Merge branch 'mk-fix-no-untracked-upload-files-error' into 'master'

Resolve "PrepareUntrackedUploads PostgreSQL syntax error"

Closes #42881

See merge request gitlab-org/gitlab-ce!17019

changelogs/unreleased/mk-fix-no-untracked-upload-files-error.yml

+---
+title: Resolve PrepareUntrackedUploads PostgreSQL syntax error
+merge_request: 17019
+author:
+type: fixed

db/migrate/20180208183958_schedule_populate_untracked_uploads_if_needed.rb

+# See http://doc.gitlab.com/ce/development/migration_style_guide.html
+# for more information on how to write migrations for GitLab.
+
+class SchedulePopulateUntrackedUploadsIfNeeded < ActiveRecord::Migration
+  include Gitlab::Database::MigrationHelpers
+
+  DOWNTIME = false
+
+  FOLLOW_UP_MIGRATION = 'PopulateUntrackedUploads'.freeze
+
+  class UntrackedFile < ActiveRecord::Base
+    include EachBatch
+
+    self.table_name = 'untracked_files_for_uploads'
+  end
+
+  def up
+    if table_exists?(:untracked_files_for_uploads)
+      process_or_remove_table
+    end
+  end
+
+  def down
+    # nothing
+  end
+
+  private
+
+  def process_or_remove_table
+    if UntrackedFile.all.empty?
+      drop_temp_table
+    else
+      schedule_populate_untracked_uploads_jobs
+    end
+  end
+
+  def drop_temp_table
+    drop_table(:untracked_files_for_uploads, if_exists: true)
+  end
+
+  def schedule_populate_untracked_uploads_jobs
+    say "Scheduling #{FOLLOW_UP_MIGRATION} background migration jobs since there are rows in untracked_files_for_uploads."
+
+    bulk_queue_background_migration_jobs_by_range(
+      UntrackedFile, FOLLOW_UP_MIGRATION)
+  end
+end

db/schema.rb

@@ -11,7 +11,7 @@
 #
 # It's strongly recommended that you check this file into your version control system.
 
-ActiveRecord::Schema.define(version: 20180206200543) do
+ActiveRecord::Schema.define(version: 20180208183958) do
 
   # These are extensions that must be enabled in order to support this database
   enable_extension "plpgsql"

lib/gitlab/background_migration/prepare_untracked_uploads.rb

@@ -43,8 +43,12 @@ module Gitlab
 
         store_untracked_file_paths
 
+        if UntrackedFile.all.empty?
+          drop_temp_table
+        else
           schedule_populate_untracked_uploads_jobs
         end
+      end
 
       private
 
@@ -92,7 +96,7 @@ module Gitlab
           end
         end
 
-        yield(paths)
+        yield(paths) if paths.any?
       end
 
       def build_find_command(search_dir)
@@ -165,6 +169,11 @@ module Gitlab
         bulk_queue_background_migration_jobs_by_range(
           UntrackedFile, FOLLOW_UP_MIGRATION)
       end
+
+      def drop_temp_table
+        UntrackedFile.connection.drop_table(:untracked_files_for_uploads,
+                                            if_exists: true)
+      end
     end
   end
 end

spec/lib/gitlab/background_migration/prepare_untracked_uploads_spec.rb

@@ -8,8 +8,6 @@ describe Gitlab::BackgroundMigration::PrepareUntrackedUploads, :sidekiq do
 
   before do
     DatabaseCleaner.clean
-
-    drop_temp_table_if_exists
   end
 
   after do
@@ -23,31 +21,21 @@ describe Gitlab::BackgroundMigration::PrepareUntrackedUploads, :sidekiq do
     end
   end
 
