Resync repositories that have been updated recently

app/workers/geo_backfill_worker.rb

@@ -2,15 +2,18 @@ class GeoBackfillWorker
   include Sidekiq::Worker
   include CronjobQueue
 
-  RUN_TIME = 5.minutes.to_i.freeze
-  BATCH_SIZE = 100.freeze
+  RUN_TIME = 5.minutes.to_i
+  BATCH_SIZE = 100
+  LAST_SYNC_INTERVAL = 24.hours
 
   def perform
     return unless Gitlab::Geo.configured?
     return unless Gitlab::Geo.primary_node.present?
 
-    start_time  = Time.now
-    project_ids = find_project_ids
+    start_time = Time.now
+    project_ids_not_synced = find_project_ids_not_synced
+    project_ids_updated_recently = find_synced_project_ids_updated_recently
+    project_ids = interleave(project_ids_not_synced, project_ids_updated_recently)
 
     logger.info "Started Geo backfilling for #{project_ids.length} project(s)"
 
@@ -38,12 +41,34 @@ class GeoBackfillWorker
 
   private
 
-  def find_project_ids
+  def find_project_ids_not_synced
     Project.where.not(id: Geo::ProjectRegistry.synced.pluck(:project_id))
            .limit(BATCH_SIZE)
            .pluck(:id)
   end
 
+  def find_synced_project_ids_updated_recently
+    Geo::ProjectRegistry.where(project_id: find_project_ids_updated_recently)
+                        .where('last_repository_synced_at <= ?', LAST_SYNC_INTERVAL.ago)
+                        .order(last_repository_synced_at: :asc)
+                        .limit(BATCH_SIZE)
+                        .pluck(:project_id)
+  end
+
+  def find_project_ids_updated_recently
+    Project.where(id: Geo::ProjectRegistry.synced.pluck(:project_id))
+           .where('last_repository_updated_at >= ?', LAST_SYNC_INTERVAL.ago)
+           .pluck(:id)
+  end
+
+  def interleave(first, second)
+    if first.length >= second.length
+      first.zip(second)
+    else
+      second.zip(first).map(&:reverse)
+    end.flatten(1).compact.take(BATCH_SIZE)
+  end
+
   def over_time?(start_time)
     Time.now - start_time >= RUN_TIME
   end

spec/workers/geo/geo_backfill_worker_spec.rb

@@ -3,7 +3,8 @@ require 'spec_helper'
 describe Geo::GeoBackfillWorker, services: true do
   let!(:primary)   { create(:geo_node, :primary, host: 'primary-geo-node') }
   let!(:secondary) { create(:geo_node, :current) }
-  let!(:projects)  { create_list(:empty_project, 2) }
+  let!(:project_1) { create(:empty_project) }
+  let!(:project_2) { create(:empty_project) }
 
   subject { described_class.new }
 
@@ -20,7 +21,7 @@ describe Geo::GeoBackfillWorker, services: true do
 
     it 'performs Geo::RepositoryBackfillService for projects where last attempt to backfill failed' do
       Geo::ProjectRegistry.create(
-        project: Project.first,
+        project: project_1,
         last_repository_synced_at: DateTime.now,
         last_repository_successful_sync_at: nil
       )
@@ -30,6 +31,27 @@ describe Geo::GeoBackfillWorker, services: true do
       subject.perform
     end
 
+    it 'performs Geo::RepositoryBackfillService for backfilled projects updated recently' do
+      Geo::ProjectRegistry.create(
+        project: project_1,
+        last_repository_synced_at: 2.days.ago,
+        last_repository_successful_sync_at: 2.days.ago
+      )
+
+      Geo::ProjectRegistry.create(
+        project: project_2,
+        last_repository_synced_at: 2.days.ago,
+        last_repository_successful_sync_at: 2.days.ago
+      )
+
+      project_1.update_attribute(:last_repository_updated_at, 2.days.ago)
+      project_2.update_attribute(:last_repository_updated_at, 10.minutes.ago)
+
+      expect(Geo::RepositoryBackfillService).to receive(:new).once.and_return(spy)
+
+      subject.perform
+    end
+
     it 'does not perform Geo::RepositoryBackfillService when tracking DB is not available' do
       allow(Rails.configuration).to receive(:respond_to?).with(:geo_database) { false }