Remove project restrictions from the repository backfill service

app/services/geo/repository_backfill_service.rb

@@ -10,18 +10,6 @@ module Geo
     end
 
     def execute
-      # When Geo customers upgrade to 9.0, the secondaries nodes that are
-      # enabled will start the backfilling process automatically. We need
-      # to populate the tracking database correctly for projects synced
-      # before the process being started or projects created during the
-      # backfilling. Otherwise, the query to retrieve the projects will
-      # always return the same projects because they don't have entries
-      # in the tracking database
-      if backfilled?
-        update_registry(DateTime.now, DateTime.now)
-        return
-      end
-
       try_obtain_lease do
         log('Started repository sync')
         started_at, finished_at = fetch_repositories
@@ -76,7 +64,6 @@ module Geo
 
     def try_obtain_lease
       log('Trying to obtain lease to sync repository')
-
       repository_lease = Gitlab::ExclusiveLease.new(lease_key, timeout: LEASE_TIMEOUT).try_obtain
 
       if repository_lease.nil?
@@ -93,25 +80,7 @@ module Geo
       Gitlab::ExclusiveLease.cancel(lease_key, repository_lease)
     end
 
-    def backfilled?
-      return false unless project.repository.exists?
-      return false if project.repository.exists? && project.repository.empty?
-      return false if failed_registry_exists?
-
-      true
-    end
-
-    def failed_registry_exists?
-      Geo::ProjectRegistry.failed.where(project_id: project_id).any?
-    end
-
-    def synced_registry_exists?
-      Geo::ProjectRegistry.synced.where(project_id: project_id).any?
-    end
-
     def update_registry(started_at, finished_at)
-      return if synced_registry_exists?
-
       log('Updating registry information')
       registry = Geo::ProjectRegistry.find_or_initialize_by(project_id: project_id)
       registry.last_repository_synced_at = started_at

spec/services/geo/repository_backfill_service_spec.rb

@@ -67,35 +67,67 @@ describe Geo::RepositoryBackfillService, services: true do
     context 'when repository exists and is not empty' do
       let(:project) { create(:project) }
 
-      it 'does not fetch the project repositories' do
-        expect_any_instance_of(Repository).not_to receive(:fetch_geo_mirror)
+      it 'fetches project repositories' do
+        fetch_count = 0
+
+        allow_any_instance_of(Repository).to receive(:fetch_geo_mirror) do
+          fetch_count += 1
+        end
 
         subject.execute
+
+        expect(fetch_count).to eq 2
       end
 
       context 'tracking database' do
-        it 'tracks missing repository sync' do
+        it 'tracks repository sync' do
           expect { subject.execute }.to change(Geo::ProjectRegistry, :count).by(1)
         end
+
+        it 'stores last_repository_successful_sync_at when succeed' do
+          allow_any_instance_of(Repository).to receive(:fetch_geo_mirror) { true }
+
+          subject.execute
+
+          registry = Geo::ProjectRegistry.find_by(project_id: project.id)
+
+          expect(registry.last_repository_successful_sync_at).not_to be_nil
+        end
+
+        it 'reset last_repository_successful_sync_at when fail' do
+          allow_any_instance_of(Repository).to receive(:fetch_geo_mirror) { raise Gitlab::Shell::Error }
+
+          subject.execute
+
+          registry = Geo::ProjectRegistry.find_by(project_id: project.id)
+
+          expect(registry.last_repository_successful_sync_at).to be_nil
+        end
       end
     end
 
     context 'when repository was backfilled successfully' do
       let(:project) { create(:project) }
-      let(:last_repository_successful_sync_at) { 5.days.ago }
+      let(:last_repository_synced_at) { 5.days.ago }
 
       let!(:registry) do
         Geo::ProjectRegistry.create(
           project: project,
-          last_repository_synced_at: 5.days.ago,
-          last_repository_successful_sync_at: last_repository_successful_sync_at
+          last_repository_synced_at: last_repository_synced_at,
+          last_repository_successful_sync_at: last_repository_synced_at
         )
       end
 
-      it 'does not fetch the project repositories' do
-        expect_any_instance_of(Repository).not_to receive(:fetch_geo_mirror)
+      it 'fetches project repositories' do
+        fetch_count = 0
+
+        allow_any_instance_of(Repository).to receive(:fetch_geo_mirror) do
+          fetch_count += 1
+        end
 
         subject.execute
+
+        expect(fetch_count).to eq 2
       end
 
       context 'tracking database' do
@@ -103,10 +135,26 @@ describe Geo::RepositoryBackfillService, services: true do
           expect { subject.execute }.not_to change(Geo::ProjectRegistry, :count)
         end
 
-        it 'does not update last_repository_successful_sync_at' do
+        it 'updates registry when succeed' do
+          allow_any_instance_of(Repository).to receive(:fetch_geo_mirror) { true }
+
           subject.execute
 
-          expect(registry.reload.last_repository_successful_sync_at).to be_within(1.second).of(last_repository_successful_sync_at)
+          registry.reload
+
+          expect(registry.last_repository_synced_at).to be_within(1.minute).of(Time.now)
+          expect(registry.last_repository_successful_sync_at).to be_within(1.minute).of(Time.now)
+        end
+
+        it 'does not update registry last_repository_successful_sync_at when fail' do
+          allow_any_instance_of(Repository).to receive(:fetch_geo_mirror) { raise Gitlab::Shell::Error }
+
+          subject.execute
+
+          registry.reload
+
+          expect(registry.last_repository_synced_at).to be_within(1.minute).of(Time.now)
+          expect(registry.last_repository_successful_sync_at).to be_within(1.minute).of(last_repository_synced_at)
         end
       end
     end