Merge branch...

Merge branch '8798-geo-implement-selective-sync-support-for-fdw-queries-to-find-verified-registries' into 'master'

Geo - Add selective sync support for FDW queries to find verified registries

See merge request gitlab-org/gitlab-ee!10255

ee/app/finders/geo/legacy_project_registry_verified_finder.rb

+# frozen_string_literal: true
+
+# Finder for retrieving project registries that have been verified
+# scoped to a type (repository or wiki) using cross-database joins
+# for selective sync.
+#
+# Basic usage:
+#
+#     Geo::LegacyProjectRegistryVerifiedFinder.new(current_node: Gitlab::Geo.current_node, :repository).execute
+#
+# Valid `type` values are:
+#
+# * `:repository`
+# * `:wiki`
+#
+# Any other value will be ignored.
+module Geo
+  class LegacyProjectRegistryVerifiedFinder < RegistryFinder
+    def initialize(current_node:, type:)
+      super(current_node: current_node)
+      @type = type.to_s.to_sym
+    end
+
+    def execute
+      if selective_sync?
+        verified_registries_for_selective_sync
+      else
+        verified_registries
+      end
+    end
+
+    private
+
+    attr_reader :type
+
+    def verified_registries
+      Geo::ProjectRegistry.verified(type)
+    end
+
+    # rubocop: disable CodeReuse/ActiveRecord
+    def verified_registries_for_selective_sync
+      legacy_inner_join_registry_ids(
+        verified_registries,
+        current_node.projects.pluck(:id),
+        Geo::ProjectRegistry,
+        foreign_key: :project_id
+      )
+    end
+    # rubocop: enable CodeReuse/ActiveRecord
+  end
+end

ee/app/finders/geo/project_registry_finder.rb

@@ -27,25 +27,11 @@ module Geo
     end
 
     def count_verified_repositories
-      relation =
-        if use_legacy_queries?
-          legacy_find_verified_repositories
-        else
-          find_verified_repositories
-        end
-
-      relation.count
+      registries_for_verified_projects(:repository).count
     end
 
     def count_verified_wikis
-      relation =
-        if use_legacy_queries?
-          legacy_find_verified_wikis
-        else
-          fdw_find_verified_wikis
-        end
-
-      relation.count
+      registries_for_verified_projects(:wiki).count
     end
 
     def count_repositories_checksum_mismatch
@@ -125,38 +111,6 @@ module Geo
 
     protected
 
-    def finder_klass_for_synced_registries
-      if Gitlab::Geo::Fdw.enabled_for_selective_sync?
-        Geo::ProjectRegistrySyncedFinder
-      else
-        Geo::LegacyProjectRegistrySyncedFinder
-      end
-    end
-
-    def registries_for_synced_projects(type)
-      finder_klass_for_synced_registries
-        .new(current_node: current_node, type: type)
-        .execute
-    end
-
-    def finder_klass_for_failed_registries
-      if Gitlab::Geo::Fdw.enabled_for_selective_sync?
-        Geo::ProjectRegistrySyncFailedFinder
-      else
-        Geo::LegacyProjectRegistrySyncFailedFinder
-      end
-    end
-
-    def registries_for_failed_projects(type)
-      finder_klass_for_failed_registries
-        .new(current_node: current_node, type: type)
-        .execute
-    end
-
-    def find_verified_repositories
-      Geo::ProjectRegistry.verified_repos
-    end
-
     def find_filtered_verification_failed_project_registries(type = nil)
       case type
       when 'repository'
@@ -221,11 +175,6 @@ module Geo
     end
     # rubocop: enable CodeReuse/ActiveRecord
 
-    # @return [ActiveRecord::Relation<Geo::ProjectRegistry>]
-    def fdw_find_verified_wikis
-      Geo::ProjectRegistry.verified_wikis
-    end
-
     def fdw_inner_join_projects
       local_registry_table
         .join(fdw_project_table, Arel::Nodes::InnerJoin)
@@ -280,27 +229,6 @@ module Geo
       ::Gitlab::SQL::Glob.q(value)
     end
 
