect(model).to receive(:disable_statement_timeout).and_call_original
expect(model).to receive(:statement_timeout_disabled?).and_return(false)
expect(model).to receive(:execute).with(/SET statement_timeout TO/)
expect(model).to receive(:execute).ordered.with(/NOT VALID/)
expect(model).to receive(:execute).ordered.with(/VALIDATE CONSTRAINT/)
expect(model).to receive(:execute).ordered.with(/RESET statement_timeout/)
model.add_concurrent_foreign_key(:projects, :users, column: :user_id)
end
it 'does not create a foreign key if it exists already' do
name = model.concurrent_foreign_key_name(:projects, :user_id)
expect(model).to receive(:foreign_key_exists?).with(:projects, :users,
column: :user_id,
on_update: nil,
on_delete: :cascade,
name: name,
primary_key: :id).and_return(true)
expect(model).not_to receive(:execute).with(/ADD CONSTRAINT/)
expect(model).to receive(:execute).with(/VALIDATE CONSTRAINT/)
model.add_concurrent_foreign_key(:projects, :users, column: :user_id)
end
end
context 'when a custom key name is supplied' do
context 'for creating a new foreign key for a column that does not presently exist' do
it 'creates a new foreign key' do
expect(model).to receive(:with_lock_retries).and_call_original
expect(model).to receive(:disable_statement_timeout).and_call_original
expect(model).to receive(:statement_timeout_disabled?).and_return(false)
expect(model).to receive(:execute).with(/SET statement_timeout TO/)
expect(model).to receive(:execute).ordered.with(/NOT VALID/)
expect(model).to receive(:execute).ordered.with(/VALIDATE CONSTRAINT.+foo/)
expect(model).to receive(:execute).ordered.with(/RESET statement_timeout/)
model.add_concurrent_foreign_key(:projects, :users, column: :user_id, name: :foo)
end
end
context 'for creating a duplicate foreign key for a column that presently exists' do
context 'when the supplied key name is the same as the existing foreign key name' do
it 'does not create a new foreign key' do
expect(model).to receive(:foreign_key_exists?).with(:projects, :users,
name: :foo,
primary_key: :id,
on_update: nil,
on_delete: :cascade,
column: :user_id).and_return(true)
expect(model).not_to receive(:execute).with(/ADD CONSTRAINT/)
expect(model).to receive(:execute).with(/VALIDATE CONSTRAINT/)
model.add_concurrent_foreign_key(:projects, :users, column: :user_id, name: :foo)
end
end
context 'when the supplied key name is different from the existing foreign key name' do
it 'creates a new foreign key' do
expect(model).to receive(:with_lock_retries).and_call_original
expect(model).to receive(:disable_statement_timeout).and_call_original
expect(model).to receive(:statement_timeout_disabled?).and_return(false)
expect(model).to receive(:execute).with(/SET statement_timeout TO/)
expect(model).to receive(:execute).ordered.with(/NOT VALID/)
expect(model).to receive(:execute).ordered.with(/VALIDATE CONSTRAINT.+bar/)
expect(model).to receive(:execute).ordered.with(/RESET statement_timeout/)
model.add_concurrent_foreign_key(:projects, :users, column: :user_id, name: :bar)
end
end
end
end
describe 'validate option' do
let(:args) { [:projects, :users] }
let(:options) { { column: :user_id, on_delete: nil } }
context 'when validate is supplied with a falsey value' do
it_behaves_like 'skips validation', validate: false
it_behaves_like 'skips validation', validate: nil
end
context 'when validate is supplied with a truthy value' do
it_behaves_like 'performs validation', validate: true
it_behaves_like 'performs validation', validate: :whatever
end
context 'when validate is not supplied' do
it_behaves_like 'performs validation', {}
end
context "when a ForeignKeyViolation occurs" do
let(:source) { 'projects' }
let(:constraint_name) { 'fk_projects_users_id' }
let(:options) { { column: :user_id, name: constraint_name } }
it "drops the constraint and raises an error", :aggregate_failures do
expect(model).to receive(:disable_statement_timeout).and_call_original
expect(model).to receive(:statement_timeout_disabled?).and_return(false)
expect(model).to receive(:execute).with(
"ALTER TABLE projects ADD CONSTRAINT fk_projects_users_id FOREIGN KEY (user_id) REFERENCES users (id) ON DELETE CASCADE NOT VALID;"
)
expect(model).to receive(:execute).with(/SET statement_timeout TO/).ordered
expect(model).to receive(:execute).with(/ALTER TABLE .* VALIDATE CONSTRAINT/).and_raise(PG::ForeignKeyViolation.new("foreign key violation")).ordered
expect(model).to receive(:execute).with(/RESET statement_timeout/).ordered
expect(model).to receive(:execute).with(/ALTER TABLE #{source} DROP CONSTRAINT #{constraint_name}/).ordered
expect do
model.add_concurrent_foreign_key(source, :users, **options)
end.to raise_error %r{Migration failed intentionally due to ForeignKeyViolation}
end
end
end
context 'when the reverse_lock_order flag is set' do
it 'explicitly locks the tables in target-source order', :aggregate_failures do
expect(model).to receive(:with_lock_retries).and_call_original
expect(model).to receive(:disable_statement_timeout).and_call_original
expect(model).to receive(:statement_timeout_disabled?).and_return(false)
expect(model).to receive(:execute).with(/SET statement_timeout TO/)
expect(model).to receive(:execute).ordered.with(/VALIDATE CONSTRAINT/)
expect(model).to receive(:execute).ordered.with(/RESET statement_timeout/)
expect(model).to receive(:execute).with('LOCK TABLE users, projects IN SHARE ROW EXCLUSIVE MODE')
expect(model).to receive(:execute).with(/REFERENCES users \(id\)/)
model.add_concurrent_foreign_key(:projects, :users, column: :user_id, reverse_lock_order: true)
end
end
context 'when creating foreign key for a group of columns' do
it 'references the custom target columns when provided', :aggregate_failures do
expect(model).to receive(:with_lock_retries).and_yield
expect(model).to receive(:execute).with(
"ALTER TABLE projects " \
"ADD CONSTRAINT fk_multiple_columns " \
"FOREIGN KEY \(partition_number, user_id\) " \
"REFERENCES users \(partition_number, id\) " \
"ON UPDATE CASCADE " \
"ON DELETE CASCADE " \
"NOT VALID;"
)
model.add_concurrent_foreign_key(
:projects,
:users,
column: [:partition_number, :user_id],
target_column: [:partition_number, :id],
validate: false,
name: :fk_multiple_columns,
on_update: :cascade
)
end
context 'when foreign key is already defined' do
before do
expect(model).to receive(:foreign_key_exists?).with(
:projects,
:users,
{
column: [:partition_number, :user_id],
name: :fk_multiple_columns,
on_update: :cascade,
on_delete: :cascade,
primary_key: [:partition_number, :id]
}
).and_return(true)
end
it 'does not create foreign key', :aggregate_failures do
expect(model).not_to receive(:with_lock_retries).and_yield
expect(model).not_to receive(:execute).with(/FOREIGN KEY/)
model.add_concurrent_foreign_key(
:projects,
:users,
column: [:partition_number, :user_id],
target_column: [:partition_number, :id],
on_update: :cascade,
validate: false,
name: :fk_multiple_columns
)
end
end
end
context 'when creating foreign key on a partitioned table' do
let(:source) { :_test_source_partitioned_table }
let(:dest) { :_test_dest_partitioned_table }
let(:args) { [source, dest] }
let(:options) { { column: [:partition_id, :owner_id], target_column: [:partition_id, :id] } }
before do
model.execute(<<~SQL)
CREATE TABLE public.#{source} (
id serial NOT NULL,
partition_id serial NOT NULL,
owner_id bigint NOT NULL,
PRIMARY KEY (id, partition_id)
) PARTITION BY LIST(partition_id);
CREATE TABLE #{source}_1
PARTITION OF public.#{source}
FOR VALUES IN (1);
CREATE TABLE public.#{dest} (
id serial NOT NULL,
partition_id serial NOT NULL,
PRIMARY KEY (id, partition_id)
);
SQL
end
it 'creates the FK without using NOT VALID', :aggregate_failures do
allow(model).to receive(:execute).and_call_original
expect(model).to receive(:with_lock_retries).and_yield
expect(model).to receive(:execute).with(
"ALTER TABLE #{source} " \
"ADD CONSTRAINT fk_multiple_columns " \
"FOREIGN KEY \(partition_id, owner_id\) " \
"REFERENCES #{dest} \(partition_id, id\) " \
"ON UPDATE CASCADE ON DELETE CASCADE ;"
)
model.add_concurrent_foreign_key(
*args,
name: :fk_multiple_columns,
on_update: :cascade,
allow_partitioned: true,
**options
)
end
it 'raises an error if allow_partitioned is not set' do
expect(model).not_to receive(:with_lock_retries).and_yield
expect(model).not_to receive(:execute).with(/FOREIGN KEY/)
expect { model.add_concurrent_foreign_key(*args, **options) }
.to raise_error ArgumentError, /use add_concurrent_partitioned_foreign_key/
end
context 'when the reverse_lock_order flag is set' do
it 'explicitly locks the tables in target-source order', :aggregate_failures do
expect(model).to receive(:with_lock_retries).and_call_original
expect(model).to receive(:disable_statement_timeout).and_call_original
expect(model).to receive(:statement_timeout_disabled?).and_return(false)
expect(model).to receive(:execute).with(/SET statement_timeout TO/)
expect(model).to receive(:execute).ordered.with(/VALIDATE CONSTRAINT/)
expect(model).to receive(:execute).ordered.with(/RESET statement_timeout/)
expect(model).to receive(:execute).with("LOCK TABLE #{dest}, #{source} IN ACCESS EXCLUSIVE MODE")
expect(model).to receive(:execute).with(/REFERENCES #{dest} \(partition_id, id\)/)
model.add_concurrent_foreign_key(*args, reverse_lock_order: true, allow_partitioned: true, **options)
end
end
end
end
end
describe '#validate_foreign_key' do
context 'when name is provided' do
it 'does not infer the foreign key constraint name' do
expect(model).to receive(:foreign_key_exists?).with(:projects, name: :foo).and_return(true)
aggregate_failures do
expect(model).not_to receive(:concurrent_foreign_key_name)
expect(model).to receive(:disable_statement_timeout).and_call_original
expect(model).to receive(:statement_timeout_disabled?).and_return(false)
expect(model).to receive(:execute).with(/SET statement_timeout TO/)
expect(model).to receive(:execute).ordered.with(/ALTER TABLE projects VALIDATE CONSTRAINT/)
expect(model).to receive(:execute).ordered.with(/RESET statement_timeout/)
end
model.validate_foreign_key(:projects, :user_id, name: :foo)
end
end
context 'when name is not provided' do
it 'infers the foreign key constraint name' do
expect(model).to receive(:foreign_key_exists?).with(:projects, name: anything).and_return(true)
aggregate_failures do
expect(model).to receive(:concurrent_foreign_key_name)
expect(model).to receive(:disable_statement_timeout).and_call_original
expect(model).to receive(:statement_timeout_disabled?).and_return(false)
expect(model).to receive(:execute).with(/SET statement_timeout TO/)
expect(model).to receive(:execute).ordered.with(/ALTER TABLE projects VALIDATE CONSTRAINT/)
expect(model).to receive(:execute).ordered.with(/RESET statement_timeout/)
end
model.validate_foreign_key(:projects, :user_id)
end
context 'when the inferred foreign key constraint does not exist' do
it 'raises an error' do
expect(model).to receive(:foreign_key_exists?).and_return(false)
error_message = /Could not find foreign key "fk_name" on table "projects"/
expect { model.validate_foreign_key(:projects, :user_id, name: :fk_name) }.to raise_error(error_message)
end
end
end
end
describe '#concurrent_foreign_key_name' do
