Merge branch 'fj-move-gitgarbagecollectworker-to-projects-namespace' into 'master'

Move GitGarbageCollectWorker to Projects namespace

See merge request gitlab-org/gitlab!52064

.rubocop_manual_todo.yml

@@ -1435,7 +1435,6 @@ RSpec/AnyInstanceOf:
     - 'spec/workers/emails_on_push_worker_spec.rb'
     - 'spec/workers/error_tracking_issue_link_worker_spec.rb'
     - 'spec/workers/expire_pipeline_cache_worker_spec.rb'
-    - 'spec/workers/git_garbage_collect_worker_spec.rb'
     - 'spec/workers/group_export_worker_spec.rb'
     - 'spec/workers/group_import_worker_spec.rb'
     - 'spec/workers/namespaceless_project_destroy_worker_spec.rb'

app/services/projects/cleanup_service.rb

@@ -40,7 +40,7 @@ module Projects
       apply_bfg_object_map!
 
       # Remove older objects that are no longer referenced
-      GitGarbageCollectWorker.new.perform(project.id, :prune, "project_cleanup:gc:#{project.id}")
+      Projects::GitGarbageCollectWorker.new.perform(project.id, :prune, "project_cleanup:gc:#{project.id}")
 
       # The cache may now be inaccurate, and holding onto it could prevent
       # bugs assuming the presence of some object from manifesting for some

app/services/repositories/housekeeping_service.rb

@@ -45,7 +45,7 @@ module Repositories
     private
 
     def execute_gitlab_shell_gc(lease_uuid)
-      GitGarbageCollectWorker.perform_async(@resource.id, task, lease_key, lease_uuid)
+      Projects::GitGarbageCollectWorker.perform_async(@resource.id, task, lease_key, lease_uuid)
     ensure
       if pushes_since_gc >= gc_period
         Gitlab::Metrics.measure(:reset_pushes_since_gc) do

app/workers/all_queues.yml

@@ -1999,6 +1999,14 @@
   :weight: 1
   :idempotent: true
   :tags: []
+- :name: projects_git_garbage_collect
+  :feature_category: :gitaly
+  :has_external_dependencies:
+  :urgency: :low
+  :resource_boundary: :unknown
+  :weight: 1
+  :idempotent:
+  :tags: []
 - :name: prometheus_create_default_alerts
   :feature_category: :incident_management
   :has_external_dependencies:

app/workers/git_garbage_collect_worker.rb

 # frozen_string_literal: true
 
+# According to our docs, we can only remove workers on major releases
+# https://docs.gitlab.com/ee/development/sidekiq_style_guide.html#removing-workers.
+#
+# We need to still maintain this until 14.0 but with the current functionality.
+#
+# In https://gitlab.com/gitlab-org/gitlab/-/issues/299290 we track that removal.
 class GitGarbageCollectWorker # rubocop:disable Scalability/IdempotentWorker
   include ApplicationWorker
 
@@ -7,117 +13,7 @@ class GitGarbageCollectWorker # rubocop:disable Scalability/IdempotentWorker
   feature_category :gitaly
   loggable_arguments 1, 2, 3
 
-  # Timeout set to 24h
-  LEASE_TIMEOUT = 86400
-
   def perform(project_id, task = :gc, lease_key = nil, lease_uuid = nil)
-    lease_key ||= "git_gc:#{task}:#{project_id}"
-    project = Project.find(project_id)
-    active_uuid = get_lease_uuid(lease_key)
-
-    if active_uuid
-      return unless active_uuid == lease_uuid
-
-      renew_lease(lease_key, active_uuid)
-    else
-      lease_uuid = try_obtain_lease(lease_key)
-
-      return unless lease_uuid
-    end
-
-    task = task.to_sym
-
-    if gc?(task)
-      ::Projects::GitDeduplicationService.new(project).execute
-      cleanup_orphan_lfs_file_references(project)
-    end
-
-    gitaly_call(task, project)
-
-    # Refresh the branch cache in case garbage collection caused a ref lookup to fail
-    flush_ref_caches(project) if gc?(task)
-
-    update_repository_statistics(project) if task != :pack_refs
-
-    # In case pack files are deleted, release libgit2 cache and open file
-    # descriptors ASAP instead of waiting for Ruby garbage collection
-    project.cleanup
-  ensure
-    cancel_lease(lease_key, lease_uuid) if lease_key.present? && lease_uuid.present?
-  end
-
-  private
-
-  def gc?(task)
-    task == :gc || task == :prune
-  end
-
-  def try_obtain_lease(key)
-    ::Gitlab::ExclusiveLease.new(key, timeout: LEASE_TIMEOUT).try_obtain
-  end
-
-  def renew_lease(key, uuid)
-    ::Gitlab::ExclusiveLease.new(key, uuid: uuid, timeout: LEASE_TIMEOUT).renew
-  end
-
-  def cancel_lease(key, uuid)
-    ::Gitlab::ExclusiveLease.cancel(key, uuid)
-  end
-
-  def get_lease_uuid(key)
-    ::Gitlab::ExclusiveLease.get_uuid(key)
-  end
-
-  def gitaly_call(task, project)
-    repository = project.repository.raw_repository
-
-    client = if task == :pack_refs
-               Gitlab::GitalyClient::RefService.new(repository)
-             else
-               Gitlab::GitalyClient::RepositoryService.new(repository)
-             end
-
-    case task
-    when :prune, :gc
-      client.garbage_collect(bitmaps_enabled?, prune: task == :prune)
-    when :full_repack
-      client.repack_full(bitmaps_enabled?)
-    when :incremental_repack
-      client.repack_incremental
-    when :pack_refs
-      client.pack_refs
-    end
-  rescue GRPC::NotFound => e
-    Gitlab::GitLogger.error("#{__method__} failed:\nRepository not found")
-    raise Gitlab::Git::Repository::NoRepository.new(e)
-  rescue GRPC::BadStatus => e
-    Gitlab::GitLogger.error("#{__method__} failed:\n#{e}")
-    raise Gitlab::Git::CommandError.new(e)
-  end
-
-  def cleanup_orphan_lfs_file_references(project)
-    return if Gitlab::Database.read_only? # GitGarbageCollectWorker may be run on a Geo secondary
-
-    ::Gitlab::Cleanup::OrphanLfsFileReferences.new(project, dry_run: false, logger: logger).run!
-  rescue => err
-    Gitlab::GitLogger.warn(message: "Cleaning up orphan LFS objects files failed", error: err.message)
-    Gitlab::ErrorTracking.track_and_raise_for_dev_exception(err)
-  end
-
-  def flush_ref_caches(project)
-    project.repository.expire_branches_cache
-    project.repository.branch_names
-    project.repository.has_visible_content?
-  end
-
-  def update_repository_statistics(project)
-    project.repository.expire_statistics_caches
-    return if Gitlab::Database.read_only? # GitGarbageCollectWorker may be run on a Geo secondary
-
-    Projects::UpdateStatisticsService.new(project, nil, statistics: [:repository_size, :lfs_objects_size]).execute
-  end
-
-  def bitmaps_enabled?
-    Gitlab::CurrentSettings.housekeeping_bitmaps_enabled
+    ::Projects::GitGarbageCollectWorker.new.perform(project_id, task, lease_key, lease_uuid)
   end
 end