-  # E.g. The installation is in use at the time of migration, and someone has
-  # just uploaded a file
-  shared_examples 'does not add files in /uploads/tmp' do
-    let(:tmp_file) { Rails.root.join(described_class::ABSOLUTE_UPLOAD_DIR, 'tmp', 'some_file.jpg') }
+  shared_examples 'prepares the untracked_files_for_uploads table' do
+    context 'when files were uploaded before and after hashed storage was enabled' do
+      let!(:appearance) { create_or_update_appearance(logo: uploaded_file, header_logo: uploaded_file) }
+      let!(:user) { create(:user, :with_avatar) }
+      let!(:project1) { create(:project, :with_avatar, :legacy_storage) }
+      let(:project2) { create(:project) } # instantiate after enabling hashed_storage
 
       before do
-      FileUtils.mkdir(File.dirname(tmp_file))
-      FileUtils.touch(tmp_file)
-    end
-
-    after do
-      FileUtils.rm(tmp_file)
-    end
-
-    it 'does not add files from /uploads/tmp' do
-      described_class.new.perform
+        # Markdown upload before enabling hashed_storage
+        UploadService.new(project1, uploaded_file, FileUploader).execute
 
-      expect(untracked_files_for_uploads.count).to eq(5)
-    end
-  end
+        stub_application_setting(hashed_storage_enabled: true)
 
-  it 'ensures the untracked_files_for_uploads table exists' do
-    expect do
-      described_class.new.perform
-    end.to change { ActiveRecord::Base.connection.table_exists?(:untracked_files_for_uploads) }.from(false).to(true)
+        # Markdown upload after enabling hashed_storage
+        UploadService.new(project2, uploaded_file, FileUploader).execute
       end
 
       it 'has a path field long enough for really long paths' do
@@ -65,31 +53,6 @@ describe Gitlab::BackgroundMigration::PrepareUntrackedUploads, :sidekiq do
         expect(record.reload.path.size).to eq(519)
       end
 
-  context "test bulk insert with ON CONFLICT DO NOTHING or IGNORE" do
-    around do |example|
-      # If this is CI, we use Postgres 9.2 so this whole context should be
-      # skipped since we're unable to use ON CONFLICT DO NOTHING or IGNORE.
-      if described_class.new.send(:can_bulk_insert_and_ignore_duplicates?)
-        example.run
-      end
-    end
-
-    context 'when files were uploaded before and after hashed storage was enabled' do
-      let!(:appearance) { create_or_update_appearance(logo: uploaded_file, header_logo: uploaded_file) }
-      let!(:user) { create(:user, :with_avatar) }
-      let!(:project1) { create(:project, :with_avatar, :legacy_storage) }
-      let(:project2) { create(:project) } # instantiate after enabling hashed_storage
-
-      before do
-        # Markdown upload before enabling hashed_storage
-        UploadService.new(project1, uploaded_file, FileUploader).execute
-
-        stub_application_setting(hashed_storage_enabled: true)
-
-        # Markdown upload after enabling hashed_storage
-        UploadService.new(project2, uploaded_file, FileUploader).execute
-      end
-
       it 'adds unhashed files to the untracked_files_for_uploads table' do
         described_class.new.perform
 
@@ -130,91 +93,66 @@ describe Gitlab::BackgroundMigration::PrepareUntrackedUploads, :sidekiq do
         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
-        it_behaves_like 'does not add files in /uploads/tmp'
-      end
-    end
-  end
+        let(:tmp_file) { Rails.root.join(described_class::ABSOLUTE_UPLOAD_DIR, 'tmp', 'some_file.jpg') }
 
-  context 'test bulk insert without ON CONFLICT DO NOTHING or IGNORE' do
         before do
-      # If this is CI, we use Postgres 9.2 so this stub has no effect.
-      #
-      # If this is being run on Postgres 9.5+ or MySQL, then this stub allows us
-      # to test the bulk insert functionality without ON CONFLICT DO NOTHING or
-      # IGNORE.
-      allow_any_instance_of(described_class).to receive(:postgresql_pre_9_5?).and_return(true)
+          FileUtils.mkdir(File.dirname(tmp_file))
+          FileUtils.touch(tmp_file)
         end
 