it 'returns the name for a foreign key' do
name = model.concurrent_foreign_key_name(:this_is_a_very_long_table_name,
:with_a_very_long_column_name)
expect(name).to be_an_instance_of(String)
expect(name.length).to eq(13)
end
context 'when using multiple columns' do
it 'returns the name of the foreign key', :aggregate_failures do
result = model.concurrent_foreign_key_name(:table_name, [:partition_number, :id])
expect(result).to be_an_instance_of(String)
expect(result.length).to eq(13)
end
end
end
describe '#foreign_key_exists?' do
let(:referenced_table_name) { :_test_gitlab_main_referenced }
let(:referencing_table_name) { :_test_gitlab_main_referencing }
let(:schema) { 'public' }
let(:identifier) { "#{schema}.#{referencing_table_name}" }
before do
model.connection.execute(<<~SQL)
create table #{referenced_table_name} (
id bigserial primary key not null
);
create table #{referencing_table_name} (
id bigserial primary key not null,
non_standard_id bigint not null,
constraint fk_referenced foreign key (non_standard_id)
references #{referenced_table_name}(id) on delete cascade
);
SQL
end
shared_examples_for 'foreign key checks' do
it 'finds existing foreign keys by column' do
expect(model.foreign_key_exists?(referencing_table_name, target_table, column: :non_standard_id)).to be_truthy
end
it 'finds existing foreign keys by name' do
expect(model.foreign_key_exists?(referencing_table_name, target_table, name: :fk_referenced)).to be_truthy
end
it 'finds existing foreign_keys by name and column' do
expect(model.foreign_key_exists?(referencing_table_name, target_table, name: :fk_referenced, column: :non_standard_id)).to be_truthy
end
it 'finds existing foreign_keys by name, column and on_delete' do
expect(model.foreign_key_exists?(referencing_table_name, target_table, name: :fk_referenced, column: :non_standard_id, on_delete: :cascade)).to be_truthy
end
it 'finds existing foreign keys by target table only' do
expect(model.foreign_key_exists?(referencing_table_name, target_table)).to be_truthy
end
it 'finds existing foreign_keys by identifier' do
expect(model.foreign_key_exists?(identifier, target_table)).to be_truthy
end
it 'compares by column name if given' do
expect(model.foreign_key_exists?(referencing_table_name, target_table, column: :user_id)).to be_falsey
end
it 'compares by target column name if given' do
expect(model.foreign_key_exists?(referencing_table_name, target_table, primary_key: :user_id)).to be_falsey
expect(model.foreign_key_exists?(referencing_table_name, target_table, primary_key: :id)).to be_truthy
end
it 'compares by foreign key name if given' do
expect(model.foreign_key_exists?(referencing_table_name, target_table, name: :non_existent_foreign_key_name)).to be_falsey
end
it 'compares by foreign key name and column if given' do
expect(model.foreign_key_exists?(referencing_table_name, target_table, name: :non_existent_foreign_key_name, column: :non_standard_id)).to be_falsey
end
it 'compares by foreign key name, column and on_delete if given' do
expect(model.foreign_key_exists?(referencing_table_name, target_table, name: :fk_referenced, column: :non_standard_id, on_delete: :nullify)).to be_falsey
end
end
context 'without specifying a target table' do
let(:target_table) { nil }
it_behaves_like 'foreign key checks'
end
context 'specifying a target table' do
let(:target_table) { referenced_table_name }
it_behaves_like 'foreign key checks'
end
context 'if the schema cache does not include the constrained_columns column' do
let(:target_table) { nil }
around do |ex|
model.transaction do
require_relative '../../../fixtures/migrations/db/migrate/schema_cache_migration_test'
# Uses the init_schema migration, as it is always present in the codebase (not affected by squashing process)
require_migration!('init_schema')
InitSchema.prepend(SchemaCacheMigrationTest)
InitSchema.new.down
Gitlab::Database::PostgresForeignKey.reset_column_information
Gitlab::Database::PostgresForeignKey.columns_hash # Force populate the column hash in the old schema
InitSchema.new.up
# Rolling back reverts the schema cache information, so we need to run the example here before the rollback.
ex.run
raise ActiveRecord::Rollback
end
# make sure that we're resetting the schema cache here so that we don't leak the change to other tests.
Gitlab::Database::PostgresForeignKey.reset_column_information
# Double-check that the column information is back to normal
expect(Gitlab::Database::PostgresForeignKey.columns_hash.keys).to include('constrained_columns')
end
# This test verifies that the situation we're trying to set up for the shared examples is actually being
# set up correctly
it 'correctly sets up the test without the column in the columns_hash' do
expect(Gitlab::Database::PostgresForeignKey.columns_hash.keys).not_to include('constrained_columns')
end
it_behaves_like 'foreign key checks'
end
it 'compares by target table if no column given' do
expect(model.foreign_key_exists?(:projects, :other_table)).to be_falsey
end
it 'raises an error if an invalid on_delete is specified' do
# The correct on_delete key is "nullify"
expect { model.foreign_key_exists?(referenced_table_name, on_delete: :set_null) }.to raise_error(ArgumentError)
end
context 'with foreign key using multiple columns' do
let(:p_referenced_table_name) { :_test_gitlab_main_p_referenced }
let(:p_referencing_table_name) { :_test_gitlab_main_p_referencing }
before do
model.connection.execute(<<~SQL)
create table #{p_referenced_table_name} (
id bigserial not null,
partition_number bigint not null default 100,
primary key (partition_number, id)
);
create table #{p_referencing_table_name} (
id bigserial primary key not null,
partition_number bigint not null,
constraint fk_partitioning foreign key (partition_number, id)
references #{p_referenced_table_name} (partition_number, id) on delete cascade
);
SQL
end
it 'finds existing foreign keys by columns' do
expect(model.foreign_key_exists?(p_referencing_table_name, p_referenced_table_name,
column: [:partition_number, :id])).to be_truthy
end
it 'finds existing foreign keys by name' do
expect(model.foreign_key_exists?(p_referencing_table_name, p_referenced_table_name,
name: :fk_partitioning)).to be_truthy
end
it 'finds existing foreign_keys by name and column' do
expect(model.foreign_key_exists?(p_referencing_table_name, p_referenced_table_name,
name: :fk_partitioning, column: [:partition_number, :id])).to be_truthy
end
it 'finds existing foreign_keys by name, column and on_delete' do
expect(model.foreign_key_exists?(p_referencing_table_name, p_referenced_table_name,
name: :fk_partitioning, column: [:partition_number, :id], on_delete: :cascade)).to be_truthy
end
it 'finds existing foreign keys by target table only' do
expect(model.foreign_key_exists?(p_referencing_table_name, p_referenced_table_name)).to be_truthy
end
it 'compares by column name if given' do
expect(model.foreign_key_exists?(p_referencing_table_name, p_referenced_table_name,
column: :id)).to be_falsey
end
it 'compares by target column name if given' do
expect(model.foreign_key_exists?(p_referencing_table_name, p_referenced_table_name,
primary_key: :user_id)).to be_falsey
expect(model.foreign_key_exists?(p_referencing_table_name, p_referenced_table_name,
primary_key: [:partition_number, :id])).to be_truthy
end
it 'compares by foreign key name if given' do
expect(model.foreign_key_exists?(p_referencing_table_name, p_referenced_table_name,
name: :non_existent_foreign_key_name)).to be_falsey
end
it 'compares by foreign key name and column if given' do
expect(model.foreign_key_exists?(p_referencing_table_name, p_referenced_table_name,
name: :non_existent_foreign_key_name, column: [:partition_number, :id])).to be_falsey
end
it 'compares by foreign key name, column and on_delete if given' do
expect(model.foreign_key_exists?(p_referencing_table_name, p_referenced_table_name,
name: :fk_partitioning, column: [:partition_number, :id], on_delete: :nullify)).to be_falsey
end
end
end
describe '#true_value' do
it 'returns the appropriate value' do
expect(model.true_value).to eq("'t'")
end
end
describe '#false_value' do
it 'returns the appropriate value' do
expect(model.false_value).to eq("'f'")
end
end
describe '#update_column_in_batches' do
context 'when running outside of a transaction' do
before do
expect(model).to receive(:transaction_open?).and_return(false)
create_list(:project, 5)
end
it 'updates all the rows in a table' do
model.update_column_in_batches(:projects, :description_html, 'foo')
expect(Project.where(description_html: 'foo').count).to eq(5)
end
it 'updates boolean values correctly' do
model.update_column_in_batches(:projects, :archived, true)
expect(Project.where(archived: true).count).to eq(5)
end
context 'when a block is supplied' do
it 'yields an Arel table and query object to the supplied block' do
first_id = Project.first.id
model.update_column_in_batches(:projects, :archived, true) do |t, query|
query.where(t[:id].eq(first_id))
end
expect(Project.where(archived: true).count).to eq(1)
end
end
context 'when the value is Arel.sql (Arel::Nodes::SqlLiteral)' do
it 'updates the value as a SQL expression' do
model.update_column_in_batches(:projects, :star_count, Arel.sql('1+1'))
expect(Project.sum(:star_count)).to eq(2 * Project.count)
end
end
context 'when the table is write-locked' do
let(:test_table) { :_test_table }
let(:lock_writes_manager) do
Gitlab::Database::LockWritesManager.new(
table_name: test_table,
connection: model.connection,
database_name: 'main',
with_retries: false
)
end
before do
model.connection.execute(<<~SQL)
CREATE TABLE #{test_table} (id integer NOT NULL, value integer NOT NULL DEFAULT 0);
INSERT INTO #{test_table} (id, value)
VALUES (1, 1), (2, 2), (3, 3)
SQL
lock_writes_manager.lock_writes
end
it 'disables the write-lock trigger function' do
expect do
model.update_column_in_batches(test_table, :value, Arel.sql('1+1'), disable_lock_writes: true)
end.not_to raise_error
end
it 'raises an error if it does not disable the trigger function' do
expect do
model.update_column_in_batches(test_table, :value, Arel.sql('1+1'), disable_lock_writes: false)
end.to raise_error(ActiveRecord::StatementInvalid, /Table: "#{test_table}" is write protected/)
end
end
end
context 'when running inside the transaction' do
it 'raises RuntimeError' do
expect(model).to receive(:transaction_open?).and_return(true)
expect do
model.update_column_in_batches(:projects, :star_count, Arel.sql('1+1'))
end.to raise_error(RuntimeError)
end
end
end
describe '#rename_column_concurrently' do
context 'in a transaction' do
it 'raises RuntimeError' do
allow(model).to receive(:transaction_open?).and_return(true)
expect { model.rename_column_concurrently(:users, :old, :new) }
.to raise_error(RuntimeError)
end
end
context 'outside a transaction' do
let(:old_column) do
double(:column,
type: :integer,
limit: 8,
default: 0,
null: false,
precision: 5,
scale: 1)
end
let(:trigger_name) { model.rename_trigger_name(:users, :old, :new) }
before do
allow(model).to receive(:transaction_open?).and_return(false)
end
context 'when the column to rename exists' do
before do
allow(model).to receive(:column_for).and_return(old_column)
end
