Add cron worker for cleaning up unit test tables

Adds a scheduled worker that runs once daily to delete
unit test failures that are older than 14 days.

Also deletes unit test records that don't have any
associated unit test failures after cleaning up.

Changelog: added

app/models/ci/unit_test.rb

@@ -14,6 +14,7 @@ module Ci
     belongs_to :project
 
     scope :by_project_and_keys, -> (project, keys) { where(project_id: project.id, key_hash: keys) }
+    scope :deletable, -> { where('NOT EXISTS (?)', Ci::UnitTestFailure.select(1).where("#{Ci::UnitTestFailure.table_name}.unit_test_id = #{table_name}.id")) }
 
     class << self
       def find_or_create_by_batch(project, unit_test_attrs)

app/models/ci/unit_test_failure.rb

@@ -11,6 +11,8 @@ module Ci
     belongs_to :unit_test, class_name: "Ci::UnitTest", foreign_key: :unit_test_id
     belongs_to :build, class_name: "Ci::Build", foreign_key: :build_id
 
+    scope :deletable, -> { where('failed_at < ?', REPORT_WINDOW.ago) }
+
     def self.recent_failures_count(project:, unit_test_keys:, date_range: REPORT_WINDOW.ago..Time.current)
       joins(:unit_test)
         .where(

app/services/ci/delete_unit_tests_service.rb

+# frozen_string_literal: true
+
+module Ci
+  class DeleteUnitTestsService
+    include EachBatch
+
+    BATCH_SIZE = 100
+
+    def execute
+      purge_data!(Ci::UnitTestFailure)
+      purge_data!(Ci::UnitTest)
+    end
+
+    private
+
+    def purge_data!(klass)
+      loop do
+        break unless delete_batch!(klass)
+      end
+    end
+
+    # rubocop: disable CodeReuse/ActiveRecord
+    def delete_batch!(klass)
+      deleted = 0
+
+      ActiveRecord::Base.transaction do
+        ids = klass.deletable.lock('FOR UPDATE SKIP LOCKED').limit(BATCH_SIZE).pluck(:id)
+        break if ids.empty?
+
+        deleted = klass.where(id: ids).delete_all
+      end
+
+      deleted > 0
+    end
+    # rubocop: enable CodeReuse/ActiveRecord
+  end
+end

app/workers/all_queues.yml

@@ -182,6 +182,15 @@
   :weight: 1
   :idempotent:
   :tags: []
+- :name: cronjob:ci_delete_unit_tests
+  :worker_name: Ci::DeleteUnitTestsWorker
+  :feature_category: :continuous_integration
+  :has_external_dependencies:
+  :urgency: :low
+  :resource_boundary: :unknown
+  :weight: 1
+  :idempotent: true
+  :tags: []
 - :name: cronjob:ci_pipeline_artifacts_expire_artifacts
   :worker_name: Ci::PipelineArtifacts::ExpireArtifactsWorker
   :feature_category: :continuous_integration

app/workers/ci/delete_unit_tests_worker.rb

+# frozen_string_literal: true
+
+module Ci
+  class DeleteUnitTestsWorker
+    include ApplicationWorker
+    # rubocop:disable Scalability/CronWorkerContext
+    # This worker does not perform work scoped to a context
+    include CronjobQueue
+    # rubocop:enable Scalability/CronWorkerContext
+
+    feature_category :continuous_integration
+    idempotent!
+
+    def perform
+      Ci::DeleteUnitTestsService.new.execute
+    end
+  end
+end

changelogs/unreleased/eb-clean-up-unit-test-tables.yml

+---
+title: Add cron worker for cleaning up unit test tables
+merge_request: 61463
+author:
+type: added

config/initializers/1_settings.rb

@@ -572,6 +572,9 @@ Settings.cron_jobs['ssh_keys_expiring_soon_notification_worker']['job_class'] =
 Settings.cron_jobs['users_deactivate_dormant_users_worker'] ||= Settingslogic.new({})
 Settings.cron_jobs['users_deactivate_dormant_users_worker']['cron'] ||= '21,42 0-4 * * *'
 Settings.cron_jobs['users_deactivate_dormant_users_worker']['job_class'] = 'Users::DeactivateDormantUsersWorker'
+Settings.cron_jobs['ci_delete_unit_tests_worker'] ||= Settingslogic.new({})
+Settings.cron_jobs['ci_delete_unit_tests_worker']['cron'] ||= '0 0 * * *'
+Settings.cron_jobs['ci_delete_unit_tests_worker']['job_class'] = 'Ci::DeleteUnitTestsWorker'
 
