Make batch size adaptive and attempt to download LFS object lists

app/services/projects/lfs_pointers/lfs_download_link_list_service.rb

@@ -13,6 +13,7 @@ module Projects
 
       DownloadLinksError = Class.new(StandardError)
       DownloadLinkNotFound = Class.new(StandardError)
+      DownloadLinksRequestEntityTooLargeError = Class.new(StandardError)
 
       attr_reader :remote_uri
 
@@ -29,22 +30,39 @@ module Projects
       def execute(oids)
         return [] unless project&.lfs_enabled? && remote_uri && oids.present?
 
+        get_download_links_in_batches(oids)
+      end
+
+      private
+
+      def get_download_links_in_batches(oids, batch_size = REQUEST_BATCH_SIZE)
         download_links = []
 
-        oids.each_slice(REQUEST_BATCH_SIZE) do |batch|
+        oids.each_slice(batch_size) do |batch|
           download_links += get_download_links(batch)
         end
 
         download_links
-      end
 
-      private
+      rescue DownloadLinksRequestEntityTooLargeError => e
+        # Log this exceptions to see how open it happens
+        Gitlab::ErrorTracking
+          .track_exception(e, project_id: project&.id, batch_size: batch_size, oids_count: oids.count)
+
+        # Try again with a smaller batch
+        batch_size /= 2
+
+        retry if batch_size > REQUEST_BATCH_SIZE / 3
+
+        raise DownloadLinksError, 'Unable to download due to RequestEntityTooLarge errors'
+      end
 
       def get_download_links(oids)
         response = Gitlab::HTTP.post(remote_uri,
                                      body: request_body(oids),
                                      headers: headers)
 
+        raise DownloadLinksRequestEntityTooLargeError if response.request_entity_too_large?
         raise DownloadLinksError, response.message unless response.success?
 
         # Since the LFS Batch API may return a Content-Ttpe of

spec/services/projects/lfs_pointers/lfs_download_link_list_service_spec.rb

@@ -8,11 +8,15 @@ describe Projects::LfsPointers::LfsDownloadLinkListService do
   let(:new_oids) { { 'oid1' => 123, 'oid2' => 125 } }
   let(:remote_uri) { URI.parse(lfs_endpoint) }
 
+  let(:request_object) { HTTParty::Request.new(Net::HTTP::Post, '/') }
+  let(:parsed_block) { lambda {} }
+  let(:success_net_response) { Net::HTTPOK.new('', '', '') }
+  let(:response) { Gitlab::HTTP::Response.new(request_object, net_response, parsed_block) }
+
   def objects_response(oids)
     body = oids.map do |oid, size|
       {
-        'oid' => oid,
-        'size' => size,
+        'oid' => oid, 'size' => size,
         'actions' => {
           'download' => { 'href' => "#{import_url}/gitlab-lfs/objects/#{oid}" }
         }
@@ -22,6 +26,11 @@ describe Projects::LfsPointers::LfsDownloadLinkListService do
     Struct.new(:success?, :objects).new(true, body).to_json
   end
 
+  def custom_response(net_response, body = nil)
+    allow(net_response).to receive(:body).and_return(body)
+    Gitlab::HTTP::Response.new(request_object, net_response, parsed_block)
+  end
+
   let(:invalid_object_response) do
     [
       'oid' => 'whatever',
@@ -33,9 +42,8 @@ describe Projects::LfsPointers::LfsDownloadLinkListService do
 
   before do
     allow(project).to receive(:lfs_enabled?).and_return(true)
-    response = instance_double(Gitlab::HTTP::Response)
-    allow(response).to receive(:body).and_return(objects_response(new_oids))
-    allow(response).to receive(:success?).and_return(true)
+
+    response = custom_response(success_net_response, objects_response(new_oids))
     allow(Gitlab::HTTP).to receive(:post).and_return(response)
   end
 
@@ -47,8 +55,18 @@ describe Projects::LfsPointers::LfsDownloadLinkListService do
     end
 
     context 'when lfs objects size is larger than the batch size' do
-      def stub_request(batch)
-        response = instance_double Gitlab::HTTP::Response, body: objects_response(batch), success?: true
+      def stub_successful_request(batch)
+        response = custom_response(success_net_response, objects_response(batch))
+        stub_request(batch, response)
+      end
+
+      def stub_entity_too_large_error_request(batch)
+        entity_too_large_net_response = Net::HTTPRequestEntityTooLarge.new('', '', '')
+        response = custom_response(entity_too_large_net_response)
+        stub_request(batch, response)
+      end
+
+      def stub_request(batch, response)
         expect(Gitlab::HTTP).to receive(:post).with(
           remote_uri,
           {
@@ -60,17 +78,74 @@ describe Projects::LfsPointers::LfsDownloadLinkListService do
 
       let(:new_oids) { { 'oid1' => 123, 'oid2' => 125, 'oid3' => 126, 'oid4' => 127, 'oid5' => 128 } }
 
