Merge branch 'lm-improve-cancelable-pending-pipelines-query' into 'master'

Splits up auto_cancelable_pipelines query and adds limit for pipelines

See merge request gitlab-org/gitlab!68585

app/models/ci/pipeline.rb

@@ -308,6 +308,7 @@ module Ci
     scope :ci_and_parent_sources, -> { where(source: Enums::Ci::Pipeline.ci_and_parent_sources.values) }
     scope :for_user, -> (user) { where(user: user) }
     scope :for_sha, -> (sha) { where(sha: sha) }
+    scope :where_not_sha, -> (sha) { where.not(sha: sha) }
     scope :for_source_sha, -> (source_sha) { where(source_sha: source_sha) }
     scope :for_sha_or_source_sha, -> (sha) { for_sha(sha).or(for_source_sha(sha)) }
     scope :for_ref, -> (ref) { where(ref: ref) }

lib/gitlab/ci/pipeline/chain/cancel_pending_pipelines.rb

@@ -7,15 +7,19 @@ module Gitlab
         class CancelPendingPipelines < Chain::Base
           include Chain::Helpers
 
+          BATCH_SIZE = 25
+
+          # rubocop: disable CodeReuse/ActiveRecord
           def perform!
             return unless project.auto_cancel_pending_pipelines?
 
             Gitlab::OptimisticLocking.retry_lock(auto_cancelable_pipelines, name: 'cancel_pending_pipelines') do |cancelables|
-              cancelables.find_each do |cancelable|
-                cancelable.auto_cancel_running(pipeline)
+              cancelables.select(:id).each_batch(of: BATCH_SIZE) do |cancelables_batch|
+                auto_cancel_interruptible_pipelines(cancelables_batch.ids)
               end
             end
           end
+          # rubocop: enable CodeReuse/ActiveRecord
 
           def break?
             false
@@ -23,16 +27,21 @@ module Gitlab
 
           private
 
-          # rubocop: disable CodeReuse/ActiveRecord
           def auto_cancelable_pipelines
-            project.all_pipelines.ci_and_parent_sources
-              .where(ref: pipeline.ref)
-              .where.not(id: pipeline.same_family_pipeline_ids)
-              .where.not(sha: project.commit(pipeline.ref).try(:id))
+            project.all_pipelines.created_after(1.week.ago)
+              .ci_and_parent_sources
+              .for_ref(pipeline.ref)
+              .id_not_in(pipeline.same_family_pipeline_ids)
+              .where_not_sha(project.commit(pipeline.ref).try(:id))
               .alive_or_scheduled
+          end
+
+          def auto_cancel_interruptible_pipelines(pipeline_ids)
+            ::Ci::Pipeline
+              .id_in(pipeline_ids)
               .with_only_interruptible_builds
+              .each { |cancelable| cancelable.auto_cancel_running(pipeline) }
           end
-          # rubocop: enable CodeReuse/ActiveRecord
         end
       end
     end

spec/lib/gitlab/ci/pipeline/chain/cancel_pending_pipelines_spec.rb

@@ -44,6 +44,14 @@ RSpec.describe Gitlab::Ci::Pipeline::Chain::CancelPendingPipelines do
         expect(build_statuses(pipeline)).to contain_exactly('pending')
       end
 
+      it 'cancels the builds with 2 queries to avoid query timeout' do
+        second_query_regex = /WHERE "ci_pipelines"\."id" = \d+ AND \(NOT EXISTS/
+        recorder = ActiveRecord::QueryRecorder.new { perform }
+        second_query = recorder.occurrences.keys.filter { |occ| occ =~ second_query_regex }
+
+        expect(second_query).to be_one
+      end
+
       context 'when the previous pipeline has a child pipeline' do
         let(:child_pipeline) { create(:ci_pipeline, child_of: prev_pipeline) }
 

spec/models/ci/pipeline_spec.rb

@@ -183,6 +183,28 @@ RSpec.describe Ci::Pipeline, :mailer, factory_default: :keep do
     end
   end
 
+  describe '.where_not_sha' do
+    let_it_be(:pipeline) { create(:ci_pipeline, sha: 'abcx') }
+    let_it_be(:pipeline_2) { create(:ci_pipeline, sha: 'abc') }
+
+    let(:sha) { 'abc' }
+
+    subject { described_class.where_not_sha(sha) }
+
+    it 'returns the pipeline without the specified sha' do
+      is_expected.to contain_exactly(pipeline)
+    end
+
+    context 'when argument is array' do
+      let(:sha) { %w[abc abcx] }
+
+      it 'returns the pipelines without the specified shas' do
+        pipeline_3 = create(:ci_pipeline, sha: 'abcy')
+        is_expected.to contain_exactly(pipeline_3)
+      end
+    end
+  end
+
   describe '.for_source_sha' do
     subject { described_class.for_source_sha(source_sha) }