app/workers/projects/git_garbage_collect_worker.rb

+# frozen_string_literal: true
+
+module Projects
+  class GitGarbageCollectWorker # rubocop:disable Scalability/IdempotentWorker
+    include ApplicationWorker
+
+    sidekiq_options retry: false
+    feature_category :gitaly
+    loggable_arguments 1, 2, 3
+
+    # Timeout set to 24h
+    LEASE_TIMEOUT = 86400
+
+    def perform(project_id, task = :gc, lease_key = nil, lease_uuid = nil)
+      lease_key ||= "git_gc:#{task}:#{project_id}"
+      project = find_project(project_id)
+      active_uuid = get_lease_uuid(lease_key)
+
+      if active_uuid
+        return unless active_uuid == lease_uuid
+
+        renew_lease(lease_key, active_uuid)
+      else
+        lease_uuid = try_obtain_lease(lease_key)
+
+        return unless lease_uuid
+      end
+
+      task = task.to_sym
+
+      if gc?(task)
+        ::Projects::GitDeduplicationService.new(project).execute
+        cleanup_orphan_lfs_file_references(project)
+      end
+
+      gitaly_call(task, project)
+
+      # Refresh the branch cache in case garbage collection caused a ref lookup to fail
+      flush_ref_caches(project) if gc?(task)
+
+      update_repository_statistics(project) if task != :pack_refs
+
+      # In case pack files are deleted, release libgit2 cache and open file
+      # descriptors ASAP instead of waiting for Ruby garbage collection
+      project.cleanup
+    ensure
+      cancel_lease(lease_key, lease_uuid) if lease_key.present? && lease_uuid.present?
+    end
+
+    private
+
+    def find_project(project_id)
+      Project.find(project_id)
+    end
+
+    def gc?(task)
+      task == :gc || task == :prune
+    end
+
+    def try_obtain_lease(key)
+      ::Gitlab::ExclusiveLease.new(key, timeout: LEASE_TIMEOUT).try_obtain
+    end
+
+    def renew_lease(key, uuid)
+      ::Gitlab::ExclusiveLease.new(key, uuid: uuid, timeout: LEASE_TIMEOUT).renew
+    end
+
+    def cancel_lease(key, uuid)
+      ::Gitlab::ExclusiveLease.cancel(key, uuid)
+    end
+
+    def get_lease_uuid(key)
+      ::Gitlab::ExclusiveLease.get_uuid(key)
+    end
+
+    def gitaly_call(task, project)
+      repository = project.repository.raw_repository
+      client = get_gitaly_client(task, repository)
+
+      case task
+      when :prune, :gc
+        client.garbage_collect(bitmaps_enabled?, prune: task == :prune)
+      when :full_repack
+        client.repack_full(bitmaps_enabled?)
+      when :incremental_repack
+        client.repack_incremental
+      when :pack_refs
+        client.pack_refs
+      end
+    rescue GRPC::NotFound => e
+      Gitlab::GitLogger.error("#{__method__} failed:\nRepository not found")
+      raise Gitlab::Git::Repository::NoRepository.new(e)
+    rescue GRPC::BadStatus => e
+      Gitlab::GitLogger.error("#{__method__} failed:\n#{e}")
+      raise Gitlab::Git::CommandError.new(e)
+    end
+
+    def get_gitaly_client(task, repository)
+      if task == :pack_refs
+        Gitlab::GitalyClient::RefService
+      else
+        Gitlab::GitalyClient::RepositoryService
+      end.new(repository)
+    end
+
+    def cleanup_orphan_lfs_file_references(project)
+      return if Gitlab::Database.read_only? # GitGarbageCollectWorker may be run on a Geo secondary
+
+      ::Gitlab::Cleanup::OrphanLfsFileReferences.new(project, dry_run: false, logger: logger).run!
+    rescue => err
+      Gitlab::GitLogger.warn(message: "Cleaning up orphan LFS objects files failed", error: err.message)
+      Gitlab::ErrorTracking.track_and_raise_for_dev_exception(err)
+    end
+
+    def flush_ref_caches(project)
+      project.repository.expire_branches_cache
+      project.repository.branch_names
+      project.repository.has_visible_content?
+    end
+
+    def update_repository_statistics(project)
+      project.repository.expire_statistics_caches
+      return if Gitlab::Database.read_only? # GitGarbageCollectWorker may be run on a Geo secondary
+
+      Projects::UpdateStatisticsService.new(project, nil, statistics: [:repository_size, :lfs_objects_size]).execute
+    end
+
+    def bitmaps_enabled?
+      Gitlab::CurrentSettings.housekeeping_bitmaps_enabled
+    end
+  end
+end

config/sidekiq_queues.yml

@@ -270,6 +270,8 @@
   - 1
 - - project_update_repository_storage
   - 1
+- - projects_git_garbage_collect
+  - 1
 - - prometheus_create_default_alerts
   - 1
 - - propagate_integration

doc/administration/housekeeping.md

@@ -49,7 +49,7 @@ It is ultimately performed by the Gitaly RPC `FetchIntoObjectPool`.
 This is the current call stack by which it is invoked:
 