it 'renames a column concurrently' do
expect(Gitlab::Database::QueryAnalyzers::GitlabSchemasValidateConnection).to receive(:with_suppressed).and_yield
expect(model).to receive(:check_trigger_permissions!).with(:users)
expect(model).to receive(:install_rename_triggers)
.with(:users, :old, :new)
expect(model).to receive(:add_column)
.with(:users, :new, :integer,
limit: old_column.limit,
precision: old_column.precision,
scale: old_column.scale)
expect(model).to receive(:change_column_default)
.with(:users, :new, old_column.default)
expect(model).to receive(:update_column_in_batches)
expect(model).to receive(:add_not_null_constraint).with(:users, :new)
expect(model).to receive(:copy_indexes).with(:users, :old, :new)
expect(model).to receive(:copy_foreign_keys).with(:users, :old, :new)
expect(model).to receive(:copy_check_constraints).with(:users, :old, :new)
model.rename_column_concurrently(:users, :old, :new)
end
context 'with existing records and type casting' do
let(:trigger_name) { model.rename_trigger_name(:users, :id, :new) }
let(:user) { create(:user) }
let(:copy_trigger) { double('copy trigger') }
let(:connection) { ActiveRecord::Migration.connection }
before do
expect(Gitlab::Database::QueryAnalyzers::GitlabSchemasValidateConnection).to receive(:with_suppressed).and_yield
expect(Gitlab::Database::UnidirectionalCopyTrigger).to receive(:on_table)
.with(:users, connection: connection).and_return(copy_trigger)
end
it 'copies the value to the new column using the type_cast_function', :aggregate_failures do
expect(model).to receive(:copy_indexes).with(:users, :id, :new)
expect(model).to receive(:add_not_null_constraint).with(:users, :new)
expect(model).to receive(:execute).with("SELECT set_config('lock_writes.users', 'false', true)")
expect(model).to receive(:execute).with("UPDATE \"users\" SET \"new\" = cast_to_jsonb_with_default(\"users\".\"id\") WHERE \"users\".\"id\" >= #{user.id}")
expect(copy_trigger).to receive(:create).with(:id, :new, trigger_name: nil)
model.rename_column_concurrently(:users, :id, :new, type_cast_function: 'cast_to_jsonb_with_default')
end
end
it 'passes the batch_column_name' do
expect(model).to receive(:column_exists?).with(:users, :other_batch_column).and_return(true)
expect(model).to receive(:check_trigger_permissions!).and_return(true)
expect(model).to receive(:create_column_from).with(
:users, :old, :new, type: nil, batch_column_name: :other_batch_column, type_cast_function: nil
).and_return(true)
expect(model).to receive(:install_rename_triggers).and_return(true)
model.rename_column_concurrently(:users, :old, :new, batch_column_name: :other_batch_column)
end
it 'passes the type_cast_function' do
expect(model).to receive(:create_column_from).with(
:users, :old, :new, type: nil, batch_column_name: :id, type_cast_function: 'JSON'
).and_return(true)
model.rename_column_concurrently(:users, :old, :new, type_cast_function: 'JSON')
end
it 'raises an error with invalid batch_column_name' do
expect do
model.rename_column_concurrently(:users, :old, :new, batch_column_name: :invalid)
end.to raise_error(RuntimeError, /Column invalid does not exist on users/)
end
context 'when default is false' do
let(:old_column) do
double(:column,
type: :boolean,
limit: nil,
default: false,
null: false,
precision: nil,
scale: nil)
end
it 'copies the default to the new column' do
expect(Gitlab::Database::QueryAnalyzers::GitlabSchemasValidateConnection).to receive(:with_suppressed).and_yield
expect(model).to receive(:change_column_default)
.with(:users, :new, old_column.default)
expect(model).to receive(:copy_check_constraints)
.with(:users, :old, :new)
model.rename_column_concurrently(:users, :old, :new)
end
end
end
context 'when the table in the other database is write-locked' do
let(:test_table) { :_test_table }
let(:lock_writes_manager) do
Gitlab::Database::LockWritesManager.new(
table_name: test_table,
connection: model.connection,
database_name: 'main',
with_retries: false
)
end
before do
model.connection.execute(<<~SQL)
CREATE TABLE #{test_table} (id integer NOT NULL, value integer NOT NULL DEFAULT 0);
INSERT INTO #{test_table} (id, value)
VALUES (1, 1), (2, 2), (3, 3)
SQL
lock_writes_manager.lock_writes
end
it 'does not raise an error when renaming the column' do
expect do
model.rename_column_concurrently(test_table, :value, :new_value)
end.not_to raise_error
end
end
context 'when the column to be renamed does not exist' do
before do
allow(model).to receive(:columns).and_return([])
end
it 'raises an error with appropriate message' do
expect(model).to receive(:check_trigger_permissions!).with(:users)
error_message = /Could not find column "missing_column" on table "users"/
expect { model.rename_column_concurrently(:users, :missing_column, :new) }.to raise_error(error_message)
end
end
end
end
describe '#undo_rename_column_concurrently' do
it 'reverses the operations of rename_column_concurrently' do
expect(model).to receive(:check_trigger_permissions!).with(:users)
expect(model).to receive(:remove_rename_triggers)
.with(:users, /trigger_.{12}/)
expect(model).to receive(:remove_column).with(:users, :new)
model.undo_rename_column_concurrently(:users, :old, :new)
end
end
describe '#cleanup_concurrent_column_rename' do
it 'cleans up the renaming procedure' do
expect(model).to receive(:check_trigger_permissions!).with(:users)
expect(model).to receive(:remove_rename_triggers)
.with(:users, /trigger_.{12}/)
expect(model).to receive(:remove_column).with(:users, :old)
model.cleanup_concurrent_column_rename(:users, :old, :new)
end
end
describe '#undo_cleanup_concurrent_column_rename' do
context 'in a transaction' do
it 'raises RuntimeError' do
allow(model).to receive(:transaction_open?).and_return(true)
expect { model.undo_cleanup_concurrent_column_rename(:users, :old, :new) }
.to raise_error(RuntimeError)
end
end
context 'outside a transaction' do
let(:new_column) do
double(:column,
type: :integer,
limit: 8,
default: 0,
null: false,
precision: 5,
scale: 1)
end
let(:trigger_name) { model.rename_trigger_name(:users, :old, :new) }
before do
allow(model).to receive(:transaction_open?).and_return(false)
allow(model).to receive(:column_for).and_return(new_column)
end
it 'reverses the operations of cleanup_concurrent_column_rename' do
expect(Gitlab::Database::QueryAnalyzers::GitlabSchemasValidateConnection).to receive(:with_suppressed).and_yield
expect(model).to receive(:check_trigger_permissions!).with(:users)
expect(model).to receive(:install_rename_triggers)
.with(:users, :old, :new)
expect(model).to receive(:add_column)
.with(:users, :old, :integer,
limit: new_column.limit,
precision: new_column.precision,
scale: new_column.scale)
expect(model).to receive(:change_column_default)
.with(:users, :old, new_column.default)
expect(model).to receive(:update_column_in_batches)
expect(model).to receive(:add_not_null_constraint).with(:users, :old)
expect(model).to receive(:copy_indexes).with(:users, :new, :old)
expect(model).to receive(:copy_foreign_keys).with(:users, :new, :old)
expect(model).to receive(:copy_check_constraints).with(:users, :new, :old)
model.undo_cleanup_concurrent_column_rename(:users, :old, :new)
end
it 'passes the batch_column_name' do
expect(model).to receive(:column_exists?).with(:users, :other_batch_column).and_return(true)
expect(model).to receive(:check_trigger_permissions!).and_return(true)
expect(model).to receive(:create_column_from).with(
:users, :new, :old, type: nil, batch_column_name: :other_batch_column
).and_return(true)
expect(model).to receive(:install_rename_triggers).and_return(true)
model.undo_cleanup_concurrent_column_rename(:users, :old, :new, batch_column_name: :other_batch_column)
end
it 'raises an error with invalid batch_column_name' do
expect do
model.undo_cleanup_concurrent_column_rename(:users, :old, :new, batch_column_name: :invalid)
end.to raise_error(RuntimeError, /Column invalid does not exist on users/)
end
context 'when default is false' do
let(:new_column) do
double(:column,
type: :boolean,
limit: nil,
default: false,
null: false,
precision: nil,
scale: nil)
end
it 'copies the default to the old column' do
expect(Gitlab::Database::QueryAnalyzers::GitlabSchemasValidateConnection).to receive(:with_suppressed).and_yield
expect(model).to receive(:change_column_default)
.with(:users, :old, new_column.default)
expect(model).to receive(:copy_check_constraints)
.with(:users, :new, :old)
model.undo_cleanup_concurrent_column_rename(:users, :old, :new)
end
end
end
end
describe '#change_column_type_concurrently' do
it 'changes the column type' do
expect(model).to receive(:rename_column_concurrently)
.with('users', 'username', 'username_for_type_change', type: :text, type_cast_function: nil, batch_column_name: :id)
model.change_column_type_concurrently('users', 'username', :text)
end
it 'passed the batch column name' do
expect(model).to receive(:rename_column_concurrently)
.with('users', 'username', 'username_for_type_change', type: :text, type_cast_function: nil, batch_column_name: :user_id)
model.change_column_type_concurrently('users', 'username', :text, batch_column_name: :user_id)
end
context 'with type cast' do
it 'changes the column type with casting the value to the new type' do
expect(model).to receive(:rename_column_concurrently)
.with('users', 'username', 'username_for_type_change', type: :text, type_cast_function: 'JSON', batch_column_name: :id)
model.change_column_type_concurrently('users', 'username', :text, type_cast_function: 'JSON')
end
end
end
describe '#undo_change_column_type_concurrently' do
it 'reverses the operations of change_column_type_concurrently' do
expect(model).to receive(:check_trigger_permissions!).with(:users)
expect(model).to receive(:remove_rename_triggers)
.with(:users, /trigger_.{12}/)
expect(model).to receive(:remove_column).with(:users, "old_for_type_change")
model.undo_change_column_type_concurrently(:users, :old)
end
end
describe '#cleanup_concurrent_column_type_change' do
it 'cleans up the type changing procedure' do
expect(model).to receive(:cleanup_concurrent_column_rename)
.with('users', 'username', 'username_for_type_change')
expect(model).to receive(:rename_column)
.with('users', 'username_for_type_change', 'username')
model.cleanup_concurrent_column_type_change('users', 'username')
end
end
describe '#undo_cleanup_concurrent_column_type_change' do
context 'in a transaction' do
it 'raises RuntimeError' do
allow(model).to receive(:transaction_open?).and_return(true)
expect { model.undo_cleanup_concurrent_column_type_change(:users, :old, :new) }
.to raise_error(RuntimeError)
end
end
context 'outside a transaction' do
let(:temp_column) { "old_for_type_change" }
let(:temp_undo_cleanup_column) do
identifier = "users_old_for_type_change"
hashed_identifier = Digest::SHA256.hexdigest(identifier).first(10)
"tmp_undo_cleanup_column_#{hashed_identifier}"
end
let(:trigger_name) { model.rename_trigger_name(:users, :old, :old_for_type_change) }
before do
allow(model).to receive(:transaction_open?).and_return(false)
end
it 'reverses the operations of cleanup_concurrent_column_type_change' do
expect(Gitlab::Database::QueryAnalyzers::RestrictAllowedSchemas).to receive(:require_ddl_mode!)