 Gitlab.com do
   Settings.cron_jobs['batched_background_migrations_worker'] ||= Settingslogic.new({})

spec/services/ci/delete_unit_tests_service_spec.rb

+# frozen_string_literal: true
+
+require 'spec_helper'
+
+RSpec.describe Ci::DeleteUnitTestsService do
+  describe '#execute' do
+    let!(:unit_test_1) { create(:ci_unit_test) }
+    let!(:unit_test_2) { create(:ci_unit_test) }
+    let!(:unit_test_3) { create(:ci_unit_test) }
+    let!(:unit_test_4) { create(:ci_unit_test) }
+    let!(:unit_test_1_recent_failure) { create(:ci_unit_test_failure, unit_test: unit_test_1) }
+    let!(:unit_test_1_old_failure) { create(:ci_unit_test_failure, unit_test: unit_test_1, failed_at: 15.days.ago) }
+    let!(:unit_test_2_old_failure) { create(:ci_unit_test_failure, unit_test: unit_test_2, failed_at: 15.days.ago) }
+    let!(:unit_test_3_old_failure) { create(:ci_unit_test_failure, unit_test: unit_test_3, failed_at: 15.days.ago) }
+    let!(:unit_test_4_old_failure) { create(:ci_unit_test_failure, unit_test: unit_test_4, failed_at: 15.days.ago) }
+
+    before do
+      stub_const("#{described_class.name}::BATCH_SIZE", 2)
+
+      described_class.new.execute
+    end
+
+    it 'does not delete unit test failures not older than 14 days' do
+      expect(unit_test_1_recent_failure.reload).to be_persisted
+    end
+
+    it 'deletes unit test failures older than 14 days' do
+      ids = [
+        unit_test_1_old_failure,
+        unit_test_2_old_failure,
+        unit_test_3_old_failure,
+        unit_test_4_old_failure
+      ].map(&:id)
+
+      result = Ci::UnitTestFailure.where(id: ids)
+
+      expect(result).to be_empty
+    end
+
+    it 'deletes unit tests that have no more associated unit test failures' do
+      ids = [
+        unit_test_2,
+        unit_test_3,
+        unit_test_4
+      ].map(&:id)
+
+      result = Ci::UnitTest.where(id: ids)
+
+      expect(result).to be_empty
+    end
+  end
+end

spec/workers/ci/delete_unit_tests_worker_spec.rb

+# frozen_string_literal: true
+
+require 'spec_helper'
+
+RSpec.describe Ci::DeleteUnitTestsWorker do
+  let(:worker) { described_class.new }
+
+  describe '#perform' do
+    it 'executes a service' do
+      expect_next_instance_of(Ci::DeleteUnitTestsService) do |instance|
+        expect(instance).to receive(:execute)
+      end
+
+      worker.perform
+    end
+  end
+
+  it_behaves_like 'an idempotent worker' do
+    let!(:unit_test_1) { create(:ci_unit_test) }
+    let!(:unit_test_2) { create(:ci_unit_test) }
+    let!(:unit_test_1_recent_failure) { create(:ci_unit_test_failure, unit_test: unit_test_1) }
+    let!(:unit_test_2_old_failure) { create(:ci_unit_test_failure, unit_test: unit_test_2, failed_at: 15.days.ago) }
+
+    it 'only deletes old unit tests and their failures' do
+      subject
+
+      expect(unit_test_1.reload).to be_persisted
+      expect(unit_test_1_recent_failure.reload).to be_persisted
+      expect(Ci::UnitTest.find_by(id: unit_test_2.id)).to be_nil
+      expect(Ci::UnitTestFailure.find_by(id: unit_test_2_old_failure.id)).to be_nil
+    end
+  end
+end