 1. `Repositories::HousekeepingService#execute_gitlab_shell_gc`
-1. `GitGarbageCollectWorker#perform`
+1. `Projects::GitGarbageCollectWorker#perform`
 1. `Projects::GitDeduplicationService#fetch_from_source`
 1. `ObjectPool#fetch`
 1. `ObjectPoolService#fetch`

ee/app/services/geo/project_housekeeping_service.rb

@@ -52,7 +52,7 @@ module Geo
     end
 
     def execute_gitlab_shell_gc(lease_uuid)
-      GitGarbageCollectWorker.perform_async(project.id, task, lease_key, lease_uuid)
+      ::Projects::GitGarbageCollectWorker.perform_async(project.id, task, lease_key, lease_uuid)
     ensure
       if should_reset?
         Gitlab::Metrics.measure(:geo_reset_syncs_since_gc) do

ee/spec/services/geo/project_housekeeping_service_spec.rb

@@ -52,13 +52,13 @@ RSpec.describe Geo::ProjectHousekeepingService do
         stub_exclusive_lease(:the_lease_key, :the_uuid)
 
         # At fetch 200
-        expect(GitGarbageCollectWorker).to receive(:perform_async).with(project.id, :gc, :the_lease_key, :the_uuid)
+        expect(::Projects::GitGarbageCollectWorker).to receive(:perform_async).with(project.id, :gc, :the_lease_key, :the_uuid)
           .once
         # At fetch 50, 100, 150
-        expect(GitGarbageCollectWorker).to receive(:perform_async).with(project.id, :full_repack, :the_lease_key, :the_uuid)
+        expect(::Projects::GitGarbageCollectWorker).to receive(:perform_async).with(project.id, :full_repack, :the_lease_key, :the_uuid)
           .exactly(3).times
         # At fetch 10, 20, ... (except those above)
-        expect(GitGarbageCollectWorker).to receive(:perform_async).with(project.id, :incremental_repack, :the_lease_key, :the_uuid)
+        expect(::Projects::GitGarbageCollectWorker).to receive(:perform_async).with(project.id, :incremental_repack, :the_lease_key, :the_uuid)
           .exactly(16).times
 
         201.times do
@@ -76,7 +76,7 @@ RSpec.describe Geo::ProjectHousekeepingService do
         allow(service).to receive(:lease_key).and_return(:the_lease_key)
         stub_exclusive_lease(:the_lease_key, :the_uuid)
 
-        expect(GitGarbageCollectWorker).to receive(:perform_async).with(project.id, :gc, :the_lease_key, :the_uuid).once
+        expect(::Projects::GitGarbageCollectWorker).to receive(:perform_async).with(project.id, :gc, :the_lease_key, :the_uuid).once
 
         service.execute
       end
@@ -88,7 +88,7 @@ RSpec.describe Geo::ProjectHousekeepingService do
       it 'does not run gc for a non-new repository' do
         stub_exclusive_lease(:the_lease_key, :the_uuid)
 
-        expect(GitGarbageCollectWorker).not_to receive(:perform_async)
+        expect(::Projects::GitGarbageCollectWorker).not_to receive(:perform_async)
 
         service.execute
       end
@@ -102,7 +102,7 @@ RSpec.describe Geo::ProjectHousekeepingService do
       end
 
       it 'does not enqueue a job' do
-        expect(GitGarbageCollectWorker).not_to receive(:perform_async)
+        expect(::Projects::GitGarbageCollectWorker).not_to receive(:perform_async)
 
         expect(service.send(:do_housekeeping)).to be_falsey
       end
@@ -119,10 +119,10 @@ RSpec.describe Geo::ProjectHousekeepingService do
       expect(service).to receive(:try_obtain_lease).and_return(:the_uuid)
       expect(service).to receive(:lease_key).and_return(:the_lease_key)
       expect(service).to receive(:task).and_return(:incremental_repack)
-      expect(GitGarbageCollectWorker).to receive(:perform_async).with(project.id, :incremental_repack, :the_lease_key, :the_uuid).and_call_original
+      expect(::Projects::GitGarbageCollectWorker).to receive(:perform_async).with(project.id, :incremental_repack, :the_lease_key, :the_uuid).and_call_original
 
       Sidekiq::Testing.fake! do
-        expect { service.send(:do_housekeeping) }.to change(GitGarbageCollectWorker.jobs, :size).by(1)
+        expect { service.send(:do_housekeeping) }.to change(::Projects::GitGarbageCollectWorker.jobs, :size).by(1)
       end
     end
 

spec/services/projects/cleanup_service_spec.rb

@@ -88,7 +88,7 @@ RSpec.describe Projects::CleanupService do
     end
 
     it 'runs garbage collection on the repository' do
-      expect_next_instance_of(GitGarbageCollectWorker) do |worker|
+      expect_next_instance_of(Projects::GitGarbageCollectWorker) do |worker|
         expect(worker).to receive(:perform).with(project.id, :prune, "project_cleanup:gc:#{project.id}")
       end
 

spec/support/shared_examples/services/projects/update_repository_storage_service_shared_examples.rb

@@ -71,7 +71,7 @@ RSpec.shared_examples 'moves repository to another storage' do |repository_type|
 
       it 'does not enqueue a GC run' do
         expect { subject.execute }
-          .not_to change(GitGarbageCollectWorker.jobs, :count)
+          .not_to change(Projects::GitGarbageCollectWorker.jobs, :count)
       end
     end
 