expect(model).to receive(:check_trigger_permissions!).with(:users)
expect(model).to receive(:create_column_from).with(
:users,
:old,
temp_undo_cleanup_column,
type: :string,
batch_column_name: :id,
type_cast_function: nil,
limit: nil
).and_return(true)
expect(model).to receive(:rename_column)
.with(:users, :old, temp_column)
expect(model).to receive(:rename_column)
.with(:users, temp_undo_cleanup_column, :old)
expect(model).to receive(:install_rename_triggers)
.with(:users, :old, 'old_for_type_change')
model.undo_cleanup_concurrent_column_type_change(:users, :old, :string)
end
it 'passes the type_cast_function, batch_column_name and limit' do
expect(Gitlab::Database::QueryAnalyzers::RestrictAllowedSchemas).to receive(:require_ddl_mode!)
expect(model).to receive(:column_exists?).with(:users, :other_batch_column).and_return(true)
expect(model).to receive(:check_trigger_permissions!).with(:users)
expect(model).to receive(:create_column_from).with(
:users,
:old,
temp_undo_cleanup_column,
type: :string,
batch_column_name: :other_batch_column,
type_cast_function: :custom_type_cast_function,
limit: 8
).and_return(true)
expect(model).to receive(:rename_column)
.with(:users, :old, temp_column)
expect(model).to receive(:rename_column)
.with(:users, temp_undo_cleanup_column, :old)
expect(model).to receive(:install_rename_triggers)
.with(:users, :old, 'old_for_type_change')
model.undo_cleanup_concurrent_column_type_change(
:users,
:old,
:string,
type_cast_function: :custom_type_cast_function,
batch_column_name: :other_batch_column,
limit: 8
)
end
it 'raises an error with invalid batch_column_name' do
expect do
model.undo_cleanup_concurrent_column_type_change(:users, :old, :new, batch_column_name: :invalid)
end.to raise_error(RuntimeError, /Column invalid does not exist on users/)
end
end
end
describe '#install_rename_triggers' do
let(:connection) { ActiveRecord::Migration.connection }
it 'installs the triggers' do
copy_trigger = double('copy trigger')
expect(Gitlab::Database::UnidirectionalCopyTrigger).to receive(:on_table)
.with(:users, connection: connection).and_return(copy_trigger)
expect(copy_trigger).to receive(:create).with(:old, :new, trigger_name: 'foo')
model.install_rename_triggers(:users, :old, :new, trigger_name: 'foo')
end
end
describe '#remove_rename_triggers' do
let(:connection) { ActiveRecord::Migration.connection }
it 'removes the function and trigger' do
copy_trigger = double('copy trigger')
expect(Gitlab::Database::UnidirectionalCopyTrigger).to receive(:on_table)
.with('bar', connection: connection).and_return(copy_trigger)
expect(copy_trigger).to receive(:drop).with('foo')
model.remove_rename_triggers('bar', 'foo')
end
end
describe '#rename_trigger_name' do
it 'returns a String' do
expect(model.rename_trigger_name(:users, :foo, :bar))
.to match(/trigger_.{12}/)
end
end
describe '#install_sharding_key_assignment_trigger' do
let(:trigger) { double }
let(:connection) { ActiveRecord::Base.connection }
it do
expect(Gitlab::Database::Triggers::AssignDesiredShardingKey).to receive(:new)
.with(table: :test_table, sharding_key: :project_id, parent_table: :parent_table, parent_table_primary_key: :project_id,
parent_sharding_key: :parent_project_id, foreign_key: :foreign_key, connection: connection,
trigger_name: 'trigger_name').and_return(trigger)
expect(trigger).to receive(:create)
model.install_sharding_key_assignment_trigger(table: :test_table, sharding_key: :project_id, parent_table: :parent_table,
parent_table_primary_key: :project_id, parent_sharding_key: :parent_project_id, foreign_key: :foreign_key,
trigger_name: 'trigger_name')
end
end
describe '#remove_sharding_key_assignment_trigger' do
let(:trigger) { double }
let(:connection) { ActiveRecord::Base.connection }
it do
expect(Gitlab::Database::Triggers::AssignDesiredShardingKey).to receive(:new)
.with(table: :test_table, sharding_key: :project_id, parent_table: :parent_table, parent_table_primary_key: :project_id,
parent_sharding_key: :parent_project_id, foreign_key: :foreign_key, connection: connection,
trigger_name: 'trigger_name').and_return(trigger)
expect(trigger).to receive(:drop)
model.remove_sharding_key_assignment_trigger(table: :test_table, sharding_key: :project_id, parent_table: :parent_table,
parent_table_primary_key: :project_id, parent_sharding_key: :parent_project_id, foreign_key: :foreign_key,
trigger_name: 'trigger_name')
end
end
describe '#indexes_for' do
it 'returns the indexes for a column' do
idx1 = double(:idx, columns: %w[project_id])
idx2 = double(:idx, columns: %w[user_id])
allow(model).to receive(:indexes).with('table').and_return([idx1, idx2])
expect(model.indexes_for('table', :user_id)).to eq([idx2])
end
end
describe '#foreign_keys_for' do
it 'returns the foreign keys for a column' do
fk1 = double(:fk, column: 'project_id')
fk2 = double(:fk, column: 'user_id')
allow(model).to receive(:foreign_keys).with('table').and_return([fk1, fk2])
expect(model.foreign_keys_for('table', :user_id)).to eq([fk2])
end
end
describe '#copy_indexes' do
context 'when index name is too long' do
it 'does not fail' do
index = double(:index,
columns: %w[uuid],
name: 'index_vuln_findings_on_uuid_including_vuln_id_1',
using: nil,
where: nil,
opclasses: {},
unique: true,
lengths: [],
orders: [])
allow(model).to receive(:indexes_for).with(:vulnerability_occurrences, 'uuid')
.and_return([index])
expect(model).to receive(:add_concurrent_index)
.with(:vulnerability_occurrences,
%w[tmp_undo_cleanup_column_8cbf300838],
{
unique: true,
name: 'idx_copy_191a1af1a0',
length: [],
order: []
})
model.copy_indexes(:vulnerability_occurrences, :uuid, :tmp_undo_cleanup_column_8cbf300838)
end
end
context 'using a regular index using a single column' do
it 'copies the index' do
index = double(:index,
columns: %w[project_id],
name: 'index_on_issues_project_id',
using: nil,
where: nil,
opclasses: {},
unique: false,
lengths: [],
orders: [])
allow(model).to receive(:indexes_for).with(:issues, 'project_id')
.and_return([index])
expect(model).to receive(:add_concurrent_index)
.with(:issues,
%w[gl_project_id],
{
unique: false,
name: 'index_on_issues_gl_project_id',
length: [],
order: []
})
model.copy_indexes(:issues, :project_id, :gl_project_id)
end
end
context 'using a regular index with multiple columns' do
it 'copies the index' do
index = double(:index,
columns: %w[project_id foobar],
name: 'index_on_issues_project_id_foobar',
using: nil,
where: nil,
opclasses: {},
unique: false,
lengths: [],
orders: [])
allow(model).to receive(:indexes_for).with(:issues, 'project_id')
.and_return([index])
expect(model).to receive(:add_concurrent_index)
.with(:issues,
%w[gl_project_id foobar],
{
unique: false,
name: 'index_on_issues_gl_project_id_foobar',
length: [],
order: []
})
model.copy_indexes(:issues, :project_id, :gl_project_id)
end
end
context 'using an index with a WHERE clause' do
it 'copies the index' do
index = double(:index,
columns: %w[project_id],
name: 'index_on_issues_project_id',
using: nil,
where: 'foo',
opclasses: {},
unique: false,
lengths: [],
orders: [])
allow(model).to receive(:indexes_for).with(:issues, 'project_id')
.and_return([index])
expect(model).to receive(:add_concurrent_index)
.with(:issues,
%w[gl_project_id],
{
unique: false,
name: 'index_on_issues_gl_project_id',
length: [],
order: [],
where: 'foo'
})
model.copy_indexes(:issues, :project_id, :gl_project_id)
end
end
context 'using an index with a USING clause' do
it 'copies the index' do
index = double(:index,
columns: %w[project_id],
name: 'index_on_issues_project_id',
where: nil,
using: 'foo',
opclasses: {},
unique: false,
lengths: [],
orders: [])
allow(model).to receive(:indexes_for).with(:issues, 'project_id')
.and_return([index])
expect(model).to receive(:add_concurrent_index)
.with(:issues,
%w[gl_project_id],
{
unique: false,
name: 'index_on_issues_gl_project_id',
length: [],
order: [],
using: 'foo'
})
model.copy_indexes(:issues, :project_id, :gl_project_id)
end
end
context 'using an index with custom operator classes' do
it 'copies the index' do
index = double(:index,
columns: %w[project_id],
name: 'index_on_issues_project_id',
using: nil,
where: nil,
opclasses: { 'project_id' => 'bar' },
unique: false,
lengths: [],
orders: [])