-    context 'when files were uploaded before and after hashed storage was enabled' do
-      let!(:appearance) { create_or_update_appearance(logo: uploaded_file, header_logo: uploaded_file) }
-      let!(:user) { create(:user, :with_avatar) }
-      let!(:project1) { create(:project, :with_avatar, :legacy_storage) }
-      let(:project2) { create(:project) } # instantiate after enabling hashed_storage
-
-      before do
-        # Markdown upload before enabling hashed_storage
-        UploadService.new(project1, uploaded_file, FileUploader).execute
-
-        stub_application_setting(hashed_storage_enabled: true)
-
-        # Markdown upload after enabling hashed_storage
-        UploadService.new(project2, uploaded_file, FileUploader).execute
+        after do
+          FileUtils.rm(tmp_file)
         end
 
-      it 'adds unhashed files to the untracked_files_for_uploads table' do
+        it 'does not add files from /uploads/tmp' do
           described_class.new.perform
 
           expect(untracked_files_for_uploads.count).to eq(5)
         end
-
-      it 'adds files with paths relative to CarrierWave.root' do
-        described_class.new.perform
-        untracked_files_for_uploads.all.each do |file|
-          expect(file.path.start_with?('uploads/')).to be_truthy
-        end
       end
 
-      it 'does not add hashed files to the untracked_files_for_uploads table' do
-        described_class.new.perform
+      context 'when the last batch size exactly matches the max batch size' do
+        it 'does not raise error' do
+          stub_const("#{described_class}::FIND_BATCH_SIZE", 5)
 
-        hashed_file_path = project2.uploads.where(uploader: 'FileUploader').first.path
-        expect(untracked_files_for_uploads.where("path like '%#{hashed_file_path}%'").exists?).to be_falsey
-      end
-
-      it 'correctly schedules the follow-up background migration jobs' do
+          expect do
             described_class.new.perform
+          end.not_to raise_error
 
-        expect(described_class::FOLLOW_UP_MIGRATION).to be_scheduled_migration(1, 5)
-        expect(BackgroundMigrationWorker.jobs.size).to eq(1)
+          expect(untracked_files_for_uploads.count).to eq(5)
         end
-
-      # E.g. from a previous failed run of this background migration
-      context 'when there is existing data in untracked_files_for_uploads' do
-        before do
-          described_class.new.perform
       end
-
-        it 'does not error or produce duplicates of existing data' do
-          expect do
-            described_class.new.perform
-          end.not_to change { untracked_files_for_uploads.count }.from(5)
     end
   end
 
-      context 'when there are files in /uploads/tmp' do
-        it_behaves_like 'does not add files in /uploads/tmp'
+  # If running on Postgres 9.2 (like on CI), this whole context is skipped
+  # since we're unable to use ON CONFLICT DO NOTHING or IGNORE.
+  context "test bulk insert with ON CONFLICT DO NOTHING or IGNORE", if: described_class.new.send(:can_bulk_insert_and_ignore_duplicates?) do
+    it_behaves_like 'prepares the untracked_files_for_uploads table'
   end
+
+  # If running on Postgres 9.2 (like on CI), the stubbed method has no effect.
+  #
+  # If running on Postgres 9.5+ or MySQL, then this context effectively tests
+  # the bulk insert functionality without ON CONFLICT DO NOTHING or IGNORE.
+  context 'test bulk insert without ON CONFLICT DO NOTHING or IGNORE' do
+    before do
+      allow_any_instance_of(described_class).to receive(:postgresql_pre_9_5?).and_return(true)
     end
+
+    it_behaves_like 'prepares the untracked_files_for_uploads table'
   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 untracked_files_for_uploads table (and does not raise error)' do
+    it 'deletes the `untracked_files_for_uploads` table (and does not raise error)' do
       described_class.new.perform
 
-      expect(untracked_files_for_uploads.count).to eq(0)
+      expect(untracked_files_for_uploads.connection.table_exists?(:untracked_files_for_uploads)).to be_falsey
     end
   end
 end