Merge branch 'detect-orphaned-repositories' into 'master'

changelogs/unreleased/detect-orphaned-repositories.yml

+---
+title: Scripts to detect orphaned repositories
+merge_request: 14204
+author:
+type: added

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 |storage_name, repository_storage|
+          $stdout.puts
+          $stdout.puts "* Storage: #{storage_name} (#{repository_storage['path']})".color(:yellow)
+          toplevel_namespace_dirs = disk_namespaces(repository_storage['path'])
+
+          orphans = (toplevel_namespace_dirs - existing_namespaces)
+          print_orphans(orphans, storage_name)
+        end
+
+        clear_namespaces! # releases memory when check finishes
+      end
+
+      private
+
+      def print_orphans(orphans, storage_name)
+        if orphans.empty?
+          $stdout.puts "* No orphaned namespaces for #{storage_name} storage".color(:green)
+          return
+        end
+
+        orphans.each do |orphan|
+          $stdout.puts " - #{orphan}".color(:red)
+        end
+      end
+
+      def disk_namespaces(storage_path)
+        fetch_disk_namespaces(storage_path).each_with_object([]) do |namespace_path, result|
+          namespace = File.basename(namespace_path)
+          next if namespace.eql?('@hashed')
+
+          result << namespace
+        end
+      end
+
+      def fetch_disk_namespaces(storage_path)
+        Dir.glob(File.join(storage_path, '*'))
+      end
+
+      def existing_namespaces
+        @namespaces ||= Namespace.where(parent: nil).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:'
+      attr_accessor :orphans
+
+      def multi_check
+        Gitlab.config.repositories.storages.each do |storage_name, repository_storage|
+          $stdout.puts
+          $stdout.puts "* Storage: #{storage_name} (#{repository_storage['path']})".color(:yellow)
+
+          repositories = disk_repositories(repository_storage['path'])
+          orphans = (repositories - fetch_repositories(storage_name))
+
+          print_orphans(orphans, storage_name)
+        end
+      end
+
+      private
+
+      def print_orphans(orphans, storage_name)
+        if orphans.empty?
+          $stdout.puts "* No orphaned repositories for #{storage_name} storage".color(:green)
+          return
+        end
+
+        orphans.each do |orphan|
+          $stdout.puts " - #{orphan}".color(:red)
+        end
+      end
+
+      def disk_repositories(storage_path)
+        fetch_disk_namespaces(storage_path).each_with_object([]) do |namespace_path, result|
+          namespace = File.basename(namespace_path)
+          next if namespace.eql?('@hashed')
+
+          fetch_disk_repositories(namespace_path).each do |repo|
+            result << "#{namespace}/#{File.basename(repo)}"
+          end
+        end
+      end
+
+      def fetch_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 AND
+                  n.parent_id IS NULL)
+          WHERE (p.repository_storage LIKE ?)
+        "
+
+        query = ActiveRecord::Base.send(:sanitize_sql_array, [sql, storage_name]) # rubocop:disable GitlabSecurity/PublicSend
+        ActiveRecord::Base.connection.select_all(query).rows.try(:flatten!) || []
+      end
+
+      def fetch_disk_namespaces(storage_path)
+        Dir.glob(File.join(storage_path, '*'))
+      end
+
+      def fetch_disk_repositories(namespace_path)
+        Dir.glob(File.join(namespace_path, '*'))
+      end
+    end
+  end
+end

