Add methods to retrieve design repositories to sync

ee/app/finders/geo/design_registry_finder.rb

@@ -20,6 +20,41 @@ module Geo
       registries.count
     end
 
+    # Returns Geo::DesignRegistry records that have never been synced.
+    #
+    # Does not care about selective sync, because it considers the Registry
+    # table to be the single source of truth. The contract is that other
+    # processes need to ensure that the table only contains records that should
+    # be synced.
+    #
+    # Any registries that have ever been synced that currently need to be
+    # resynced will be handled by other find methods (like
+    # #find_retryable_dirty_registries)
+    #
+    # You can pass a list with `except_ids:` so you can exclude items you
+    # already scheduled but haven't finished and aren't persisted to the database yet
+    #
+    # @param [Integer] batch_size used to limit the results returned
+    # @param [Array<Integer>] except_ids ids that will be ignored from the query
+    # rubocop:disable CodeReuse/ActiveRecord
+    def find_never_synced_registries(batch_size:, except_ids: [])
+      Geo::DesignRegistry
+        .never_synced
+        .model_id_not_in(except_ids)
+        .limit(batch_size)
+    end
+    # rubocop:enable CodeReuse/ActiveRecord
+
+    # rubocop:disable CodeReuse/ActiveRecord
+    def find_retryable_dirty_registries(batch_size:, except_ids: [])
+      Geo::DesignRegistry
+        .updated_recently
+        .model_id_not_in(except_ids)
+        .order(Gitlab::Database.nulls_first_order(:last_synced_at))
+        .limit(batch_size)
+    end
+    # rubocop:enable CodeReuse/ActiveRecord
+
     private
 
     def registries

ee/app/models/geo/design_registry.rb

@@ -10,6 +10,7 @@ class Geo::DesignRegistry < Geo::BaseRegistry
 
   belongs_to :project
 
+  scope :never_synced, -> { with_state(:pending).where(last_synced_at: nil) }
   scope :pending, -> { with_state(:pending) }
   scope :failed, -> { with_state(:failed) }
   scope :synced, -> { with_state(:synced) }

ee/spec/finders/geo/design_registry_finder_spec.rb

@@ -2,9 +2,17 @@
 require 'spec_helper'
 
 RSpec.describe Geo::DesignRegistryFinder, :geo, :geo_fdw do
-  include ::EE::GeoHelpers
-
   let!(:secondary) { create(:geo_node) }
+  let!(:project_1) { create(:project) }
+  let!(:project_2) { create(:project) }
+  let!(:project_3) { create(:project) }
+  let!(:project_4) { create(:project) }
+  let!(:project_5) { create(:project) }
+  let!(:project_6) { create(:project) }
+
+  subject { described_class.new(current_node_id: secondary.id) }
+
+  context 'count all the things' do
     let!(:failed_registry) { create(:geo_design_registry, :sync_failed) }
     let!(:synced_registry) { create(:geo_design_registry, :synced) }
 
@@ -13,13 +21,6 @@ RSpec.describe Geo::DesignRegistryFinder, :geo, :geo_fdw do
     let(:synced_project) { create(:project, group: synced_group) }
     let(:unsynced_project) { create(:project, :broken_storage, group: unsynced_group) }
 
-  subject { described_class.new(current_node_id: secondary.id) }
-
-  before do
-    stub_current_geo_node(secondary)
-  end
-
-  context 'count all the things' do
     describe '#count_syncable' do
       it 'returns number of design repositories' do
         # One more design for the same project to assert absence of duplicates
@@ -103,4 +104,46 @@ RSpec.describe Geo::DesignRegistryFinder, :geo, :geo_fdw do
       end
     end
   end
+
+  describe '#find_never_synced_registries' do
+    let!(:registry_project_1) { create(:geo_design_registry, :synced, project_id: project_1.id) }
+    let!(:registry_project_2) { create(:geo_design_registry, :sync_failed, project_id: project_2.id) }
+    let!(:registry_project_3) { create(:geo_design_registry, project_id: project_3.id, last_synced_at: nil) }
+    let!(:registry_project_4) { create(:geo_design_registry, project_id: project_4.id, last_synced_at: 3.days.ago, retry_at: 2.days.ago) }
+    let!(:registry_project_5) { create(:geo_design_registry, project_id: project_5.id, last_synced_at: 6.days.ago) }
+    let!(:registry_project_6) { create(:geo_design_registry, project_id: project_6.id, last_synced_at: nil) }
+
+    it 'returns registries for projects that have never been synced' do
+      registries = subject.find_never_synced_registries(batch_size: 10)
+
+      expect(registries).to match_ids(registry_project_3, registry_project_6)
+    end
+
+    it 'excludes except_ids' do
+      registries = subject.find_never_synced_registries(batch_size: 10, except_ids: [project_3.id])
+
+      expect(registries).to match_ids(registry_project_6)
+    end
+  end
+
+  describe '#find_retryable_dirty_registries' do
+    let!(:registry_project_1) { create(:geo_design_registry, :synced, project_id: project_1.id) }
+    let!(:registry_project_2) { create(:geo_design_registry, :sync_failed, project_id: project_2.id) }
+    let!(:registry_project_3) { create(:geo_design_registry, project_id: project_3.id, last_synced_at: nil) }
+    let!(:registry_project_4) { create(:geo_design_registry, project_id: project_4.id, last_synced_at: 3.days.ago, retry_at: 2.days.ago) }
+    let!(:registry_project_5) { create(:geo_design_registry, project_id: project_5.id, last_synced_at: 6.days.ago) }
+    let!(:registry_project_6) { create(:geo_design_registry, project_id: project_6.id, last_synced_at: nil) }
+
+    it 'returns registries for projects that have been recently updated' do
+      registries = subject.find_retryable_dirty_registries(batch_size: 10)
+
+      expect(registries).to match_ids(registry_project_2, registry_project_3, registry_project_4, registry_project_5, registry_project_6)
+    end
+
+    it 'excludes except_ids' do
+      registries = subject.find_retryable_dirty_registries(batch_size: 10, except_ids: [project_4.id, project_5.id, project_6.id])
+
+      expect(registries).to match_ids(registry_project_2, registry_project_3)
+    end
+  end
 end