allow(model).to receive(:indexes_for).with(:issues, 'project_id')
.and_return([index])
expect(model).to receive(:add_concurrent_index)
.with(:issues,
%w[gl_project_id],
{
unique: false,
name: 'index_on_issues_gl_project_id',
length: [],
order: [],
opclass: { 'gl_project_id' => 'bar' }
})
model.copy_indexes(:issues, :project_id, :gl_project_id)
end
end
context 'using an index with multiple columns and custom operator classes' do
it 'copies the index' do
index = double(:index,
{
columns: %w[project_id foobar],
name: 'index_on_issues_project_id_foobar',
using: :gin,
where: nil,
opclasses: { 'project_id' => 'bar', 'foobar' => :gin_trgm_ops },
unique: false,
lengths: [],
orders: []
})
allow(model).to receive(:indexes_for).with(:issues, 'project_id')
.and_return([index])
expect(model).to receive(:add_concurrent_index)
.with(:issues,
%w[gl_project_id foobar],
{
unique: false,
name: 'index_on_issues_gl_project_id_foobar',
length: [],
order: [],
opclass: { 'gl_project_id' => 'bar', 'foobar' => :gin_trgm_ops },
using: :gin
})
model.copy_indexes(:issues, :project_id, :gl_project_id)
end
end
context 'using an index with multiple columns and a custom operator class on the non affected column' do
it 'copies the index' do
index = double(:index,
{
columns: %w[project_id foobar],
name: 'index_on_issues_project_id_foobar',
using: :gin,
where: nil,
opclasses: { 'foobar' => :gin_trgm_ops },
unique: false,
lengths: [],
orders: []
})
allow(model).to receive(:indexes_for).with(:issues, 'project_id')
.and_return([index])
expect(model).to receive(:add_concurrent_index)
.with(:issues,
%w[gl_project_id foobar],
{
unique: false,
name: 'index_on_issues_gl_project_id_foobar',
length: [],
order: [],
opclass: { 'foobar' => :gin_trgm_ops },
using: :gin
})
model.copy_indexes(:issues, :project_id, :gl_project_id)
end
end
describe 'using an index of which the name does not contain the source column' do
it 'raises RuntimeError' do
index = double(:index,
columns: %w[project_id],
name: 'index_foobar_index',
using: nil,
where: nil,
opclasses: {},
unique: false,
lengths: [],
orders: [])
allow(model).to receive(:indexes_for).with(:issues, 'project_id')
.and_return([index])
expect { model.copy_indexes(:issues, :project_id, :gl_project_id) }
.to raise_error(RuntimeError)
end
end
end
describe '#copy_foreign_keys' do
it 'copies foreign keys from one column to another' do
fk = double(:fk,
from_table: 'issues',
to_table: 'projects',
on_delete: :cascade)
allow(model).to receive(:foreign_keys_for).with(:issues, :project_id)
.and_return([fk])
expect(model).to receive(:add_concurrent_foreign_key)
.with('issues', 'projects', column: :gl_project_id, on_delete: :cascade)
model.copy_foreign_keys(:issues, :project_id, :gl_project_id)
end
end
describe '#column_for' do
it 'returns a column object for an existing column' do
column = model.column_for(:users, :id)
expect(column.name).to eq('id')
end
it 'raises an error when a column does not exist' do
error_message = /Could not find column "kittens" on table "users"/
expect { model.column_for(:users, :kittens) }.to raise_error(error_message)
end
end
describe '#replace_sql' do
it 'builds the sql with correct functions' do
expect(model.replace_sql(Arel::Table.new(:users)[:first_name], "Alice", "Eve").to_s)
.to include('regexp_replace')
end
describe 'results' do
let!(:user) { create(:user, name: 'Kathy Alice Aliceson') }
it 'replaces the correct part of the string' do
allow(model).to receive(:transaction_open?).and_return(false)
query = model.replace_sql(Arel::Table.new(:users)[:name], 'Alice', 'Eve')
model.update_column_in_batches(:users, :name, query)
expect(user.reload.name).to eq('Kathy Eve Aliceson')
end
end
end
describe '#check_trigger_permissions!' do
it 'does nothing when the user has the correct permissions' do
expect { model.check_trigger_permissions!('users') }
.not_to raise_error
end
it 'raises RuntimeError when the user does not have the correct permissions' do
allow(Gitlab::Database::Grant).to receive(:create_and_execute_trigger?)
.with('kittens')
.and_return(false)
expect { model.check_trigger_permissions!('kittens') }
.to raise_error(RuntimeError, /Your database user is not allowed/)
end
end
describe '#convert_to_bigint_column' do
it 'returns the name of the temporary column used to convert to bigint' do
expect(model.convert_to_bigint_column(:id)).to eq('id_convert_to_bigint')
end
end
describe '#convert_to_type_column' do
it 'returns the name of the temporary column used to convert to bigint' do
expect(model.convert_to_type_column(:id, :int, :bigint)).to eq('id_convert_int_to_bigint')
end
it 'returns the name of the temporary column used to convert to uuid' do
expect(model.convert_to_type_column(:uuid, :string, :uuid)).to eq('uuid_convert_string_to_uuid')
end
end
describe '#index_exists_by_name?' do
it 'returns true if an index exists' do
ActiveRecord::Migration.connection.execute(
'CREATE INDEX test_index_for_index_exists ON projects (path);'
)
expect(model.index_exists_by_name?(:projects, 'test_index_for_index_exists'))
.to be_truthy
end
it 'returns false if the index does not exist' do
expect(model.index_exists_by_name?(:projects, 'this_does_not_exist'))
.to be_falsy
end
context 'when an index with a function exists' do
before do
ActiveRecord::Migration.connection.execute(
'CREATE INDEX test_index ON projects (LOWER(path));'
)
end
it 'returns true if an index exists' do
expect(model.index_exists_by_name?(:projects, 'test_index'))
.to be_truthy
end
end
context 'when an index exists for a table with the same name in another schema' do
before do
ActiveRecord::Migration.connection.execute(
'CREATE SCHEMA new_test_schema'
)
ActiveRecord::Migration.connection.execute(
'CREATE TABLE new_test_schema.projects (id integer, name character varying)'
)
ActiveRecord::Migration.connection.execute(
'CREATE INDEX test_index_on_name ON new_test_schema.projects (LOWER(name));'
)
end
it 'returns false if the index does not exist in the current schema' do
expect(model.index_exists_by_name?(:projects, 'test_index_on_name'))
.to be_falsy
end
end
end
describe '#create_or_update_plan_limit' do
before do
stub_const('Plan', Class.new(ActiveRecord::Base))
stub_const('PlanLimits', Class.new(ActiveRecord::Base))
Plan.class_eval do
self.table_name = 'plans'
end
PlanLimits.class_eval do
self.table_name = 'plan_limits'
end
end
it 'properly escapes names' do
expect(model).to receive(:execute).with <<~SQL
INSERT INTO plan_limits (plan_id, "project_hooks")
SELECT id, '10' FROM plans WHERE name = 'free' LIMIT 1
ON CONFLICT (plan_id) DO UPDATE SET "project_hooks" = EXCLUDED."project_hooks";
SQL
model.create_or_update_plan_limit('project_hooks', 'free', 10)
end
context 'when plan does not exist' do
it 'does not create any plan limits' do
expect { model.create_or_update_plan_limit('project_hooks', 'plan_name', 10) }
.not_to change { PlanLimits.count }
end
end
context 'when plan does exist' do
let!(:plan) { Plan.create!(name: 'plan_name') }
context 'when limit does not exist' do
it 'inserts a new plan limits' do
expect { model.create_or_update_plan_limit('project_hooks', 'plan_name', 10) }
.to change { PlanLimits.count }.by(1)
expect(PlanLimits.pluck(:project_hooks)).to contain_exactly(10)
end
end
context 'when limit does exist' do
let!(:plan_limit) { PlanLimits.create!(plan_id: plan.id) }
it 'updates an existing plan limits' do
expect { model.create_or_update_plan_limit('project_hooks', 'plan_name', 999) }
.not_to change { PlanLimits.count }
expect(plan_limit.reload.project_hooks).to eq(999)
end
end
end
end
describe '#backfill_iids' do
include MigrationsHelpers
let_it_be(:issue_base_type_enum) { 0 }
let_it_be(:issue_type) { table(:work_item_types).find_by(base_type: issue_base_type_enum) }
let(:issue_class) do
Class.new(ActiveRecord::Base) do
include AtomicInternalId
self.table_name = 'issues'
self.inheritance_column = :_type_disabled
belongs_to :project, class_name: "::Project", inverse_of: nil
has_internal_id :iid,
scope: :project,
init: ->(s, _scope) { s&.project&.issues&.maximum(:iid) },
presence: false
before_validation -> { self.work_item_type_id = ::WorkItems::Type.default_issue_type.id }
end
end
let_it_be(:organizations) { table(:organizations) }
let_it_be(:namespaces) { table(:namespaces) }
