Backfill / update timelogs.spent_at where NULL

db/post_migrate/20211215090620_schedule_update_timelogs_null_spent_at.rb

+# frozen_string_literal: true
+
+class ScheduleUpdateTimelogsNullSpentAt < Gitlab::Database::Migration[1.0]
+  DOWNTIME = false
+  BATCH_SIZE = 5_000
+  DELAY_INTERVAL = 2.minutes
+  MIGRATION = 'UpdateTimelogsNullSpentAt'
+
+  disable_ddl_transaction!
+
+  def up
+    queue_background_migration_jobs_by_range_at_intervals(
+      define_batchable_model('timelogs').where(spent_at: nil),
+      MIGRATION,
+      DELAY_INTERVAL,
+      batch_size: BATCH_SIZE
+    )
+  end
+
+  def down
+    # no-op
+  end
+end

db/schema_migrations/20211215090620

+43ae6290e11e3944b23ce2865b5c466a29c9ba3cfd2e0b58bd834568414b5bf2
\ No newline at end of file

lib/gitlab/background_migration/update_timelogs_null_spent_at.rb

+# frozen_string_literal: true
+
+module Gitlab
+  module BackgroundMigration
+    # Class to populate spent_at for timelogs
+    class UpdateTimelogsNullSpentAt
+      include Gitlab::Database::DynamicModelHelpers
+
+      BATCH_SIZE = 100
+
+      def perform(start_id, stop_id)
+        define_batchable_model('timelogs').where(spent_at: nil, id: start_id..stop_id).each_batch(of: 100) do |subbatch|
+          batch_start, batch_end = subbatch.pluck('min(id), max(id)').first
+
+          update_timelogs(batch_start, batch_end)
+        end
+      end
+
+      def update_timelogs(batch_start, batch_stop)
+        execute(<<~SQL)
+          UPDATE timelogs
+          SET spent_at = created_at
+          WHERE spent_at IS NULL
+          AND timelogs.id BETWEEN #{batch_start} AND #{batch_stop};
+        SQL
+      end
+
+      def execute(sql)
+        @connection ||= ::ActiveRecord::Base.connection
+        @connection.execute(sql)
+      end
+    end
+  end
+end

spec/lib/gitlab/background_migration/update_timelogs_null_spent_at_spec.rb

+# frozen_string_literal: true
+
+require 'spec_helper'
+
+RSpec.describe Gitlab::BackgroundMigration::UpdateTimelogsNullSpentAt, schema: 20211215090620 do
+  let_it_be(:previous_time) { 10.days.ago }
+  let_it_be(:namespace) { table(:namespaces).create!(name: 'namespace', path: 'namespace') }
+  let_it_be(:project) { table(:projects).create!(namespace_id: namespace.id) }
+  let_it_be(:issue) { table(:issues).create!(project_id: project.id) }
+  let_it_be(:merge_request) { table(:merge_requests).create!(target_project_id: project.id, source_branch: 'master', target_branch: 'feature') }
+  let_it_be(:timelog1) { create_timelog!(issue_id: issue.id) }
+  let_it_be(:timelog2) { create_timelog!(merge_request_id: merge_request.id) }
+  let_it_be(:timelog3) { create_timelog!(issue_id: issue.id, spent_at: previous_time) }
+  let_it_be(:timelog4) { create_timelog!(merge_request_id: merge_request.id, spent_at: previous_time) }
+
+  subject(:background_migration) { described_class.new }
+
+  before_all do
+    table(:timelogs).where.not(id: [timelog3.id, timelog4.id]).update_all(spent_at: nil)
+  end
+
+  describe '#perform' do
+    it 'sets correct spent_at' do
+      background_migration.perform(timelog1.id, timelog4.id)
+
+      expect(timelog1.reload.spent_at).to be_like_time(timelog1.created_at)
+      expect(timelog2.reload.spent_at).to be_like_time(timelog2.created_at)
+      expect(timelog3.reload.spent_at).to be_like_time(previous_time)
+      expect(timelog4.reload.spent_at).to be_like_time(previous_time)
+      expect(timelog3.reload.spent_at).not_to be_like_time(timelog3.created_at)
+      expect(timelog4.reload.spent_at).not_to be_like_time(timelog4.created_at)
+    end
+  end
+
+  private
+
+  def create_timelog!(**args)
+    table(:timelogs).create!(**args, time_spent: 1)
+  end
+end

spec/migrations/schedule_update_timelogs_null_spent_at_spec.rb

+# frozen_string_literal: true
+
+require 'spec_helper'
+require_migration!
+
+RSpec.describe ScheduleUpdateTimelogsNullSpentAt do
+  let_it_be(:namespace) { table(:namespaces).create!(name: 'namespace', path: 'namespace') }
+  let_it_be(:project) { table(:projects).create!(namespace_id: namespace.id) }
+  let_it_be(:issue) { table(:issues).create!(project_id: project.id) }
+  let_it_be(:merge_request) { table(:merge_requests).create!(target_project_id: project.id, source_branch: 'master', target_branch: 'feature') }
+  let_it_be(:timelog1) { create_timelog!(merge_request_id: merge_request.id) }
+  let_it_be(:timelog2) { create_timelog!(merge_request_id: merge_request.id) }
+  let_it_be(:timelog3) { create_timelog!(merge_request_id: merge_request.id) }
+  let_it_be(:timelog4) { create_timelog!(issue_id: issue.id) }
+  let_it_be(:timelog5) { create_timelog!(issue_id: issue.id) }
+
+  before_all do
+    table(:timelogs).where.not(id: timelog3.id).update_all(spent_at: nil)
+  end
+
+  it 'correctly schedules background migrations' do
+    stub_const("#{described_class}::BATCH_SIZE", 2)
+
+    Sidekiq::Testing.fake! do
+      freeze_time do
+        migrate!
+
+        expect(described_class::MIGRATION)
+          .to be_scheduled_delayed_migration(2.minutes, timelog1.id, timelog2.id)
+
+        expect(described_class::MIGRATION)
+          .to be_scheduled_delayed_migration(4.minutes, timelog4.id, timelog5.id)
+
+        expect(BackgroundMigrationWorker.jobs.size).to eq(2)
+      end
+    end
+  end
+
+  private
+
+  def create_timelog!(**args)
+    table(:timelogs).create!(**args, time_spent: 1)
+  end
+end