-    # @return [ActiveRecord::Relation<Geo::ProjectRegistry>] list of verified projects
-    def legacy_find_verified_repositories
-      legacy_find_project_registries(Geo::ProjectRegistry.verified_repos)
-    end
-
-    # @return [ActiveRecord::Relation<Geo::ProjectRegistry>] list of verified wikis
-    def legacy_find_verified_wikis
-      legacy_find_project_registries(Geo::ProjectRegistry.verified_wikis)
-    end
-
-    # @return [ActiveRecord::Relation<Project>] list of synced projects
-    # rubocop: disable CodeReuse/ActiveRecord
-    def legacy_find_project_registries(project_registries)
-      legacy_inner_join_registry_ids(
-        current_node.projects,
-        project_registries.pluck(:project_id),
-        Project
-      )
-    end
-    # rubocop: enable CodeReuse/ActiveRecord
-
     # @return [ActiveRecord::Relation<Geo::ProjectRegistry>] list of projects that verification has failed
     # rubocop: disable CodeReuse/ActiveRecord
     def legacy_find_filtered_verification_failed_projects(type = nil)
@@ -400,5 +328,49 @@ module Geo
     def wiki_missing_on_primary_is_not_true
       Arel::Nodes::SqlLiteral.new("project_registry.wiki_missing_on_primary IS NOT TRUE")
     end
+
+    private
+
+    def finder_klass_for_synced_registries
+      if Gitlab::Geo::Fdw.enabled_for_selective_sync?
+        Geo::ProjectRegistrySyncedFinder
+      else
+        Geo::LegacyProjectRegistrySyncedFinder
+      end
+    end
+
+    def registries_for_synced_projects(type)
+      finder_klass_for_synced_registries
+        .new(current_node: current_node, type: type)
+        .execute
+    end
+
+    def finder_klass_for_failed_registries
+      if Gitlab::Geo::Fdw.enabled_for_selective_sync?
+        Geo::ProjectRegistrySyncFailedFinder
+      else
+        Geo::LegacyProjectRegistrySyncFailedFinder
+      end
+    end
+
+    def registries_for_failed_projects(type)
+      finder_klass_for_failed_registries
+        .new(current_node: current_node, type: type)
+        .execute
+    end
+
+    def finder_klass_for_verified_registries
+      if Gitlab::Geo::Fdw.enabled_for_selective_sync?
+        Geo::ProjectRegistryVerifiedFinder
+      else
+        Geo::LegacyProjectRegistryVerifiedFinder
+      end
+    end
+
+    def registries_for_verified_projects(type)
+      finder_klass_for_verified_registries
+        .new(current_node: current_node, type: type)
+        .execute
+    end
   end
 end

ee/app/finders/geo/project_registry_verified_finder.rb

+# frozen_string_literal: true
+
+# Finder for retrieving project registries that have been verified
+# scoped to a type (repository or wiki) using FDW queries.
+#
+# Basic usage:
+#
+#     Geo::ProjectRegistryVerifiedFinder.new(current_node: Gitlab::Geo.current_node, :repository).execute
+#
+# Valid `type` values are:
+#
+# * `:repository`
+# * `:wiki`
+#
+# Any other value will be ignored.
+module Geo
+  class ProjectRegistryVerifiedFinder
+    def initialize(current_node:, type:)
+      @current_node = Geo::Fdw::GeoNode.find(current_node.id)
+      @type = type.to_sym
+    end
+
+    def execute
+      current_node.project_registries.verified(type)
+    end
+
+    private
+
+    attr_reader :current_node, :type
+  end
+end

ee/app/models/geo/project_registry.rb

@@ -112,6 +112,17 @@ class Geo::ProjectRegistry < Geo::BaseRegistry
     end
   end
 
+  def self.verified(type)
+    case type
+    when :repository
+      verified_repos
+    when :wiki
+      verified_wikis
+    else
+      none
+    end
+  end
+
   def self.flag_repositories_for_resync!
     update_all(
       resync_repository: true,

ee/changelogs/unreleased/8798-geo-implement-selective-sync-support-for-fdw-queries-to-find-verified-registries.yml

+---
+title: Geo - Add selective sync support for FDW queries to find verified registries
+merge_request: 10255
+author:
+type: changed

