Merge branch '341069-backfill-namespace-db-entries-with-correct-usernamespace-type' into 'master'

Backfill namespace DB entries with correct UserNamespace type

See merge request gitlab-org/gitlab!71510

db/post_migrate/20210930211936_backfill_user_namespace.rb

+# frozen_string_literal: true
+
+class BackfillUserNamespace < Gitlab::Database::Migration[1.0]
+  MIGRATION = 'BackfillUserNamespace'
+  INTERVAL = 2.minutes
+  BATCH_SIZE = 1_000
+  SUB_BATCH_SIZE = 200
+  DOWNTIME = false
+
+  def up
+    queue_batched_background_migration(
+      MIGRATION,
+      :namespaces,
+      :id,
+      job_interval: INTERVAL,
+      batch_size: BATCH_SIZE,
+      sub_batch_size: SUB_BATCH_SIZE
+    )
+  end
+
+  def down
+    Gitlab::Database::BackgroundMigration::BatchedMigration
+      .for_configuration(MIGRATION, :namespaces, :id, [])
+      .delete_all
+  end
+end

db/schema_migrations/20210930211936

+3aaf2a4fa834331768e2acc10f67b8d456e70aca9784787e40b55dade7b6f64c
\ No newline at end of file

lib/gitlab/background_migration/backfill_user_namespace.rb

+# frozen_string_literal: true
+
+module Gitlab
+  module BackgroundMigration
+    # Backfills the `namespaces.type` column, replacing any
+    # instances of `NULL` with `User`
+    class BackfillUserNamespace
+      include Gitlab::Database::DynamicModelHelpers
+
+      def perform(start_id, end_id, batch_table, batch_column, sub_batch_size, pause_ms)
+        parent_batch_relation = relation_scoped_to_range(batch_table, batch_column, start_id, end_id)
+        parent_batch_relation.each_batch(column: batch_column, of: sub_batch_size, order_hint: :type) do |sub_batch|
+          batch_metrics.time_operation(:update_all) do
+            sub_batch.update_all(type: 'User')
+          end
+          pause_ms = 0 if pause_ms < 0
+          sleep(pause_ms * 0.001)
+        end
+      end
+
+      def batch_metrics
+        @batch_metrics ||= Gitlab::Database::BackgroundMigration::BatchMetrics.new
+      end
+
+      private
+
+      def connection
+        ActiveRecord::Base.connection
+      end
+
+      def relation_scoped_to_range(source_table, source_key_column, start_id, stop_id)
+        define_batchable_model(source_table)
+          .where(source_key_column => start_id..stop_id)
+          .where(type: nil)
+      end
+    end
+  end
+end

spec/lib/gitlab/background_migration/backfill_user_namespace_spec.rb

+# frozen_string_literal: true
+
+require 'spec_helper'
+
+RSpec.describe Gitlab::BackgroundMigration::BackfillUserNamespace, :migration, schema: 20210930211936 do
+  let(:migration) { described_class.new }
+  let(:namespaces_table) { table(:namespaces) }
+
+  let(:table_name) { 'namespaces' }
+  let(:batch_column) { :id }
+  let(:sub_batch_size) { 100 }
+  let(:pause_ms) { 0 }
+
+  subject(:perform_migration) { migration.perform(1, 10, table_name, batch_column, sub_batch_size, pause_ms) }
+
+  before do
+    namespaces_table.create!(id: 1, name: 'test1', path: 'test1', type: nil)
+    namespaces_table.create!(id: 2, name: 'test2', path: 'test2', type: 'User')
+    namespaces_table.create!(id: 3, name: 'test3', path: 'test3', type: 'Group')
+    namespaces_table.create!(id: 4, name: 'test4', path: 'test4', type: nil)
+    namespaces_table.create!(id: 11, name: 'test11', path: 'test11', type: nil)
+  end
+
+  it 'backfills `type` for the selected records', :aggregate_failures do
+    queries = ActiveRecord::QueryRecorder.new do
+      perform_migration
+    end
+
+    expect(queries.count).to eq(3)
+    expect(namespaces_table.where(type: 'User').count).to eq 3
+    expect(namespaces_table.where(type: 'User').pluck(:id)).to match_array([1, 2, 4])
+  end
+
+  it 'tracks timings of queries' do
+    expect(migration.batch_metrics.timings).to be_empty
+
+    expect { perform_migration }.to change { migration.batch_metrics.timings }
+  end
+end

spec/migrations/backfill_user_namespace_spec.rb

+# frozen_string_literal: true
+
+require 'spec_helper'
+require_migration!
+
+RSpec.describe BackfillUserNamespace do
+  let_it_be(:migration) { described_class::MIGRATION }
+
+  describe '#up' do
+    it 'schedules background jobs for each batch of namespaces' do
+      migrate!
+
+      expect(migration).to have_scheduled_batched_migration(
+        table_name: :namespaces,
+        column_name: :id,
+        interval: described_class::INTERVAL
+      )
+    end
+  end
+
+  describe '#down' do
+    it 'deletes all batched migration records' do
+      migrate!
+      schema_migrate_down!
+
+      expect(migration).not_to have_scheduled_batched_migration
+    end
+  end
+end