lib/tasks/gitlab/check.rake

@@ -398,6 +398,35 @@ 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|

spec/lib/system_check/orphans/namespace_check_spec.rb

+require 'spec_helper'
+require 'rake_helper'
+
+describe SystemCheck::Orphans::NamespaceCheck do
+  let(:storages) { Gitlab.config.repositories.storages.reject { |key, _| key.eql? 'broken' } }
+
+  before do
+    allow(Gitlab.config.repositories).to receive(:storages).and_return(storages)
+    allow(subject).to receive(:fetch_disk_namespaces).and_return(disk_namespaces)
+    silence_output
+  end
+
+  describe '#multi_check' do
+    context 'all orphans' do
+      let(:disk_namespaces) { %w(/repos/orphan1 /repos/orphan2 repos/@hashed) }
+
+      it 'prints list of all orphaned namespaces except @hashed' do
+        expect_list_of_orphans(%w(orphan1 orphan2))
+
+        subject.multi_check
+      end
+    end
+
+    context 'few orphans with existing namespace' do
+      let!(:first_level) { create(:group, path: 'my-namespace') }
+      let(:disk_namespaces) { %w(/repos/orphan1 /repos/orphan2 /repos/my-namespace /repos/@hashed) }
+
+      it 'prints list of orphaned namespaces' do
+        expect_list_of_orphans(%w(orphan1 orphan2))
+
+        subject.multi_check
+      end
+    end
+
+    context 'few orphans with existing namespace and parents with same name as orphans' do
+      let!(:first_level) { create(:group, path: 'my-namespace') }
+      let!(:second_level) { create(:group, path: 'second-level', parent: first_level) }
+      let(:disk_namespaces) { %w(/repos/orphan1 /repos/orphan2 /repos/my-namespace /repos/second-level /repos/@hashed) }
+
+      it 'prints list of orphaned namespaces ignoring parents with same namespace as orphans' do
+        expect_list_of_orphans(%w(orphan1 orphan2 second-level))
+
+        subject.multi_check
+      end
+    end
+
+    context 'no orphans' do
+      let(:disk_namespaces) { %w(@hashed) }
+
+      it 'prints an empty list ignoring @hashed' do
+        expect_list_of_orphans([])
+
+        subject.multi_check
+      end
+    end
+  end
+
+  def expect_list_of_orphans(orphans)
+    expect(subject).to receive(:print_orphans).with(orphans, 'default')
+  end
+end

spec/lib/system_check/orphans/repository_check_spec.rb

+require 'spec_helper'
+require 'rake_helper'
+
+describe SystemCheck::Orphans::RepositoryCheck do
+  let(:storages) { Gitlab.config.repositories.storages.reject { |key, _| key.eql? 'broken' } }
+
+  before do
+    allow(Gitlab.config.repositories).to receive(:storages).and_return(storages)
+    allow(subject).to receive(:fetch_disk_namespaces).and_return(disk_namespaces)
+    allow(subject).to receive(:fetch_disk_repositories).and_return(disk_repositories)
+    # silence_output
+  end
+
+  describe '#multi_check' do
+    context 'all orphans' do
+      let(:disk_namespaces) { %w(/repos/orphan1 /repos/orphan2 repos/@hashed) }
+      let(:disk_repositories) { %w(repo1.git repo2.git) }
+
+      it 'prints list of all orphaned namespaces except @hashed' do
+        expect_list_of_orphans(%w(orphan1/repo1.git orphan1/repo2.git orphan2/repo1.git orphan2/repo2.git))
+
+        subject.multi_check
+      end
+    end
+
+    context 'few orphans with existing namespace' do
+      let!(:first_level) { create(:group, path: 'my-namespace') }
+      let!(:project) { create(:project, path: 'repo', namespace: first_level) }
+      let(:disk_namespaces) { %w(/repos/orphan1 /repos/orphan2 /repos/my-namespace /repos/@hashed) }
+      let(:disk_repositories) { %w(repo.git) }
+
+      it 'prints list of orphaned namespaces' do
+        expect_list_of_orphans(%w(orphan1/repo.git orphan2/repo.git))
+
+        subject.multi_check
+      end
+    end
+
+    context 'few orphans with existing namespace and parents with same name as orphans' do
+      let!(:first_level) { create(:group, path: 'my-namespace') }
+      let!(:second_level) { create(:group, path: 'second-level', parent: first_level) }
+      let!(:project) { create(:project, path: 'repo', namespace: first_level) }
+      let(:disk_namespaces) { %w(/repos/orphan1 /repos/orphan2 /repos/my-namespace /repos/second-level /repos/@hashed) }
+      let(:disk_repositories) { %w(repo.git) }
+
+      it 'prints list of orphaned namespaces ignoring parents with same namespace as orphans' do
+        expect_list_of_orphans(%w(orphan1/repo.git orphan2/repo.git second-level/repo.git))
+
+        subject.multi_check
+      end
+    end
+
+    context 'no orphans' do
+      let(:disk_namespaces) { %w(@hashed) }
+      let(:disk_repositories) { %w(repo.git) }
+
+      it 'prints an empty list ignoring @hashed' do
+        expect_list_of_orphans([])
+
+        subject.multi_check
+      end
+    end
+  end
+
+  def expect_list_of_orphans(orphans)
+    expect(subject).to receive(:print_orphans).with(orphans, 'default')
+  end
+end