@@ -84,12 +84,12 @@ RSpec.shared_examples 'moves repository to another storage' do |repository_type|
         stub_application_setting(housekeeping_enabled: false)
 
         expect { subject.execute }
-          .not_to change(GitGarbageCollectWorker.jobs, :count)
+          .not_to change(Projects::GitGarbageCollectWorker.jobs, :count)
       end
 
       it 'enqueues a GC run' do
         expect { subject.execute }
-          .to change(GitGarbageCollectWorker.jobs, :count).by(1)
+          .to change(Projects::GitGarbageCollectWorker.jobs, :count).by(1)
       end
     end
   end

spec/support/shared_examples/services/repositories/housekeeping_shared_examples.rb

@@ -9,10 +9,10 @@ RSpec.shared_examples 'housekeeps repository' do
         expect(subject).to receive(:try_obtain_lease).and_return(:the_uuid)
         expect(subject).to receive(:lease_key).and_return(:the_lease_key)
         expect(subject).to receive(:task).and_return(:incremental_repack)
-        expect(GitGarbageCollectWorker).to receive(:perform_async).with(resource.id, :incremental_repack, :the_lease_key, :the_uuid).and_call_original
+        expect(Projects::GitGarbageCollectWorker).to receive(:perform_async).with(resource.id, :incremental_repack, :the_lease_key, :the_uuid).and_call_original
 
         Sidekiq::Testing.fake! do
-          expect { subject.execute }.to change(GitGarbageCollectWorker.jobs, :size).by(1)
+          expect { subject.execute }.to change(Projects::GitGarbageCollectWorker.jobs, :size).by(1)
         end
       end
 
@@ -38,7 +38,7 @@ RSpec.shared_examples 'housekeeps repository' do
         end
 
         it 'does not enqueue a job' do
-          expect(GitGarbageCollectWorker).not_to receive(:perform_async)
+          expect(Projects::GitGarbageCollectWorker).not_to receive(:perform_async)
 
           expect { subject.execute }.to raise_error(Repositories::HousekeepingService::LeaseTaken)
         end
@@ -63,16 +63,16 @@ RSpec.shared_examples 'housekeeps repository' do
           allow(subject).to receive(:lease_key).and_return(:the_lease_key)
 
           # At push 200
-          expect(GitGarbageCollectWorker).to receive(:perform_async).with(resource.id, :gc, :the_lease_key, :the_uuid)
+          expect(Projects::GitGarbageCollectWorker).to receive(:perform_async).with(resource.id, :gc, :the_lease_key, :the_uuid)
             .once
           # At push 50, 100, 150
-          expect(GitGarbageCollectWorker).to receive(:perform_async).with(resource.id, :full_repack, :the_lease_key, :the_uuid)
+          expect(Projects::GitGarbageCollectWorker).to receive(:perform_async).with(resource.id, :full_repack, :the_lease_key, :the_uuid)
             .exactly(3).times
           # At push 10, 20, ... (except those above)
-          expect(GitGarbageCollectWorker).to receive(:perform_async).with(resource.id, :incremental_repack, :the_lease_key, :the_uuid)
+          expect(Projects::GitGarbageCollectWorker).to receive(:perform_async).with(resource.id, :incremental_repack, :the_lease_key, :the_uuid)
             .exactly(16).times
           # At push 6, 12, 18, ... (except those above)
-          expect(GitGarbageCollectWorker).to receive(:perform_async).with(resource.id, :pack_refs, :the_lease_key, :the_uuid)
+          expect(Projects::GitGarbageCollectWorker).to receive(:perform_async).with(resource.id, :pack_refs, :the_lease_key, :the_uuid)
             .exactly(27).times
 
           201.times do
@@ -90,7 +90,7 @@ RSpec.shared_examples 'housekeeps repository' do
         allow(housekeeping).to receive(:try_obtain_lease).and_return(:gc_uuid)
         allow(housekeeping).to receive(:lease_key).and_return(:gc_lease_key)
 
-        expect(GitGarbageCollectWorker).to receive(:perform_async).with(resource.id, :gc, :gc_lease_key, :gc_uuid).twice
+        expect(Projects::GitGarbageCollectWorker).to receive(:perform_async).with(resource.id, :gc, :gc_lease_key, :gc_uuid).twice
 
         2.times do
           housekeeping.execute

spec/workers/git_garbage_collect_worker_spec.rb

@@ -5,350 +5,22 @@ require 'fileutils'
 require 'spec_helper'
 
 RSpec.describe GitGarbageCollectWorker do
-  include GitHelpers
-
   let_it_be(:project) { create(:project, :repository) }
-  let(:shell) { Gitlab::Shell.new }
-  let!(:lease_uuid) { SecureRandom.uuid }
-  let!(:lease_key) { "project_housekeeping:#{project.id}" }
-  let(:params) { [project.id, task, lease_key, lease_uuid] }
-
-  subject { described_class.new }
-
-  shared_examples 'it calls Gitaly' do
-    specify do
-      expect_any_instance_of(Gitlab::GitalyClient::RepositoryService).to receive(gitaly_task)
-        .and_return(nil)
-
-      subject.perform(*params)
-    end
-  end
 
-  shared_examples 'it updates the project statistics' do
-    it 'updates the project statistics' do
-      expect_next_instance_of(Projects::UpdateStatisticsService, project, nil, statistics: [:repository_size, :lfs_objects_size]) do |service|
-        expect(service).to receive(:execute).and_call_original
-      end
-
-      subject.perform(*params)
-    end
+  let(:lease_uuid) { SecureRandom.uuid }
+  let(:lease_key)  { "project_housekeeping:#{project.id}" }
+  let(:task)       { :full_repack }
+  let(:params)     { [project.id, task, lease_key, lease_uuid] }
 