let_it_be(:projects) { table(:projects) }
let_it_be(:issues) { table(:issues) }
let_it_be(:organization) { organizations.create!(name: 'organization', path: 'organization') }
def setup
namespace = namespaces.create!(
name: 'foo',
path: 'foo',
type: Namespaces::UserNamespace.sti_name,
organization_id: organization.id
)
project_namespace = namespaces.create!(
name: 'project-foo',
path: 'project-foo',
type: 'Project',
organization_id: organization.id,
parent_id: namespace.id,
visibility_level: 20
)
projects.create!(
namespace_id: namespace.id,
project_namespace_id: project_namespace.id,
organization_id: organization.id
)
end
it 'generates iids properly for models created after the migration' do
project = setup
model.backfill_iids('issues')
issue = issue_class.create!(project_id: project.id, namespace_id: project.project_namespace_id)
expect(issue.iid).to eq(1)
end
it 'generates iids properly for models created after the migration when iids are backfilled' do
project = setup
issue_a = issues.create!(project_id: project.id, namespace_id: project.project_namespace_id, work_item_type_id: issue_type.id)
model.backfill_iids('issues')
issue_b = issue_class.create!(project_id: project.id, namespace_id: project.project_namespace_id)
expect(issue_a.reload.iid).to eq(1)
expect(issue_b.iid).to eq(2)
end
it 'generates iids properly for models created after the migration across multiple projects' do
project_a = setup
project_b = setup
issues.create!(project_id: project_a.id, namespace_id: project_a.project_namespace_id, work_item_type_id: issue_type.id)
issues.create!(project_id: project_b.id, namespace_id: project_b.project_namespace_id, work_item_type_id: issue_type.id)
issues.create!(project_id: project_b.id, namespace_id: project_b.project_namespace_id, work_item_type_id: issue_type.id)
model.backfill_iids('issues')
issue_a = issue_class.create!(project_id: project_a.id, namespace_id: project_a.project_namespace_id, work_item_type_id: issue_type.id)
issue_b = issue_class.create!(project_id: project_b.id, namespace_id: project_b.project_namespace_id, work_item_type_id: issue_type.id)
expect(issue_a.iid).to eq(2)
expect(issue_b.iid).to eq(3)
end
context 'when the first model is created for a project after the migration' do
it 'generates an iid' do
project_a = setup
project_b = setup
issue_a = issues.create!(project_id: project_a.id, namespace_id: project_a.project_namespace_id, work_item_type_id: issue_type.id)
model.backfill_iids('issues')
issue_b = issue_class.create!(project_id: project_b.id, namespace_id: project_b.project_namespace_id)
expect(issue_a.reload.iid).to eq(1)
expect(issue_b.reload.iid).to eq(1)
end
end
context 'when a row already has an iid set in the database' do
it 'backfills iids' do
project = setup
issue_a = issues.create!(project_id: project.id, namespace_id: project.project_namespace_id, work_item_type_id: issue_type.id, iid: 1)
issue_b = issues.create!(project_id: project.id, namespace_id: project.project_namespace_id, work_item_type_id: issue_type.id, iid: 2)
model.backfill_iids('issues')
expect(issue_a.reload.iid).to eq(1)
expect(issue_b.reload.iid).to eq(2)
end
it 'backfills for multiple projects' do
project_a = setup
project_b = setup
issue_a = issues.create!(project_id: project_a.id, namespace_id: project_a.project_namespace_id, work_item_type_id: issue_type.id, iid: 1)
issue_b = issues.create!(project_id: project_b.id, namespace_id: project_b.project_namespace_id, work_item_type_id: issue_type.id, iid: 1)
issue_c = issues.create!(project_id: project_a.id, namespace_id: project_a.project_namespace_id, work_item_type_id: issue_type.id, iid: 2)
model.backfill_iids('issues')
expect(issue_a.reload.iid).to eq(1)
expect(issue_b.reload.iid).to eq(1)
expect(issue_c.reload.iid).to eq(2)
end
end
end
describe '#add_primary_key_using_index' do
it "executes the statement to add the primary key" do
expect(model).to receive(:execute).with(/ALTER TABLE "_test_table" ADD CONSTRAINT "old_name" PRIMARY KEY USING INDEX "new_name"/)
model.add_primary_key_using_index(:_test_table, :old_name, :new_name)
end
end
context 'when changing the primary key of a given table' do
before do
model.create_table(:_test_table, primary_key: :id) do |t|
t.integer :partition_number, default: 1
end
model.add_index(:_test_table, :id, unique: true, name: :old_index_name)
model.add_index(:_test_table, [:id, :partition_number], unique: true, name: :new_index_name)
end
describe '#swap_primary_key' do
it 'executes statements to swap primary key', :aggregate_failures do
expect(model).to receive(:with_lock_retries).with(raise_on_exhaustion: true).ordered.and_yield
expect(model).to receive(:execute).with(/ALTER TABLE "_test_table" DROP CONSTRAINT "_test_table_pkey" CASCADE/).and_call_original
expect(model).to receive(:execute).with(/ALTER TABLE "_test_table" ADD CONSTRAINT "_test_table_pkey" PRIMARY KEY USING INDEX "new_index_name"/).and_call_original
model.swap_primary_key(:_test_table, :_test_table_pkey, :new_index_name)
end
context 'when new index does not exist' do
before do
model.remove_index(:_test_table, column: [:id, :partition_number])
end
it 'raises ActiveRecord::StatementInvalid' do
expect do
model.swap_primary_key(:_test_table, :_test_table_pkey, :new_index_name)
end.to raise_error(ActiveRecord::StatementInvalid)
end
end
end
describe '#unswap_primary_key' do
it 'executes statements to unswap primary key' do
expect(model).to receive(:with_lock_retries).with(raise_on_exhaustion: true).ordered.and_yield
expect(model).to receive(:execute).with(/ALTER TABLE "_test_table" DROP CONSTRAINT "_test_table_pkey" CASCADE/).ordered.and_call_original
expect(model).to receive(:execute).with(/ALTER TABLE "_test_table" ADD CONSTRAINT "_test_table_pkey" PRIMARY KEY USING INDEX "old_index_name"/).ordered.and_call_original
model.unswap_primary_key(:_test_table, :_test_table_pkey, :old_index_name)
end
end
end
describe '#drop_sequence' do
it "executes the statement to drop the sequence" do
expect(model).to receive(:execute).with(/ALTER TABLE "_test_table" ALTER COLUMN "test_column" DROP DEFAULT;\nDROP SEQUENCE IF EXISTS "_test_table_id_seq"/)
model.drop_sequence(:_test_table, :test_column, :_test_table_id_seq)
end
end
describe '#add_sequence' do
it "executes the statement to add the sequence" do
expect(model).to receive(:execute).with "CREATE SEQUENCE \"_test_table_id_seq\" START 1;\nALTER TABLE \"_test_table\" ALTER COLUMN \"test_column\" SET DEFAULT nextval(\'_test_table_id_seq\')\n"
model.add_sequence(:_test_table, :test_column, :_test_table_id_seq, 1)
end
end
describe '#remove_column_default' do
let(:test_table) { :_test_defaults_table }
let(:drop_default_statement) do
/ALTER TABLE "#{test_table}" ALTER COLUMN "#{column_name}" SET DEFAULT NULL/
end
subject(:recorder) do
ActiveRecord::QueryRecorder.new do
model.remove_column_default(test_table, column_name)
end
end
before do
model.create_table(test_table) do |t|
t.integer :int_with_default, default: 100
t.integer :int_with_default_function, default: -> { 'ceil(random () * 100)::int' }
t.integer :int_without_default
end
end
context 'with default values' do
let(:column_name) { :int_with_default }
it { expect(recorder.log).to include(drop_default_statement) }
end
context 'with default functions' do
let(:column_name) { :int_with_default_function }
it { expect(recorder.log).to include(drop_default_statement) }
end
context 'without any defaults' do
let(:column_name) { :int_without_default }
it { expect(recorder.log).to be_empty }
end
end
describe '#lock_tables' do
subject(:recorder) do
ActiveRecord::QueryRecorder.new { statement }
end
let(:statement) { model.lock_tables(:ci_builds, :ci_pipelines) }
it 'locks the tables' do
expect(recorder.log).to include(/LOCK TABLE "ci_builds", "ci_pipelines" IN ACCESS EXCLUSIVE MODE/)
end
context 'when only is provided' do
let(:statement) { model.lock_tables(:p_ci_builds, only: true) }
it 'locks the tables' do
expect(recorder.log).to include(/LOCK TABLE ONLY "p_ci_builds" IN ACCESS EXCLUSIVE MODE/)
end
end
context 'when nowait is provided' do
let(:statement) { model.lock_tables(:p_ci_builds, nowait: true) }
it 'locks the tables' do
expect(recorder.log).to include(/LOCK TABLE "p_ci_builds" IN ACCESS EXCLUSIVE MODE NOWAIT/)
