Use find instead Ruby to remove files due to performance reasons

2 files changed, 12 insertions, 32 deletions

diff --git a/app/services/repository_archive_clean_up_service.rb b/app/services/repository_archive_clean_up_service.rb
index 9764df6492d..0b56b09738d 100644
--- a/app/services/repository_archive_clean_up_service.rb
+++ b/app/services/repository_archive_clean_up_service.rb
@@ -1,5 +1,4 @@
 class RepositoryArchiveCleanUpService
-  ALLOWED_ARCHIVE_EXTENSIONS = %w[tar tar.bz2 tar.gz zip].join(',').freeze
   LAST_MODIFIED_TIME_IN_MINUTES = 120
 
   def initialize(mmin = LAST_MODIFIED_TIME_IN_MINUTES)
@@ -21,22 +20,14 @@ class RepositoryArchiveCleanUpService
   attr_reader :mmin, :path
 
   def clean_up_old_archives
-    Dir.glob("#{path}/**.git/*{#{ALLOWED_ARCHIVE_EXTENSIONS}}") do |filename|
-      File.delete(filename) if older?(filename)
-    end
-  end
-
-  def older?(filename)
-    File.exist?(filename) && File.new(filename).mtime < (Time.now - mmin * 60)
+    run(%W(find #{path} -not -path #{path} -type f \( -name \*.tar -o -name \*.bz2 -o -name \*.tar.gz -o -name \*.zip \) -maxdepth 2 -mmin +#{mmin} -delete))
   end
 
   def clean_up_empty_directories
-    Dir.glob("#{path}/**.git/").reverse_each do |dir|
-      Dir.rmdir(dir) if empty?(dir)
-    end
+    run(%W(find #{path} -not -path #{path} -type d -empty -name \*.git -maxdepth 1 -delete))
   end
 
-  def empty?(dir)
-    (Dir.entries(dir) - %w[. ..]).empty?
+  def run(cmd)
+    Gitlab::Popen.popen(cmd)
   end
 end
diff --git a/spec/services/repository_archive_clean_up_service_spec.rb b/spec/services/repository_archive_clean_up_service_spec.rb
index 321cc6daf9d..6173d6cb51c 100644
--- a/spec/services/repository_archive_clean_up_service_spec.rb
+++ b/spec/services/repository_archive_clean_up_service_spec.rb
@@ -30,13 +30,9 @@ describe RepositoryArchiveCleanUpService, services: true do
       end
 
       context 'when archives older than 2 hours exists' do
-        before do
-          allow_any_instance_of(File).to receive(:mtime).and_return(2.hours.ago)
-        end
-
         it 'removes old files that matches valid archive extensions' do
           dirname = File.join(path, 'sample.git')
-          files = create_temporary_files(dirname, %w[tar tar.bz2 tar.gz zip])
+          files = create_temporary_files(dirname, %w[tar tar.bz2 tar.gz zip], 2.hours)
 
           service.execute
 
@@ -46,7 +42,7 @@ describe RepositoryArchiveCleanUpService, services: true do
 
         it 'keeps old files that does not matches valid archive extensions' do
           dirname = File.join(path, 'sample.git')
-          files = create_temporary_files(dirname, %w[conf rb])
+          files = create_temporary_files(dirname, %w[conf rb], 2.hours)
 
           service.execute
 
@@ -56,7 +52,7 @@ describe RepositoryArchiveCleanUpService, services: true do
 
         it 'keeps old files inside invalid directories' do
           dirname = File.join(path, 'john_doe/sample.git')
-          files = create_temporary_files(dirname, %w[conf rb tar tar.gz])
+          files = create_temporary_files(dirname, %w[conf rb tar tar.gz], 2.hours)
 
           service.execute
 
@@ -66,13 +62,9 @@ describe RepositoryArchiveCleanUpService, services: true do
       end
 
       context 'when archives older than 2 hours does not exist' do
-        before do
-          allow_any_instance_of(File).to receive(:mtime).and_return(1.hour.ago)
-        end
-
         it 'keeps files that matches valid archive extensions' do
           dirname = File.join(path, 'sample.git')
-          files = create_temporary_files(dirname, %w[tar tar.bz2 tar.gz zip])
+          files = create_temporary_files(dirname, %w[tar tar.bz2 tar.gz zip], 1.hour)
 
           service.execute
 
@@ -82,7 +74,7 @@ describe RepositoryArchiveCleanUpService, services: true do
 
         it 'keeps files that does not matches valid archive extensions' do
           dirname = File.join(path, 'sample.git')
-          files = create_temporary_files(dirname, %w[conf rb])
+          files = create_temporary_files(dirname, %w[conf rb], 1.hour)
 
           service.execute
 
@@ -92,7 +84,7 @@ describe RepositoryArchiveCleanUpService, services: true do
 
         it 'keeps files inside invalid directories' do
           dirname = File.join(path, 'john_doe/sample.git')
-          files = create_temporary_files(dirname, %w[conf rb tar tar.gz])
+          files = create_temporary_files(dirname, %w[conf rb tar tar.gz], 1.hour)
 
           service.execute
 
@@ -101,12 +93,9 @@ describe RepositoryArchiveCleanUpService, services: true do
         end
       end
 
-      def create_temporary_files(dirname, extensions)
+      def create_temporary_files(dirname, extensions, mtime)
         FileUtils.mkdir_p(dirname)
-
-        extensions.flat_map do |extension|
-          FileUtils.touch(File.join(dirname, "sample.#{extension}"))
-        end
+        FileUtils.touch(extensions.map { |ext| File.join(dirname, "sample.#{ext}") }, mtime: Time.now - mtime)
       end
     end
   end