-    it 'does nothing if the database is read-only' do
-      allow(Gitlab::Database).to receive(:read_only?) { true }
-
-      expect_any_instance_of(Projects::UpdateStatisticsService).not_to receive(:execute)
-
-      subject.perform(*params)
-    end
-  end
+  subject { described_class.new }
 
   describe "#perform" do
-    let(:gitaly_task) { :garbage_collect }
-    let(:task) { :gc }
-
-    context 'with active lease_uuid' do
-      before do
-        allow(subject).to receive(:get_lease_uuid).and_return(lease_uuid)
-      end
-
-      it_behaves_like 'it calls Gitaly'
-      it_behaves_like 'it updates the project statistics'
-
-      it "flushes ref caches when the task if 'gc'" do
-        expect(subject).to receive(:renew_lease).with(lease_key, lease_uuid).and_call_original
-        expect_any_instance_of(Repository).to receive(:expire_branches_cache).and_call_original
-        expect_any_instance_of(Repository).to receive(:branch_names).and_call_original
-        expect_any_instance_of(Repository).to receive(:has_visible_content?).and_call_original
-        expect_any_instance_of(Gitlab::Git::Repository).to receive(:has_visible_content?).and_call_original
-
-        subject.perform(*params)
-      end
-
-      it 'handles gRPC errors' do
-        expect_any_instance_of(Gitlab::GitalyClient::RepositoryService).to receive(:garbage_collect).and_raise(GRPC::NotFound)
-
-        expect { subject.perform(*params) }.to raise_exception(Gitlab::Git::Repository::NoRepository)
-      end
-    end
-
-    context 'with different lease than the active one' do
-      before do
-        allow(subject).to receive(:get_lease_uuid).and_return(SecureRandom.uuid)
-      end
-
-      it 'returns silently' do
-        expect_any_instance_of(Repository).not_to receive(:expire_branches_cache).and_call_original
-        expect_any_instance_of(Repository).not_to receive(:branch_names).and_call_original
-        expect_any_instance_of(Repository).not_to receive(:has_visible_content?).and_call_original
-
-        subject.perform(*params)
-      end
-    end
-
-    context 'with no active lease' do
-      let(:params) { [project.id] }
-
-      before do
-        allow(subject).to receive(:get_lease_uuid).and_return(false)
-      end
-
-      context 'when is able to get the lease' do
-        before do
-          allow(subject).to receive(:try_obtain_lease).and_return(SecureRandom.uuid)
-        end
-
-        it_behaves_like 'it calls Gitaly'
-        it_behaves_like 'it updates the project statistics'
-
-        it "flushes ref caches when the task if 'gc'" do
-          expect(subject).to receive(:get_lease_uuid).with("git_gc:#{task}:#{project.id}").and_return(false)
-          expect_any_instance_of(Repository).to receive(:expire_branches_cache).and_call_original
-          expect_any_instance_of(Repository).to receive(:branch_names).and_call_original
-          expect_any_instance_of(Repository).to receive(:has_visible_content?).and_call_original
-          expect_any_instance_of(Gitlab::Git::Repository).to receive(:has_visible_content?).and_call_original
-
-          subject.perform(*params)
-        end
-
-        context 'when the repository has joined a pool' do
-          let!(:pool) { create(:pool_repository, :ready) }
-          let(:project) { pool.source_project }
-
-          it 'ensures the repositories are linked' do
-            expect_any_instance_of(PoolRepository).to receive(:link_repository).once
-
-            subject.perform(*params)
-          end
-        end
-
-        context 'LFS object garbage collection' do
-          before do
-            stub_lfs_setting(enabled: true)
-          end
-
-          let_it_be(:lfs_reference) { create(:lfs_objects_project, project: project) }
-          let(:lfs_object) { lfs_reference.lfs_object }
-
-          it 'cleans up unreferenced LFS objects' do
-            expect_next_instance_of(Gitlab::Cleanup::OrphanLfsFileReferences) do |svc|
-              expect(svc.project).to eq(project)
-              expect(svc.dry_run).to be_falsy
-              expect(svc).to receive(:run!).and_call_original
-            end
-
-            subject.perform(*params)
-
-            expect(project.lfs_objects.reload).not_to include(lfs_object)
-          end
-
-          it 'catches and logs exceptions' do
-            expect_any_instance_of(Gitlab::Cleanup::OrphanLfsFileReferences)
-              .to receive(:run!)
-              .and_raise(/Failed/)
-
-            expect(Gitlab::GitLogger).to receive(:warn)
-            expect(Gitlab::ErrorTracking).to receive(:track_and_raise_for_dev_exception)
-
-            subject.perform(*params)
-          end
-
-          it 'does nothing if the database is read-only' do
-            allow(Gitlab::Database).to receive(:read_only?) { true }
-            expect_any_instance_of(Gitlab::Cleanup::OrphanLfsFileReferences).not_to receive(:run!)
-
-            subject.perform(*params)
-
-            expect(project.lfs_objects.reload).to include(lfs_object)
-          end
-        end
-      end
-
-      context 'when no lease can be obtained' do
-        before do
-          expect(subject).to receive(:try_obtain_lease).and_return(false)
-        end
-
-        it 'returns silently' do
-          expect(subject).not_to receive(:command)
-          expect_any_instance_of(Repository).not_to receive(:expire_branches_cache).and_call_original
-          expect_any_instance_of(Repository).not_to receive(:branch_names).and_call_original
-          expect_any_instance_of(Repository).not_to receive(:has_visible_content?).and_call_original
-
-          subject.perform(*params)
-        end
-      end
-    end
-
-    context "repack_full" do
-      let(:task) { :full_repack }
-      let(:gitaly_task) { :repack_full }
-
-      before do
-        expect(subject).to receive(:get_lease_uuid).and_return(lease_uuid)
+    it 'calls the Projects::GitGarbageGitGarbageCollectWorker with the same params' do
+      expect_next_instance_of(Projects::GitGarbageCollectWorker) do |instance|
+        expect(instance).to receive(:perform).with(*params)
       end
 