end
end
end
describe '#column_is_nullable?' do
# This is defined as a private method of this module, and normally would not warrant
# dedicated test coverage. But that being said, it has no test coverage at all (it's
# only stubbed in the ConstraintsHelpers spec) so I'm adding testing here until we
# figure out how to test it properly through the public methods that use it.
context 'when a plain table name is passed' do
subject { model.send(:column_is_nullable?, 'table_name', 'column_name') }
it 'defaults to querying for the table defined in the current_schema' do
expect(model.connection).to receive(:select_value)
.with(/c\.table_schema = 'public'\s+AND c.table_name = 'table_name'\s+AND c.column_name = 'column_name'/)
subject
end
end
context 'when a table name is passed with a schema prefix' do
subject { model.send(:column_is_nullable?, 'schema_prefix.table_name', 'column_name') }
it 'correctly parses out the schema prefix and uses it instead of current_schema' do
expect(model.connection).to receive(:select_value)
.with(/c\.table_schema = 'schema_prefix'\s+AND c.table_name = 'table_name'\s+AND c.column_name = 'column_name'/)
subject
end
end
end
describe '#feature_flag_enabled?' do
let(:feature_flag_name) { 'test_feature_flag' }
context 'when feature flag is enabled' do
let(:result) { instance_double(PG::Result, ntuples: 1) }
it 'returns true' do
expected_sql = <<~SQL.squish
SELECT 1
FROM feature_gates
WHERE feature_key = '#{feature_flag_name}'
AND value = 'true'
LIMIT 1;
SQL
expect(model).to receive(:execute).with(expected_sql.strip).and_return(result)
expect(model.feature_flag_enabled?(feature_flag_name)).to be true
end
end
context 'when feature flag is disabled' do
let(:result) { instance_double(PG::Result, ntuples: 0) }
it 'returns false' do
expected_sql = <<~SQL.squish
SELECT 1
FROM feature_gates
WHERE feature_key = '#{feature_flag_name}'
AND value = 'true'
LIMIT 1;
SQL
expect(model).to receive(:execute).with(expected_sql.strip).and_return(result)
expect(model.feature_flag_enabled?(feature_flag_name)).to be false
end
end
end
describe 'bigint conversion helpers' do
include MigrationsHelpers
include Database::TriggerHelpers
let(:migration_class) do
Class.new(Gitlab::Database::Migration[2.2]) do
milestone '17.4'
restrict_gitlab_migration gitlab_schema: :gitlab_main
def type_from_path(_)
:regular
end
end
end
describe 'complete bigint conversion migration process' do
let(:context) { migration_class.new }
let(:table_name) { :_test_table }
let(:model) { table(table_name) }
let(:column_names) { [:id, :namespace_id, :traversal_ids, :parent_id] }
let(:create_table) do
migration_class.new.create_table table_name, id: false do |t|
t.integer :id, primary_key: true
t.integer :namespace_id, null: false
t.integer :traversal_ids, null: false, array: true, default: []
t.integer :parent_id
end
end
before do
allow_next_instance_of(migration_class) do |instance|
allow(instance).to receive_messages(
puts: nil,
transaction_open?: false
)
allow(instance.connection).to receive_messages(
puts: nil,
transaction_open?: false
)
end
# to prevent it from writing to the actual file
stub_const(
'Gitlab::Database::Migrations::Conversions::BigintConverter::YAML_FILE_PATH',
'tmp/integer_ids_not_yet_initialized_to_bigint.yml'
)
end
it 'correctly converts the integer columns to bigint' do
create_table
first_record = model.create!(namespace_id: 11, traversal_ids: [11], parent_id: nil)
second_record = model.create!(namespace_id: 22, traversal_ids: [22], parent_id: 111)
expect do
migration_class.new.initialize_conversion_of_integer_to_bigint(table_name, column_names)
end.to change { all_column_names }.to(
include(
*%w[id_convert_to_bigint namespace_id_convert_to_bigint traversal_ids_convert_to_bigint
parent_id_convert_to_bigint]
)
)
expect(context.column_for(table_name, :id_convert_to_bigint))
.to have_attributes(
sql_type: 'bigint',
null: false,
default: '0',
array: false
)
expect(context.column_for(table_name, :namespace_id_convert_to_bigint))
.to have_attributes(
sql_type: 'bigint',
null: false,
default: '0',
array: false
)
expect(context.column_for(table_name, :traversal_ids_convert_to_bigint))
.to have_attributes(
sql_type: 'bigint',
null: false,
default: '{}',
array: true
)
expect(context.column_for(table_name, :parent_id_convert_to_bigint))
.to have_attributes(
sql_type: 'bigint',
null: true,
default: nil,
array: false
)
model.reset_column_information
expect(first_record.reload)
.to have_attributes(
id_convert_to_bigint: 0,
namespace_id_convert_to_bigint: 0,
traversal_ids_convert_to_bigint: [],
parent_id_convert_to_bigint: nil
)
expect(second_record.reload)
.to have_attributes(
id_convert_to_bigint: 0,
namespace_id_convert_to_bigint: 0,
traversal_ids_convert_to_bigint: [],
parent_id_convert_to_bigint: nil
)
expect do
migration_class.new.backfill_conversion_of_integer_to_bigint(table_name, column_names)
end.to change { Gitlab::Database::BackgroundMigration::BatchedMigration.count }.by(1)
batched_migration = Gitlab::Database::BackgroundMigration::BatchedMigration.first
expect(batched_migration)
.to have_attributes(
batch_class_name: 'PrimaryKeyBatchingStrategy',
job_class_name: 'CopyColumnUsingBackgroundMigrationJob',
table_name: table_name.to_s,
column_name: 'id',
job_arguments: [
%w[
id namespace_id traversal_ids parent_id
], %w[
id_convert_to_bigint namespace_id_convert_to_bigint
traversal_ids_convert_to_bigint parent_id_convert_to_bigint
]
],
batch_size: 20_000,
sub_batch_size: 1000,
pause_ms: 100,
interval: 2.minutes
)
expect do
migration_class.new.ensure_backfill_conversion_of_integer_to_bigint_is_finished(table_name, column_names)
end.to change { batched_migration.reload.finalized? }.to(true)
expect(first_record.reload)
.to have_attributes(
id_convert_to_bigint: 1,
namespace_id_convert_to_bigint: 11,
traversal_ids_convert_to_bigint: [11],
parent_id_convert_to_bigint: nil
)
expect(second_record.reload)
.to have_attributes(
id_convert_to_bigint: 2,
namespace_id_convert_to_bigint: 22,
traversal_ids_convert_to_bigint: [22],
parent_id_convert_to_bigint: 111
)
# Swap columns
db_migration = migration_class.new.extend(Gitlab::Database::MigrationHelpers::Swapping)
db_migration.add_index(table_name, :id_convert_to_bigint, name: :idx, unique: true)
db_migration.swap_primary_key(table_name, :_test_table_pkey, :idx)
db_migration.swap_columns(table_name, :id, :id_convert_to_bigint)
db_migration.swap_columns_default(table_name, :id, :id_convert_to_bigint)
db_migration.swap_columns(table_name, :namespace_id, :namespace_id_convert_to_bigint)
db_migration.swap_columns_default(table_name, :namespace_id, :namespace_id_convert_to_bigint)
db_migration.swap_columns(table_name, :traversal_ids, :traversal_ids_convert_to_bigint)
db_migration.swap_columns_default(table_name, :traversal_ids, :traversal_ids_convert_to_bigint)
db_migration.swap_columns(table_name, :parent_id, :parent_id_convert_to_bigint)
db_migration.swap_columns_default(table_name, :parent_id, :parent_id_convert_to_bigint)
expect do
migration_class.new.cleanup_conversion_of_integer_to_bigint(table_name, column_names)
end.to change { all_column_names }.to(
not_include(
*%w[
id_convert_to_bigint namespace_id_convert_to_bigint
traversal_ids_convert_to_bigint parent_id_convert_to_bigint
]
)
)
expect(context.column_for(table_name, :id))
.to have_attributes(
sql_type: 'bigint',
null: false,
default: nil,
array: false
)
expect(context.column_for(table_name, :namespace_id))
.to have_attributes(
sql_type: 'bigint',
null: false,
default: nil,
array: false
)
expect(context.column_for(table_name, :traversal_ids))
.to have_attributes(
sql_type: 'bigint',
null: false,
default: '{}',
array: true
)
expect(context.column_for(table_name, :parent_id))
.to have_attributes(
sql_type: 'bigint',
null: true,
default: nil,
array: false
)
model.reset_column_information
expect(first_record.reload)
.to have_attributes(
id: 1,
namespace_id: 11,
traversal_ids: [11],
parent_id: nil
)
expect(second_record.reload)
.to have_attributes(
id: 2,
namespace_id: 22,
traversal_ids: [22],
parent_id: 111
)
expect do
migration_class.new.restore_conversion_of_integer_to_bigint(table_name, column_names)
end.to change { all_column_names }.to(
include(
*%w[
id_convert_to_bigint namespace_id_convert_to_bigint
traversal_ids_convert_to_bigint parent_id_convert_to_bigint
]
)
)
expect(context.column_for(table_name, :id_convert_to_bigint))
.to have_attributes(
sql_type: 'integer',
null: false,
default: '0',
array: false
)
expect(context.column_for(table_name, :namespace_id_convert_to_bigint))
.to have_attributes(
sql_type: 'integer',
null: false,
default: '0',
array: false
)
expect(context.column_for(table_name, :traversal_ids_convert_to_bigint))
.to have_attributes(
sql_type: 'integer',
null: false,
default: '{}',
array: true
)
expect(context.column_for(table_name, :parent_id_convert_to_bigint))
.to have_attributes(
sql_type: 'integer',
null: true,
default: nil,
array: false
)
end
context 'when all of them are bigint' do
let(:create_table) do
migration_class.new.create_table table_name, id: false do |t|
t.bigint :id, primary_key: true
t.bigint :namespace_id, null: false
t.bigint :traversal_ids, null: false, array: true, default: []
t.bigint :parent_id
end
end
it 'executes the migrations with doing nothing' do
create_table
model.create!(namespace_id: 11, traversal_ids: [11], parent_id: nil)
model.create!(namespace_id: 22, traversal_ids: [22], parent_id: 111)
expect do
migration_class.new.initialize_conversion_of_integer_to_bigint(table_name, column_names)
end.not_to change { all_column_names }
expect do
migration_class.new.backfill_conversion_of_integer_to_bigint(table_name, column_names)
end.to change { Gitlab::Database::BackgroundMigration::BatchedMigration.count }.by(1)
expect do
migration_class.new.ensure_backfill_conversion_of_integer_to_bigint_is_finished(table_name, column_names)
end.to change { Gitlab::Database::BackgroundMigration::BatchedMigration.first.human_status_name }.to('finalized')
expect do
migration_class.new.revert_backfill_conversion_of_integer_to_bigint(table_name, column_names)
end.to change { Gitlab::Database::BackgroundMigration::BatchedMigration.count }.by(-1)
expect do
migration_class.new.cleanup_conversion_of_integer_to_bigint(table_name, column_names)
end.not_to change { all_column_names }
end
end
context 'when there are both integer and bigint columns' do
let(:create_table) do
migration_class.new.create_table table_name, id: false do |t|
t.bigint :id, primary_key: true
t.integer :namespace_id, null: false
t.integer :traversal_ids, null: false, array: true, default: []
t.bigint :parent_id
end
end
it 'correctly converts the integer columns to bigint' do
create_table
first_record = model.create!(namespace_id: 11, traversal_ids: [11], parent_id: nil)
second_record = model.create!(namespace_id: 22, traversal_ids: [22], parent_id: 111)
expect do
migration_class.new.initialize_conversion_of_integer_to_bigint(table_name, column_names)
end.to change { all_column_names }.to(