-      before do
-        stub_const("#{described_class.name}::REQUEST_BATCH_SIZE", 2)
+      context 'when batch size' do
+        before do
+          stub_const("#{described_class.name}::REQUEST_BATCH_SIZE", 2)
 
-        stub_request([['oid1', 123], ['oid2', 125]])
-        stub_request([['oid3', 126], ['oid4', 127]])
-        stub_request([['oid5', 128]])
+          data = new_oids.to_a
+          stub_successful_request([data[0], data[1]])
+          stub_successful_request([data[2], data[3]])
+          stub_successful_request([data[4]])
+        end
+
+        it 'retreives them in batches' do
+          subject.execute(new_oids).each do |lfs_download_object|
+            expect(lfs_download_object.link).to eq "#{import_url}/gitlab-lfs/objects/#{lfs_download_object.oid}"
+          end
+        end
       end
 
-      it 'retreives them in batches' do
-        subject.execute(new_oids).each do |lfs_download_object|
-          expect(lfs_download_object.link).to eq "#{import_url}/gitlab-lfs/objects/#{lfs_download_object.oid}"
+      context 'when request fails with PayloadTooLarge error' do
+        let(:error_class) { described_class::DownloadLinksRequestEntityTooLargeError }
+
+        context 'when the smaller batch eventually works' do
+          before do
+            stub_const("#{described_class.name}::REQUEST_BATCH_SIZE", 5)
+
+            data = new_oids.to_a
+
+            # with the batch size of 5
+            stub_entity_too_large_error_request(data)
+
+            # with the batch size of 2
+            stub_successful_request([data[0], data[1]])
+            stub_successful_request([data[2], data[3]])
+            stub_successful_request([data[4]])
+          end
+
+          it 'retreives them eventually and logs exceptions' do
+            expect(Gitlab::ErrorTracking).to receive(:track_exception).with(
+              an_instance_of(error_class), project_id: project.id, batch_size: 5, oids_count: 5
+            )
+
+            subject.execute(new_oids).each do |lfs_download_object|
+              expect(lfs_download_object.link).to eq "#{import_url}/gitlab-lfs/objects/#{lfs_download_object.oid}"
+            end
+          end
+        end
+
+        context 'when batch size cannot be any smaller' do
+          before do
+            stub_const("#{described_class.name}::REQUEST_BATCH_SIZE", 5)
+
+            data = new_oids.to_a
+
+            # with the batch size of 5
+            stub_entity_too_large_error_request(data)
+
+            # with the batch size of 2
+            stub_entity_too_large_error_request([data[0], data[1]])
+          end
+
+          it 'raises an error and logs exceptions' do
+            expect(Gitlab::ErrorTracking).to receive(:track_exception).with(
+              an_instance_of(error_class), project_id: project.id, batch_size: 5, oids_count: 5
+            )
+            expect(Gitlab::ErrorTracking).to receive(:track_exception).with(
+              an_instance_of(error_class), project_id: project.id, batch_size: 2, oids_count: 5
+            )
+            expect { subject.execute(new_oids) }.to raise_error(described_class::DownloadLinksError)
+          end
         end
       end
     end
@@ -116,17 +191,22 @@ describe Projects::LfsPointers::LfsDownloadLinkListService do
   end
 
   describe '#get_download_links' do
-    it 'raise error if request fails' do
-      allow(Gitlab::HTTP).to receive(:post).and_return(Struct.new(:success?, :message).new(false, 'Failed request'))
+    context 'if request fails' do
+      before do
+        request_timeout_net_response = Net::HTTPRequestTimeout.new('', '', '')
+        response = custom_response(request_timeout_net_response)
+        allow(Gitlab::HTTP).to receive(:post).and_return(response)
+      end
 
-      expect { subject.send(:get_download_links, new_oids) }.to raise_error(described_class::DownloadLinksError)
+      it 'raises an error' do
+        expect { subject.send(:get_download_links, new_oids) }.to raise_error(described_class::DownloadLinksError)
+      end
     end
 
     shared_examples 'JSON parse errors' do |body|
-      it 'raises error' do
-        response = instance_double(Gitlab::HTTP::Response)
+      it 'raises an error' do
+        response = custom_response(success_net_response)
         allow(response).to receive(:body).and_return(body)
-        allow(response).to receive(:success?).and_return(true)
         allow(Gitlab::HTTP).to receive(:post).and_return(response)
 
         expect { subject.send(:get_download_links, new_oids) }.to raise_error(described_class::DownloadLinksError)