-      it_behaves_like 'it calls Gitaly'
-      it_behaves_like 'it updates the project statistics'
-    end
-
-    context "pack_refs" do
-      let(:task) { :pack_refs }
-      let(:gitaly_task) { :pack_refs }
-
-      before do
-        expect(subject).to receive(:get_lease_uuid).and_return(lease_uuid)
-      end
-
-      it "calls Gitaly" do
-        expect_any_instance_of(Gitlab::GitalyClient::RefService).to receive(task)
-          .and_return(nil)
-
-        subject.perform(*params)
-      end
-
-      it 'does not update the project statistics' do
-        expect(Projects::UpdateStatisticsService).not_to receive(:new)
-
-        subject.perform(*params)
-      end
-    end
-
-    context "repack_incremental" do
-      let(:task) { :incremental_repack }
-      let(:gitaly_task) { :repack_incremental }
-
-      before do
-        expect(subject).to receive(:get_lease_uuid).and_return(lease_uuid)
-      end
-
-      it_behaves_like 'it calls Gitaly'
-      it_behaves_like 'it updates the project statistics'
-    end
-
-    shared_examples 'gc tasks' do
-      before do
-        allow(subject).to receive(:get_lease_uuid).and_return(lease_uuid)
-        allow(subject).to receive(:bitmaps_enabled?).and_return(bitmaps_enabled)
-      end
-
-      it 'incremental repack adds a new packfile' do
-        create_objects(project)
-        before_packs = packs(project)
-
-        expect(before_packs.count).to be >= 1
-
-        subject.perform(project.id, 'incremental_repack', lease_key, lease_uuid)
-        after_packs = packs(project)
-
-        # Exactly one new pack should have been created
-        expect(after_packs.count).to eq(before_packs.count + 1)
-
-        # Previously existing packs are still around
-        expect(before_packs & after_packs).to eq(before_packs)
-      end
-
-      it 'full repack consolidates into 1 packfile' do
-        create_objects(project)
-        subject.perform(project.id, 'incremental_repack', lease_key, lease_uuid)
-        before_packs = packs(project)
-
-        expect(before_packs.count).to be >= 2
-
-        subject.perform(project.id, 'full_repack', lease_key, lease_uuid)
-        after_packs = packs(project)
-
-        expect(after_packs.count).to eq(1)
-
-        # Previously existing packs should be gone now
-        expect(after_packs - before_packs).to eq(after_packs)
-
-        expect(File.exist?(bitmap_path(after_packs.first))).to eq(bitmaps_enabled)
-      end
-
-      it 'gc consolidates into 1 packfile and updates packed-refs' do
-        create_objects(project)
-        before_packs = packs(project)
-        before_packed_refs = packed_refs(project)
-
-        expect(before_packs.count).to be >= 1
-
-        expect_any_instance_of(Gitlab::GitalyClient::RepositoryService)
-          .to receive(:garbage_collect)
-          .with(bitmaps_enabled, prune: false)
-          .and_call_original
-
-        subject.perform(project.id, 'gc', lease_key, lease_uuid)
-        after_packed_refs = packed_refs(project)
-        after_packs = packs(project)
-
-        expect(after_packs.count).to eq(1)
-
-        # Previously existing packs should be gone now
-        expect(after_packs - before_packs).to eq(after_packs)
-
-        # The packed-refs file should have been updated during 'git gc'
-        expect(before_packed_refs).not_to eq(after_packed_refs)
-
-        expect(File.exist?(bitmap_path(after_packs.first))).to eq(bitmaps_enabled)
-      end
-
-      it 'cleans up repository after finishing' do
-        expect_any_instance_of(Project).to receive(:cleanup).and_call_original
-
-        subject.perform(project.id, 'gc', lease_key, lease_uuid)
-      end
-
-      it 'prune calls garbage_collect with the option prune: true' do
-        expect_any_instance_of(Gitlab::GitalyClient::RepositoryService)
-          .to receive(:garbage_collect)
-          .with(bitmaps_enabled, prune: true)
-          .and_return(nil)
-
-        subject.perform(project.id, 'prune', lease_key, lease_uuid)
-      end
-    end
-
-    context 'with bitmaps enabled' do
-      let(:bitmaps_enabled) { true }
-
-      include_examples 'gc tasks'
-    end
-
-    context 'with bitmaps disabled' do
-      let(:bitmaps_enabled) { false }
-
-      include_examples 'gc tasks'
-    end
-  end
-
-  # Create a new commit on a random new branch
-  def create_objects(project)
-    rugged = rugged_repo(project.repository)
-    old_commit = rugged.branches.first.target
-    new_commit_sha = Rugged::Commit.create(
-      rugged,
-      message: "hello world #{SecureRandom.hex(6)}",
-      author: { email: 'foo@bar', name: 'baz' },
-      committer: { email: 'foo@bar', name: 'baz' },
-      tree: old_commit.tree,
-      parents: [old_commit]
-    )
-    rugged.references.create("refs/heads/#{SecureRandom.hex(6)}", new_commit_sha)
-  end
-
-  def packs(project)
-    Gitlab::GitalyClient::StorageSettings.allow_disk_access do
-      Dir["#{project.repository.path_to_repo}/objects/pack/*.pack"]
+      subject.perform(*params)
     end
   end
-
-  def packed_refs(project)
-    path = "#{project.repository.path_to_repo}/packed-refs"
-    FileUtils.touch(path)
-    File.read(path)
-  end
-
-  def bitmap_path(pack)
-    pack.sub(/\.pack\z/, '.bitmap')
-  end
 end

