Find and store unhashed upload file paths

lib/gitlab/background_migration/prepare_unhashed_uploads.rb

 module Gitlab
   module BackgroundMigration
     class PrepareUnhashedUploads
+      FILE_PATH_BATCH_SIZE = 500
+      UPLOAD_DIR = "#{CarrierWave.root}/uploads"
+
       class UnhashedUploadFile < ActiveRecord::Base
         self.table_name = 'unhashed_upload_files'
       end
@@ -8,8 +11,8 @@ module Gitlab
       def perform
         return unless migrate?
 
-        clear_unhashed_upload_files
-        store_unhashed_upload_files
+        clear_unhashed_upload_file_paths
+        store_unhashed_upload_file_paths
         schedule_populate_untracked_uploads_jobs
       end
 
@@ -19,12 +22,55 @@ module Gitlab
         UnhashedUploadFile.table_exists?
       end
 
-      def clear_unhashed_upload_files
-        # TODO
+      def clear_unhashed_upload_file_paths
+        UnhashedUploadFile.delete_all
       end
 
-      def store_unhashed_upload_files
-        # TODO
+      def store_unhashed_upload_file_paths
+        return unless Dir.exists?(UPLOAD_DIR)
+
+        file_paths = []
+        each_file_path(UPLOAD_DIR) do |file_path|
+          file_paths << file_path
+
+          if file_paths.size >= FILE_PATH_BATCH_SIZE
+            insert_file_paths(file_paths)
+            file_paths = []
+          end
+        end
+
+        insert_file_paths(file_paths) if file_paths.any?
+      end
+
+      def each_file_path(search_dir, &block)
+        cmd = build_find_command(search_dir)
+        Open3.popen2(*cmd) do |stdin, stdout, status_thread|
+          stdout.each_line("\0") do |line|
+            yield(line.chomp("\0"))
+          end
+          raise "Find command failed" unless status_thread.value.success?
+        end
+      end
+
+      def build_find_command(search_dir)
+        cmd = ['find', search_dir, '-type', 'f', '!', '-path', "#{UPLOAD_DIR}/@hashed/*", '!', '-path', "#{UPLOAD_DIR}/tmp/*", '-print0']
+
+        ['ionice', '-c', 'Idle'] + cmd if ionice_is_available?
+
+        cmd
+      end
+
+      def ionice_is_available?
+        Gitlab::Utils.which('ionice')
+      rescue StandardError
+        # In this case, returning false is relatively safe, even though it isn't very nice
+        false
+      end
+
+      def insert_file_paths(file_paths)
+        file_paths.each do |file_path|
+          UnhashedUploadFile.create!(path: file_path)
+        end
       end
 
       def schedule_populate_untracked_uploads_jobs

spec/lib/gitlab/background_migration/prepare_unhashed_uploads_spec.rb

+require 'spec_helper'
+
+describe Gitlab::BackgroundMigration::PrepareUnhashedUploads, :migration, schema: 20171103140253 do
+  let!(:unhashed_upload_files) { table(:unhashed_upload_files) }
+
+  let(:user1) { create(:user) }
+  let(:user2) { create(:user) }
+  let(:project1) { create(:project) }
+  let(:project2) { create(:project) }
+  let(:appearance) { create(:appearance) }
+
+  context 'when files were uploaded before and after hashed storage was enabled' do
+    before do
+      fixture = Rails.root.join('spec', 'fixtures', 'rails_sample.jpg')
+      uploaded_file = fixture_file_upload(fixture)
+
+      user1.update(avatar: uploaded_file)
+      project1.update(avatar: uploaded_file)
+      appearance.update(logo: uploaded_file, header_logo: uploaded_file)
+      uploaded_file = fixture_file_upload(fixture)
+      UploadService.new(project1, uploaded_file, FileUploader).execute # Markdown upload
+
+      stub_application_setting(hashed_storage_enabled: true)
+
+      # Hashed files
+      uploaded_file = fixture_file_upload(fixture)
+      UploadService.new(project2, uploaded_file, FileUploader).execute
+    end
+
+    it 'adds unhashed files to the unhashed_upload_files table' do
+      expect do
+        described_class.new.perform
+      end.to change { unhashed_upload_files.count }.from(0).to(5)
+    end
+
+    it 'does not add hashed files to the unhashed_upload_files table' do
+      described_class.new.perform
+
+      hashed_file_path = project2.uploads.where(uploader: 'FileUploader').first.path
+      expect(unhashed_upload_files.where("path like '%#{hashed_file_path}%'").exists?).to be_falsey
+    end
+
+    # E.g. from a previous failed run of this background migration
+    context 'when there is existing data in unhashed_upload_files' do
+      before do
+        unhashed_upload_files.create(path: '/foo/bar.jpg')
+      end
+
+      it 'clears existing data before adding new data' do
+        expect do
+          described_class.new.perform
+        end.to change { unhashed_upload_files.count }.from(1).to(5)
+      end
+    end
+
+    # E.g. The installation is in use at the time of migration, and someone has
+    # just uploaded a file
+    context 'when there are files in /uploads/tmp' do
+      before do
+        FileUtils.touch(Rails.root.join(described_class::UPLOAD_DIR, 'tmp', 'some_file.jpg'))
+      end
+
+      it 'does not add files from /uploads/tmp' do
+        expect do
+          described_class.new.perform
+        end.to change { unhashed_upload_files.count }.from(0).to(5)
+      end
+    end
+  end
+
+  # Very new or lightly-used installations that are running this migration
+  # may not have an upload directory because they have no uploads.
+  context 'when no files were ever uploaded' do
+    it 'does not add to the unhashed_upload_files table (and does not raise error)' do
+      expect do
+        described_class.new.perform
+      end.not_to change { unhashed_upload_files.count }.from(0)
+    end
+  end
+end