ee/spec/finders/geo/legacy_project_registry_verified_finder_spec.rb

+# frozen_string_literal: true
+
+require 'spec_helper'
+
+describe Geo::LegacyProjectRegistryVerifiedFinder, :geo do
+  include EE::GeoHelpers
+
+  describe '#execute' do
+    let(:node) { create(:geo_node) }
+    let(:group_1) { create(:group) }
+    let(:group_2) { create(:group) }
+    let(:nested_group_1) { create(:group, parent: group_1) }
+    let(:project_1) { create(:project, group: group_1) }
+    let(:project_2) { create(:project, group: nested_group_1) }
+    let(:project_3) { create(:project, group: nested_group_1) }
+    let(:project_4) { create(:project, :broken_storage, group: group_2) }
+    let(:project_5) { create(:project, :broken_storage, group: group_2) }
+    let!(:registry_verified) { create(:geo_project_registry, :repository_verified, :wiki_verified, project: project_1) }
+    let!(:registry_repository_verification_failed) { create(:geo_project_registry, :repository_verification_failed, :wiki_verified, project: project_2) }
+    let!(:registry_repository_verification_failed_broken_shard) { create(:geo_project_registry, :repository_verification_failed, :wiki_verified, project: project_5) }
+    let!(:registry_wiki_verification_failed) { create(:geo_project_registry, :repository_verified, :wiki_verification_failed, project: project_3) }
+    let!(:registry_wiki_verification_failed_broken_shard) { create(:geo_project_registry, :repository_verified, :wiki_verification_failed, project: project_4) }
+    let!(:registry_verification_failed) { create(:geo_project_registry, :repository_verification_failed, :wiki_verification_failed) }
+
+    shared_examples 'finds verified registries' do
+      context 'with repository type' do
+        subject { described_class.new(current_node: node, type: :repository) }
+
+        context 'without selective sync' do
+          it 'returns all verified registries' do
+            expect(subject.execute).to contain_exactly(registry_verified, registry_wiki_verification_failed, registry_wiki_verification_failed_broken_shard)
+          end
+        end
+
+        context 'with selective sync by namespace' do
+          it 'returns verified registries where projects belongs to the namespaces' do
+            node.update!(selective_sync_type: 'namespaces', namespaces: [group_1, nested_group_1])
+
+            expect(subject.execute).to contain_exactly(registry_verified, registry_wiki_verification_failed)
+          end
+        end
+
+        context 'with selective sync by shard' do
+          it 'returns verified registries where projects belongs to the shards' do
+            node.update!(selective_sync_type: 'shards', selective_sync_shards: ['broken'])
+
+            expect(subject.execute).to contain_exactly(registry_wiki_verification_failed_broken_shard)
+          end
+        end
+      end
+
+      context 'with wiki type' do
+        subject { described_class.new(current_node: node, type: :wiki) }
+
+        context 'without selective sync' do
+          it 'returns all verified registries' do
+            expect(subject.execute).to contain_exactly(registry_verified, registry_repository_verification_failed, registry_repository_verification_failed_broken_shard)
+          end
+        end
+
+        context 'with selective sync by namespace' do
+          it 'returns verified registries where projects belongs to the namespaces' do
+            node.update!(selective_sync_type: 'namespaces', namespaces: [group_1, nested_group_1])
+
+            expect(subject.execute).to contain_exactly(registry_verified, registry_repository_verification_failed)
+          end
+        end
+
+        context 'with selective sync by shard' do
+          it 'returns verified registries where projects belongs to the shards' do
+            node.update!(selective_sync_type: 'shards', selective_sync_shards: ['broken'])
+
+            expect(subject.execute).to contain_exactly(registry_repository_verification_failed_broken_shard)
+          end
+        end
+      end
+
+      context 'with invalid type' do
+        subject { described_class.new(current_node: node, type: :invalid) }
+
+        it 'returns nothing' do
+          expect(subject.execute).to be_empty
+        end
+      end
+    end
+
+    # Disable transactions via :delete method because a foreign table
+    # can't see changes inside a transaction of a different connection.
+    context 'FDW', :delete do
+      before do
+        skip('FDW is not configured') unless Gitlab::Geo::Fdw.enabled?
+      end
+
+      include_examples 'finds verified registries'
+    end
+
+    context 'Legacy' do
+      before do
+        stub_fdw_disabled
+      end
+
+      include_examples 'finds verified registries'
+    end
+  end
+end

ee/spec/finders/geo/project_registry_finder_spec.rb

@@ -146,22 +146,6 @@ describe Geo::ProjectRegistryFinder, :geo do
     end
 
     describe '#count_verified_repositories' do
-      it 'delegates to #find_verified_repositories when use_legacy_queries is false' do
-        expect(subject).to receive(:use_legacy_queries?).and_return(false)
-
-        expect(subject).to receive(:find_verified_repositories).and_call_original
-
-        subject.count_verified_repositories
-      end
-
-      it 'delegates to #legacy_find_verified_repositories when use_legacy_queries is true' do
-        expect(subject).to receive(:use_legacy_queries?).and_return(true)
-
-        expect(subject).to receive(:legacy_find_verified_repositories).and_call_original
-
-        subject.count_verified_repositories
-      end
-
       it 'counts projects that verified' do
         create(:geo_project_registry, :repository_verified, project: project_repository_verified)
         create(:geo_project_registry, :repository_verified, project: build(:project))
@@ -172,12 +156,6 @@ describe Geo::ProjectRegistryFinder, :geo do
     end
 