spec/workers/projects/git_garbage_collect_worker_spec.rb

+# frozen_string_literal: true
+
+require 'fileutils'
+require 'spec_helper'
+
+RSpec.describe Projects::GitGarbageCollectWorker do
+  include GitHelpers
+
+  let_it_be(:project) { create(:project, :repository) }
+
+  let!(:lease_uuid) { SecureRandom.uuid }
+  let!(:lease_key) { "project_housekeeping:#{project.id}" }
+  let(:params) { [project.id, task, lease_key, lease_uuid] }
+  let(:shell) { Gitlab::Shell.new }
+  let(:repository) { project.repository }
+
+  subject { described_class.new }
+
+  before do
+    allow(subject).to receive(:find_project).and_return(project)
+  end
+
+  shared_examples 'it calls Gitaly' do
+    specify do
+      repository_service = instance_double(Gitlab::GitalyClient::RepositoryService)
+
+      expect(subject).to receive(:get_gitaly_client).with(task, repository.raw_repository).and_return(repository_service)
+      expect(repository_service).to receive(gitaly_task)
+
+      subject.perform(*params)
+    end
+  end
+
+  shared_examples 'it updates the project statistics' do
+    it 'updates the project statistics' do
+      expect_next_instance_of(Projects::UpdateStatisticsService, project, nil, statistics: [:repository_size, :lfs_objects_size]) do |service|
+        expect(service).to receive(:execute)
+      end
+
+      subject.perform(*params)
+    end
+
+    it 'does nothing if the database is read-only' do
+      allow(Gitlab::Database).to receive(:read_only?) { true }
+
+      expect(Projects::UpdateStatisticsService).not_to receive(:new)
+
+      subject.perform(*params)
+    end
+  end
+
+  describe '#perform', :aggregate_failures do
+    let(:gitaly_task) { :garbage_collect }
+    let(:task) { :gc }
+
+    context 'with active lease_uuid' do
+      before do
+        allow(subject).to receive(:get_lease_uuid).and_return(lease_uuid)
+      end
+
+      it_behaves_like 'it calls Gitaly'
+      it_behaves_like 'it updates the project statistics'
+
+      it "flushes ref caches when the task if 'gc'" do
+        expect(subject).to receive(:renew_lease).with(lease_key, lease_uuid).and_call_original
+        expect(repository).to receive(:expire_branches_cache).and_call_original
+        expect(repository).to receive(:branch_names).and_call_original
+        expect(repository).to receive(:has_visible_content?).and_call_original
+        expect(repository.raw_repository).to receive(:has_visible_content?).and_call_original
+
+        subject.perform(*params)
+      end
+
+      it 'handles gRPC errors' do
+        allow_next_instance_of(Gitlab::GitalyClient::RepositoryService, repository.raw_repository) do |instance|
+          allow(instance).to receive(:garbage_collect).and_raise(GRPC::NotFound)
+        end
+
+        expect { subject.perform(*params) }.to raise_exception(Gitlab::Git::Repository::NoRepository)
+      end
+    end
+
+    context 'with different lease than the active one' do
+      before do
+        allow(subject).to receive(:get_lease_uuid).and_return(SecureRandom.uuid)
+      end
+
+      it 'returns silently' do
+        expect(repository).not_to receive(:expire_branches_cache).and_call_original
+        expect(repository).not_to receive(:branch_names).and_call_original
+        expect(repository).not_to receive(:has_visible_content?).and_call_original
+
+        subject.perform(*params)
+      end
+    end
+
+    context 'with no active lease' do
+      let(:params) { [project.id] }
+
+      before do
+        allow(subject).to receive(:get_lease_uuid).and_return(false)
+      end
+
+      context 'when is able to get the lease' do
+        before do
+          allow(subject).to receive(:try_obtain_lease).and_return(SecureRandom.uuid)
+        end
+
+        it_behaves_like 'it calls Gitaly'
+        it_behaves_like 'it updates the project statistics'
+
+        it "flushes ref caches when the task if 'gc'" do
+          expect(subject).to receive(:get_lease_uuid).with("git_gc:#{task}:#{project.id}").and_return(false)
+          expect(repository).to receive(:expire_branches_cache).and_call_original
+          expect(repository).to receive(:branch_names).and_call_original
+          expect(repository).to receive(:has_visible_content?).and_call_original
+          expect(repository.raw_repository).to receive(:has_visible_content?).and_call_original
+
+          subject.perform(*params)
+        end
+
+        context 'when the repository has joined a pool' do
+          let!(:pool) { create(:pool_repository, :ready) }
+          let(:project) { pool.source_project }
+
+          it 'ensures the repositories are linked' do
+            expect(project.pool_repository).to receive(:link_repository).once
+
+            subject.perform(*params)
+          end
+        end
+
+        context 'LFS object garbage collection' do
+          before do
+            stub_lfs_setting(enabled: true)
+          end
+
+          let_it_be(:lfs_reference) { create(:lfs_objects_project, project: project) }
+          let(:lfs_object) { lfs_reference.lfs_object }
+
+          it 'cleans up unreferenced LFS objects' do
+            expect_next_instance_of(Gitlab::Cleanup::OrphanLfsFileReferences) do |svc|
+              expect(svc.project).to eq(project)
+              expect(svc.dry_run).to be_falsy
+              expect(svc).to receive(:run!).and_call_original
+            end
+
+            subject.perform(*params)
+
+            expect(project.lfs_objects.reload).not_to include(lfs_object)
+          end
+
+          it 'catches and logs exceptions' do
+            allow_next_instance_of(Gitlab::Cleanup::OrphanLfsFileReferences) do |svc|
+              allow(svg).to receive(:run!).and_raise(/Failed/)
+            end
+
+            expect(Gitlab::GitLogger).to receive(:warn)
+            expect(Gitlab::ErrorTracking).to receive(:track_and_raise_for_dev_exception)
+
+            subject.perform(*params)
+          end
+
+          it 'does nothing if the database is read-only' do
+            allow(Gitlab::Database).to receive(:read_only?) { true }
+            expect(Gitlab::Cleanup::OrphanLfsFileReferences).not_to receive(:new)
+
+            subject.perform(*params)
+
+            expect(project.lfs_objects.reload).to include(lfs_object)
+          end
+        end
+      end
+
+      context 'when no lease can be obtained' do
+        it 'returns silently' do
+          expect(subject).to receive(:try_obtain_lease).and_return(false)
+
+          expect(subject).not_to receive(:command)
+          expect(repository).not_to receive(:expire_branches_cache).and_call_original