include(
*%w[namespace_id_convert_to_bigint traversal_ids_convert_to_bigint]
)
)
expect(context.column_for(table_name, :namespace_id_convert_to_bigint))
.to have_attributes(
sql_type: 'bigint',
null: false,
default: '0',
array: false
)
expect(context.column_for(table_name, :traversal_ids_convert_to_bigint))
.to have_attributes(
sql_type: 'bigint',
null: false,
default: '{}',
array: true
)
model.reset_column_information
expect(first_record.reload)
.to have_attributes(
namespace_id_convert_to_bigint: 0,
traversal_ids_convert_to_bigint: []
)
expect(second_record.reload)
.to have_attributes(
namespace_id_convert_to_bigint: 0,
traversal_ids_convert_to_bigint: []
)
expect do
migration_class.new.backfill_conversion_of_integer_to_bigint(table_name, column_names)
end.to change { Gitlab::Database::BackgroundMigration::BatchedMigration.count }.by(1)
batched_migration = Gitlab::Database::BackgroundMigration::BatchedMigration.first
expect(batched_migration)
.to have_attributes(
batch_class_name: 'PrimaryKeyBatchingStrategy',
job_class_name: 'CopyColumnUsingBackgroundMigrationJob',
table_name: table_name.to_s,
column_name: 'id',
job_arguments: [
%w[
id namespace_id traversal_ids parent_id
], %w[
id_convert_to_bigint namespace_id_convert_to_bigint
traversal_ids_convert_to_bigint parent_id_convert_to_bigint
]
],
batch_size: 20_000,
sub_batch_size: 1000,
pause_ms: 100,
interval: 2.minutes
)
expect do
migration_class.new.ensure_backfill_conversion_of_integer_to_bigint_is_finished(table_name, column_names)
end.to change { batched_migration.reload.finalized? }.to(true)
expect(first_record.reload)
.to have_attributes(
namespace_id_convert_to_bigint: 11,
traversal_ids_convert_to_bigint: [11]
)
expect(second_record.reload)
.to have_attributes(
namespace_id_convert_to_bigint: 22,
traversal_ids_convert_to_bigint: [22]
)
# Swap columns
db_migration = migration_class.new.extend(Gitlab::Database::MigrationHelpers::Swapping)
db_migration.swap_columns(table_name, :namespace_id, :namespace_id_convert_to_bigint)
db_migration.swap_columns_default(table_name, :namespace_id, :namespace_id_convert_to_bigint)
db_migration.swap_columns(table_name, :traversal_ids, :traversal_ids_convert_to_bigint)
db_migration.swap_columns_default(table_name, :traversal_ids, :traversal_ids_convert_to_bigint)
expect do
migration_class.new.cleanup_conversion_of_integer_to_bigint(table_name, column_names)
end.to change { all_column_names }.to(
not_include(
*%w[namespace_id_convert_to_bigint traversal_ids_convert_to_bigint]
)
)
expect(context.column_for(table_name, :id))
.to have_attributes(
sql_type: 'bigint',
null: false,
default: nil,
array: false
)
expect(context.column_for(table_name, :namespace_id))
.to have_attributes(
sql_type: 'bigint',
null: false,
default: nil,
array: false
)
expect(context.column_for(table_name, :traversal_ids))
.to have_attributes(
sql_type: 'bigint',
null: false,
default: '{}',
array: true
)
expect(context.column_for(table_name, :parent_id))
.to have_attributes(
sql_type: 'bigint',
null: true,
default: nil,
array: false
)
model.reset_column_information
expect(first_record.reload)
.to have_attributes(
namespace_id: 11,
traversal_ids: [11]
)
expect(second_record.reload)
.to have_attributes(
namespace_id: 22,
traversal_ids: [22]
)
end
end
context <<~DESCRIPTION do
when the initialization and backfill have started
for columns which are integer on gitlab.com but bigint on self-managed instance
before this change is introduce
DESCRIPTION
let(:create_table) do
migration_class.new.create_table table_name, id: false do |t|
t.bigint :id, primary_key: true
t.integer :namespace_id, null: false
t.integer :traversal_ids, null: false, array: true, default: []
t.integer :parent_id
t.integer :duration
end
end
let(:column_names) { [:namespace_id, :traversal_ids, :parent_id, :duration] }
it 'correctly converts the integer columns to bigint' do
create_table
first_record = model.create!(namespace_id: 11, traversal_ids: [11], parent_id: nil, duration: 888)
second_record = model.create!(namespace_id: 22, traversal_ids: [22], parent_id: 111, duration: 999)
expect do
migration_class.new.initialize_conversion_of_integer_to_bigint(table_name, column_names)
end.to change { all_column_names }.to(
include(
*%w[
namespace_id_convert_to_bigint traversal_ids_convert_to_bigint
parent_id_convert_to_bigint duration_convert_to_bigint
]
)
)
expect(context.column_for(table_name, :namespace_id_convert_to_bigint))
.to have_attributes(
sql_type: 'bigint',
null: false,
default: '0',
array: false
)
expect(context.column_for(table_name, :traversal_ids_convert_to_bigint))
.to have_attributes(
sql_type: 'bigint',
null: false,
default: '{}',
array: true
)
expect(context.column_for(table_name, :parent_id_convert_to_bigint))
.to have_attributes(
sql_type: 'bigint',
null: true,
default: nil,
array: false
)
expect(context.column_for(table_name, :duration_convert_to_bigint))
.to have_attributes(
sql_type: 'bigint',
null: true,
default: nil,
array: false
)
model.reset_column_information
expect(first_record.reload)
.to have_attributes(
namespace_id_convert_to_bigint: 0,
traversal_ids_convert_to_bigint: [],
parent_id_convert_to_bigint: nil,
duration_convert_to_bigint: nil
)
expect(second_record.reload)
.to have_attributes(
namespace_id_convert_to_bigint: 0,
traversal_ids_convert_to_bigint: [],
parent_id_convert_to_bigint: nil,
duration_convert_to_bigint: nil
)
expect do
migration_class.new.backfill_conversion_of_integer_to_bigint(table_name, column_names)
end.to change { Gitlab::Database::BackgroundMigration::BatchedMigration.count }.by(1)
batched_migration = Gitlab::Database::BackgroundMigration::BatchedMigration.first
expect(batched_migration)
.to have_attributes(
batch_class_name: 'PrimaryKeyBatchingStrategy',
job_class_name: 'CopyColumnUsingBackgroundMigrationJob',
table_name: table_name.to_s,
column_name: 'id',
job_arguments: [
%w[
namespace_id traversal_ids parent_id duration
], %w[
namespace_id_convert_to_bigint traversal_ids_convert_to_bigint
parent_id_convert_to_bigint duration_convert_to_bigint
]
],
batch_size: 20_000,
sub_batch_size: 1000,
pause_ms: 100,
interval: 2.minutes
)
db_migration = migration_class.new.extend(Gitlab::Database::MigrationHelpers::Swapping)
# Change the integer columns to bigint for self-managed
db_migration.change_column(table_name, :namespace_id, :bigint, null: false)
db_migration.change_column(table_name, :traversal_ids, :bigint, null: false, array: true, default: [])
db_migration.change_column(table_name, :parent_id, :bigint)
expect do
migration_class.new.ensure_backfill_conversion_of_integer_to_bigint_is_finished(table_name, column_names)
end.to change { batched_migration.reload.finalized? }.to(true)
expect(first_record.reload)
.to have_attributes(
namespace_id_convert_to_bigint: 11,
traversal_ids_convert_to_bigint: [11],
parent_id_convert_to_bigint: nil,
duration_convert_to_bigint: 888
)
expect(second_record.reload)
.to have_attributes(
namespace_id_convert_to_bigint: 22,
traversal_ids_convert_to_bigint: [22],
parent_id_convert_to_bigint: 111,
duration_convert_to_bigint: 999
)
# Swap columns
db_migration.swap_columns(table_name, :duration, :duration_convert_to_bigint)
db_migration.swap_columns_default(table_name, :duration, :duration_convert_to_bigint)
expect do
migration_class.new.cleanup_conversion_of_integer_to_bigint(table_name, column_names)
end.to change { all_column_names }.to(
not_include(
*%w[
namespace_id_convert_to_bigint traversal_ids_convert_to_bigint
parent_id_convert_to_bigint duration_convert_to_bigint
]
)
)
expect(context.column_for(table_name, :id))
.to have_attributes(
sql_type: 'bigint',
null: false,
default: nil,
array: false
)
expect(context.column_for(table_name, :namespace_id))
.to have_attributes(
sql_type: 'bigint',
null: false,
default: nil,
array: false
)
expect(context.column_for(table_name, :traversal_ids))
.to have_attributes(
sql_type: 'bigint',
null: false,
default: '{}',
array: true
)
expect(context.column_for(table_name, :parent_id))
.to have_attributes(
sql_type: 'bigint',
null: true,
default: nil,
array: false
)
model.reset_column_information
expect(first_record.reload)
.to have_attributes(
namespace_id: 11,
traversal_ids: [11],
parent_id: nil,
duration: 888
)
expect(second_record.reload)
.to have_attributes(
namespace_id: 22,
traversal_ids: [22],
parent_id: 111,
duration: 999
)
end
end
end
describe '#initialize_conversion_of_integer_to_bigint' do
it 'calls the converter' do
expect_next_instance_of(Gitlab::Database::Migrations::Conversions::BigintConverter) do |instance|
expect(instance).to receive(:init)
end
migration_class.new.initialize_conversion_of_integer_to_bigint('a_table', %w[column1 column2])
end
end
describe '#restore_conversion_of_integer_to_bigint' do
it 'calls the converter' do
expect_next_instance_of(Gitlab::Database::Migrations::Conversions::BigintConverter) do |instance|
expect(instance).to receive(:restore_cleanup)
end
migration_class.new.restore_conversion_of_integer_to_bigint('a_table', %w[column1 column2])
end
end
describe '#revert_initialize_conversion_of_integer_to_bigint' do
it 'calls the converter' do
expect_next_instance_of(Gitlab::Database::Migrations::Conversions::BigintConverter) do |instance|
expect(instance).to receive(:revert_init)
end
migration_class.new.revert_initialize_conversion_of_integer_to_bigint('a_table', %w[column1 column2])
end
end
describe '#cleanup_conversion_of_integer_to_bigint' do
it 'calls the converter' do
expect_next_instance_of(Gitlab::Database::Migrations::Conversions::BigintConverter) do |instance|
expect(instance).to receive(:cleanup)
end
migration_class.new.cleanup_conversion_of_integer_to_bigint('a_table', %w[column1 column2])
end
end
describe '#backfill_conversion_of_integer_to_bigint' do
it 'calls the converter' do
expect_next_instance_of(Gitlab::Database::Migrations::Conversions::BigintConverter) do |instance|
expect(instance).to receive(:backfill).with(
batch_size: 20000, job_interval: 2.minutes, pause_ms: 100, primary_key: :id, sub_batch_size: 1000
)
end
migration_class.new.backfill_conversion_of_integer_to_bigint('a_table', %w[column1 column2])
end
end
describe '#ensure_backfill_conversion_of_integer_to_bigint_is_finished' do
it 'calls the converter' do
expect_next_instance_of(Gitlab::Database::Migrations::Conversions::BigintConverter) do |instance|
expect(instance).to receive(:ensure_backfill).with(primary_key: :id)
end
migration_class.new.ensure_backfill_conversion_of_integer_to_bigint_is_finished('a_table', %w[column1 column2])
end
end
describe '#revert_backfill_conversion_of_integer_to_bigint' do
it 'calls the converter' do
expect_next_instance_of(Gitlab::Database::Migrations::Conversions::BigintConverter) do |instance|
expect(instance).to receive(:revert_backfill).with(primary_key: :id)
end
migration_class.new.revert_backfill_conversion_of_integer_to_bigint('a_table', %w[column1 column2])
end
end
private
def all_column_names
context.columns(table_name).map(&:name)
end
end
end