     describe '#count_verified_wikis' do
-      it 'delegates to the correct method' do
-        expect(subject).to receive("#{method_prefix}_find_verified_wikis".to_sym).and_call_original
-
-        subject.count_verified_wikis
-      end
-
       it 'counts wikis that verified' do
         create(:geo_project_registry, :wiki_verified, project: project_wiki_verified)
         create(:geo_project_registry, :wiki_verified, project: build(:project))

ee/spec/finders/geo/project_registry_verified_finder_spec.rb

+# frozen_string_literal: true
+
+require 'spec_helper'
+
+describe Geo::ProjectRegistryVerifiedFinder, :geo do
+  # Disable transactions via :delete method because a foreign table
+  # can't see changes inside a transaction of a different connection.
+  describe '#execute', :delete do
+    let(:node) { create(:geo_node) }
+    let(:group_1) { create(:group) }
+    let(:group_2) { create(:group) }
+    let(:nested_group_1) { create(:group, parent: group_1) }
+    let(:project_1) { create(:project, group: group_1) }
+    let(:project_2) { create(:project, group: nested_group_1) }
+    let(:project_3) { create(:project, group: nested_group_1) }
+    let(:project_4) { create(:project, :broken_storage, group: group_2) }
+    let(:project_5) { create(:project, :broken_storage, group: group_2) }
+    let!(:registry_verified) { create(:geo_project_registry, :repository_verified, :wiki_verified, project: project_1) }
+    let!(:registry_repository_verification_failed) { create(:geo_project_registry, :repository_verification_failed, :wiki_verified, project: project_2) }
+    let!(:registry_repository_verification_failed_broken_shard) { create(:geo_project_registry, :repository_verification_failed, :wiki_verified, project: project_5) }
+    let!(:registry_wiki_verification_failed) { create(:geo_project_registry, :repository_verified, :wiki_verification_failed, project: project_3) }
+    let!(:registry_wiki_verification_failed_broken_shard) { create(:geo_project_registry, :repository_verified, :wiki_verification_failed, project: project_4) }
+    let!(:registry_verification_failed) { create(:geo_project_registry, :repository_verification_failed, :wiki_verification_failed) }
+
+    before do
+      skip('FDW is not configured') unless Gitlab::Geo::Fdw.enabled?
+    end
+
+    context 'with repository type' do
+      subject { described_class.new(current_node: node, type: :repository) }
+
+      context 'without selective sync' do
+        it 'returns all verified registries' do
+          expect(subject.execute).to contain_exactly(registry_verified, registry_wiki_verification_failed, registry_wiki_verification_failed_broken_shard)
+        end
+      end
+
+      context 'with selective sync by namespace' do
+        it 'returns verified registries where projects belongs to the namespaces' do
+          node.update!(selective_sync_type: 'namespaces', namespaces: [group_1, nested_group_1])
+
+          expect(subject.execute).to contain_exactly(registry_verified, registry_wiki_verification_failed)
+        end
+      end
+
+      context 'with selective sync by shard' do
+        it 'returns verified registries where projects belongs to the shards' do
+          node.update!(selective_sync_type: 'shards', selective_sync_shards: ['broken'])
+
+          expect(subject.execute).to contain_exactly(registry_wiki_verification_failed_broken_shard)
+        end
+      end
+    end
+
+    context 'with wiki type' do
+      subject { described_class.new(current_node: node, type: :wiki) }
+
+      context 'without selective sync' do
+        it 'returns all verified registries' do
+          expect(subject.execute).to contain_exactly(registry_verified, registry_repository_verification_failed, registry_repository_verification_failed_broken_shard)
+        end
+      end
+
+      context 'with selective sync by namespace' do
+        it 'returns verified registries where projects belongs to the namespaces' do
+          node.update!(selective_sync_type: 'namespaces', namespaces: [group_1, nested_group_1])
+
+          expect(subject.execute).to contain_exactly(registry_verified, registry_repository_verification_failed)
+        end
+      end
+
+      context 'with selective sync by shard' do
+        it 'returns verified registries where projects belongs to the shards' do
+          node.update!(selective_sync_type: 'shards', selective_sync_shards: ['broken'])
+
+          expect(subject.execute).to contain_exactly(registry_repository_verification_failed_broken_shard)
+        end
+      end
+    end
+
+    context 'with invalid type' do
+      subject { described_class.new(current_node: node, type: :invalid) }
+
+      it 'returns nothing' do
+        expect(subject.execute).to be_empty
+      end
+    end
+  end
+end