+          expect(repository).not_to receive(:branch_names).and_call_original
+          expect(repository).not_to receive(:has_visible_content?).and_call_original
+
+          subject.perform(*params)
+        end
+      end
+    end
+
+    context 'repack_full' do
+      let(:task) { :full_repack }
+      let(:gitaly_task) { :repack_full }
+
+      before do
+        expect(subject).to receive(:get_lease_uuid).and_return(lease_uuid)
+      end
+
+      it_behaves_like 'it calls Gitaly'
+      it_behaves_like 'it updates the project statistics'
+    end
+
+    context 'pack_refs' do
+      let(:task) { :pack_refs }
+      let(:gitaly_task) { :pack_refs }
+
+      before do
+        expect(subject).to receive(:get_lease_uuid).and_return(lease_uuid)
+      end
+
+      it 'calls Gitaly' do
+        repository_service = instance_double(Gitlab::GitalyClient::RefService)
+
+        expect(subject).to receive(:get_gitaly_client).with(task, repository.raw_repository).and_return(repository_service)
+        expect(repository_service).to receive(gitaly_task)
+
+        subject.perform(*params)
+      end
+
+      it 'does not update the project statistics' do
+        expect(Projects::UpdateStatisticsService).not_to receive(:new)
+
+        subject.perform(*params)
+      end
+    end
+
+    context 'repack_incremental' do
+      let(:task) { :incremental_repack }
+      let(:gitaly_task) { :repack_incremental }
+
+      before do
+        expect(subject).to receive(:get_lease_uuid).and_return(lease_uuid)
+      end
+
+      it_behaves_like 'it calls Gitaly'
+      it_behaves_like 'it updates the project statistics'
+    end
+
+    shared_examples 'gc tasks' do
+      before do
+        allow(subject).to receive(:get_lease_uuid).and_return(lease_uuid)
+        allow(subject).to receive(:bitmaps_enabled?).and_return(bitmaps_enabled)
+      end
+
+      it 'incremental repack adds a new packfile' do
+        create_objects(project)
+        before_packs = packs(project)
+
+        expect(before_packs.count).to be >= 1
+
+        subject.perform(project.id, 'incremental_repack', lease_key, lease_uuid)
+        after_packs = packs(project)
+
+        # Exactly one new pack should have been created
+        expect(after_packs.count).to eq(before_packs.count + 1)
+
+        # Previously existing packs are still around
+        expect(before_packs & after_packs).to eq(before_packs)
+      end
+
+      it 'full repack consolidates into 1 packfile' do
+        create_objects(project)
+        subject.perform(project.id, 'incremental_repack', lease_key, lease_uuid)
+        before_packs = packs(project)
+
+        expect(before_packs.count).to be >= 2
+
+        subject.perform(project.id, 'full_repack', lease_key, lease_uuid)
+        after_packs = packs(project)
+
+        expect(after_packs.count).to eq(1)
+
+        # Previously existing packs should be gone now
+        expect(after_packs - before_packs).to eq(after_packs)
+
+        expect(File.exist?(bitmap_path(after_packs.first))).to eq(bitmaps_enabled)
+      end
+
+      it 'gc consolidates into 1 packfile and updates packed-refs' do
+        create_objects(project)
+        before_packs = packs(project)
+        before_packed_refs = packed_refs(project)
+
+        expect(before_packs.count).to be >= 1
+
+        # It's quite difficult to use `expect_next_instance_of` in this place
+        # because the RepositoryService is instantiated several times to do
+        # some repository calls like `exists?`, `create_repository`, ... .
+        # Therefore, since we're instantiating the object several times,
+        # RSpec has troubles figuring out which instance is the next and which
+        # one we want to mock.
+        # Besides, at this point, we actually want to perform the call to Gitaly,
+        # otherwise we would just use `instance_double` like in other parts of the
+        # spec file.
+        expect_any_instance_of(Gitlab::GitalyClient::RepositoryService) # rubocop:disable RSpec/AnyInstanceOf
+          .to receive(:garbage_collect)
+          .with(bitmaps_enabled, prune: false)
+          .and_call_original
+
+        subject.perform(project.id, 'gc', lease_key, lease_uuid)
+        after_packed_refs = packed_refs(project)
+        after_packs = packs(project)
+
+        expect(after_packs.count).to eq(1)
+
+        # Previously existing packs should be gone now
+        expect(after_packs - before_packs).to eq(after_packs)
+
+        # The packed-refs file should have been updated during 'git gc'
+        expect(before_packed_refs).not_to eq(after_packed_refs)
+
+        expect(File.exist?(bitmap_path(after_packs.first))).to eq(bitmaps_enabled)
+      end
+
+      it 'cleans up repository after finishing' do
+        expect(project).to receive(:cleanup).and_call_original
+
+        subject.perform(project.id, 'gc', lease_key, lease_uuid)
+      end
+
+      it 'prune calls garbage_collect with the option prune: true' do
+        repository_service = instance_double(Gitlab::GitalyClient::RepositoryService)
+
+        expect(subject).to receive(:get_gitaly_client).with(:prune, repository.raw_repository).and_return(repository_service)
+        expect(repository_service).to receive(:garbage_collect).with(bitmaps_enabled, prune: true)
+
+        subject.perform(project.id, 'prune', lease_key, lease_uuid)
+      end
+    end
+
+    context 'with bitmaps enabled' do
+      let(:bitmaps_enabled) { true }
+
+      include_examples 'gc tasks'
+    end
+
+    context 'with bitmaps disabled' do
+      let(:bitmaps_enabled) { false }
+
+      include_examples 'gc tasks'
+    end
+  end
+
+  # Create a new commit on a random new branch
+  def create_objects(project)
+    rugged = rugged_repo(project.repository)
+    old_commit = rugged.branches.first.target
+    new_commit_sha = Rugged::Commit.create(
+      rugged,
+      message: "hello world #{SecureRandom.hex(6)}",
+      author: { email: 'foo@bar', name: 'baz' },
+      committer: { email: 'foo@bar', name: 'baz' },
+      tree: old_commit.tree,
+      parents: [old_commit]
+    )
+    rugged.references.create("refs/heads/#{SecureRandom.hex(6)}", new_commit_sha)
+  end
+
+  def packs(project)
+    Dir["#{path_to_repo}/objects/pack/*.pack"]
+  end
+
+  def packed_refs(project)
+    path = File.join(path_to_repo, 'packed-refs')
+    FileUtils.touch(path)
+    File.read(path)
+  end
+
+  def path_to_repo
+    @path_to_repo ||= File.join(TestEnv.repos_path, project.repository.relative_path)
+  end
+
+  def bitmap_path(pack)
+    pack.sub(/\.pack\z/, '.bitmap')
+  end
+end