Detect orphaned repositories and namespaces in any storage

lib/system_check/orphans/namespace_check.rb

+module SystemCheck
+  module Orphans
+    class NamespaceCheck < SystemCheck::BaseCheck
+      set_name 'Orphaned namespaces:'
+
+      def multi_check
+        Gitlab.config.repositories.storages.each do |name, repository_storage|
+          $stdout.puts
+          $stdout.puts "* Storage: #{name} (#{repository_storage['path']})".color(:yellow)
+          toplevel_namespace_dirs = Dir.glob(File.join(repository_storage['path'], '*')).map{|p| File.basename(p)}
+
+          orphans = (toplevel_namespace_dirs - existing_namespaces)
+          if orphans.empty?
+            $stdout.puts "* No orphaned namespaces for #{name} storage".color(:green)
+            next
+          end
+
+          orphans.each do |orphan|
+            $stdout.puts " - #{orphan}".color(:red)
+          end
+        end
+
+        clear_namespaces! # releases memory when check finishes
+      end
+
+      private
+
+      def existing_namespaces
+        @namespaces ||= Namespace.all.pluck(:path)
+      end
+
+      def clear_namespaces!
+        @namespaces = nil
+      end
+    end
+  end
+end

lib/system_check/orphans/repository_check.rb

+module SystemCheck
+  module Orphans
+    class RepositoryCheck < SystemCheck::BaseCheck
+      set_name 'Orphaned repositories:'
+
+      def multi_check
+        Gitlab.config.repositories.storages.each do |name, repository_storage|
+          $stdout.puts
+          $stdout.puts "* Storage: #{name} (#{repository_storage['path']})".color(:yellow)
+
+          repositories = toplevel_namespace_dirs(repository_storage['path']).map do |path|
+            namespace = File.basename(path)
+            Dir.glob(File.join(path, '*')).map {|repo| "#{namespace}/#{File.basename(repo)}"}
+          end.try(:flatten!)
+
+          orphans = (repositories - list_repositories(name))
+          if orphans.empty?
+            $stdout.puts "* No orphaned repositories for #{name} storage".color(:green)
+            next
+          end
+
+          orphans.each do |orphan|
+            $stdout.puts " - #{orphan}".color(:red)
+          end
+        end
+      end
+
+      private
+
+      def list_repositories(storage_name)
+        sql = "
+          SELECT
+            CONCAT(n.path, '/', p.path, '.git') repo,
+            CONCAT(n.path, '/', p.path, '.wiki.git') wiki
+          FROM projects p
+            JOIN namespaces n
+              ON (p.namespace_id = n.id)
+          WHERE (p.repository_storage LIKE ?)
+        "
+
+        query = ActiveRecord::Base.send(:sanitize_sql_array, [sql, storage_name])
+        ActiveRecord::Base.connection.select_all(query).rows.try(:flatten!)
+      end
+
+      def toplevel_namespace_dirs(storage_path)
+        Dir.glob(File.join(storage_path, '*'))
+      end
+    end
+  end
+end

lib/tasks/gitlab/check.rake

@@ -398,6 +398,36 @@ namespace :gitlab do
     end
   end
 
+  namespace :orphans do
+    desc 'Gitlab | Check for orphaned namespaces and repositories'
+    task check: :environment do
+      warn_user_is_not_gitlab
+      checks = [
+        SystemCheck::Orphans::NamespaceCheck,
+        SystemCheck::Orphans::RepositoryCheck
+      ]
+
+      SystemCheck.run('Orphans', checks)
+    end
+
+    desc 'GitLab | Check for orphaned namespaces in the repositories path'
+    task check_namespaces: :environment do
+      warn_user_is_not_gitlab
+      checks = [SystemCheck::Orphans::NamespaceCheck]
+
+      SystemCheck.run('Orphans', checks)
+    end
+
+    desc 'GitLab | Check for orphaned repositories in the repositories path'
+    task check_repositories: :environment do
+      warn_user_is_not_gitlab
+      checks = [SystemCheck::Orphans::RepositoryCheck]
+
+      SystemCheck.run('Orphans', checks)
+    end
+  end
+
+
   namespace :user do
     desc "GitLab | Check the integrity of a specific user's repositories"
     task :check_repos, [:username] => :environment do |t, args|