Geo Batch Operations

Added Resync all and Recheck all actions to the Geo > Projects page.

Added ProjectRegistryBatchWorker that is responsible for doing mass
update on the ProjectRegistry entries to flag them as requiring recheck
or resync.

app/workers/all_queues.yml

@@ -169,6 +169,7 @@
 - geo:geo_file_registry_removal
 - geo:geo_hashed_storage_attachments_migration
 - geo:geo_hashed_storage_migration
+- geo:geo_project_registry_batch
 - geo:geo_project_sync
 - geo:geo_rename_repository
 - geo:geo_repositories_clean_up

ee/app/controllers/admin/geo/projects_controller.rb

@@ -55,6 +55,18 @@ class Admin::Geo::ProjectsController < Admin::ApplicationController
     redirect_back_or_admin_geo_projects(notice: s_('Geo|%{name} is scheduled for forced re-download') % { name: @registry.project.full_name })
   end
 
+  def recheck_all
+    Geo::ProjectRegistryBatchWorker.perform_async(:recheck_repositories)
+
+    redirect_back_or_admin_geo_projects(notice: s_('Geo|All projects are being scheduled for re-check'))
+  end
+
+  def resync_all
+    Geo::ProjectRegistryBatchWorker.perform_async(:resync_repositories)
+
+    redirect_back_or_admin_geo_projects(notice: s_('Geo|All projects are being scheduled for re-sync'))
+  end
+
   private
 
   def check_license

ee/app/models/geo/project_registry.rb

@@ -88,6 +88,27 @@ class Geo::ProjectRegistry < Geo::BaseRegistry
     where(project: Geo::Fdw::Project.search(query))
   end
 
+  def self.flag_repositories_for_resync!
+    update_all(
+      resync_repository: true,
+      repository_verification_checksum_sha: nil,
+      repository_checksum_mismatch: false,
+      last_repository_verification_failure: nil,
+      repository_verification_retry_count: nil,
+      resync_repository_was_scheduled_at: Time.now,
+      repository_retry_count: nil,
+      repository_retry_at: nil
+    )
+  end
+
+  def self.flag_repositories_for_recheck!
+    update_all(
+      repository_verification_checksum_sha: nil,
+      last_repository_verification_failure: nil,
+      repository_checksum_mismatch: false
+    )
+  end
+
   # Must be run before fetching the repository to avoid a race condition
   #
   # @param [String] type must be one of the values in TYPES

ee/app/views/admin/geo/projects/index.html.haml

@@ -24,6 +24,15 @@
           = s_('Geo|Never')
     .nav-controls
       = render(partial: 'shared/projects/search_form', autofocus: true)
+      .dropdown
+        %a.btn.dropdown-toggle{ href: '#', data: { toggle: 'dropdown' }, 'aria-haspopup' => 'true', 'aria-expanded' => 'false' }
+          = icon('gears')
+          = s_('Geo|Batch operations')
+          = icon("caret-down")
+        %ul.dropdown-menu.dropdown-menu-right
+          %li.dropdown-header= s_('Geo|Batch operations')
+          %li= link_to s_('Geo|Resync all projects'), resync_all_admin_geo_projects_path, method: :post
+          %li= link_to s_('Geo|Recheck all projects'), recheck_all_admin_geo_projects_path, method: :post
 
   - case params[:sync_status]
   - when 'never'

ee/app/workers/geo/project_registry_batch_worker.rb

+# frozen_string_literal: true
+
+module Geo
+  # Responsible for scheduling multiple jobs to mark Project Registries as requiring syncing or verification.
+  #
+  # This class includes an Exclusive Lease guard and only one can be executed at the same time
+  # If multiple jobs are scheduled, only one will run and the others will drop forever.
+  class ProjectRegistryBatchWorker
+    include ApplicationWorker
+    include GeoQueue
+    include ExclusiveLeaseGuard
+    include ::Gitlab::Geo::LogHelpers
+
+    BATCH_SIZE = 1000
+    LEASE_TIMEOUT = 8.hours
+    OPERATIONS = [:resync_repositories, :recheck_repositories].freeze
+
+    def perform(operation)
+      try_obtain_lease do
+        case operation.to_sym
+        when :resync_repositories
+          resync_repositories
+        when :recheck_repositories
+          recheck_repositories
+        else
+          fail_invalid_operation!(operation)
+        end
+      end
+    end
+
+    private
+
+    def resync_repositories
+      Geo::ProjectRegistry.each_batch(of: BATCH_SIZE) do |batch|
+        batch.flag_repositories_for_resync!
+      end
+    end
+
+    def recheck_repositories
+      Geo::ProjectRegistry.each_batch(of: BATCH_SIZE) do |batch|
+        batch.flag_repositories_for_recheck!
+      end
+    end
+
+    def lease_timeout
+      LEASE_TIMEOUT
+    end
+
+    def fail_invalid_operation!(operation)
+      raise ArgumentError, "Invalid operation: '#{operation.inspect}' informed. Must be one of the following: #{OPERATIONS.map { |valid_op| "'#{valid_op}'" }.join(', ')}"
+    end
+  end
+end

ee/changelogs/unreleased/7126-batch-operations.yml

+---
+title: 'Geo: Admin > Geo > Projects support for batch operations'
+merge_request: 7806
+author:
+type: added

ee/config/routes/admin.rb

@@ -34,6 +34,11 @@ namespace :admin do
         post :resync
         post :force_redownload
       end
+
+      collection do
+        post :recheck_all
+        post :resync_all
+      end
     end
   end
 

ee/lib/ee/gitlab/middleware/read_only/controller.rb

@@ -10,7 +10,7 @@ module EE
           }.freeze
 
           WHITELISTED_GEO_ROUTES_TRACKING_DB = {
-            'admin/geo/projects' => %w{destroy resync recheck force_redownload}
+            'admin/geo/projects' => %w{destroy resync recheck force_redownload resync_all recheck_all}
           }.freeze
 
           private

ee/spec/controllers/admin/geo/projects_controller_spec.rb

@@ -27,7 +27,7 @@ describe Admin::Geo::ProjectsController, :geo do
       render_views
 
       before do
-        allow(Gitlab::Geo).to receive(:license_allows?).and_return(true)
+        stub_licensed_features(geo: true)
       end
 
       context 'without sync_status specified' do
@@ -87,7 +87,7 @@ describe Admin::Geo::ProjectsController, :geo do
 
     context 'with a valid license' do
       before do
-        allow(Gitlab::Geo).to receive(:license_allows?).and_return(true)
+        stub_licensed_features(geo: true)
       end
 
       context 'with an orphaned registry' do
@@ -117,7 +117,7 @@ describe Admin::Geo::ProjectsController, :geo do
 
     context 'with a valid license' do
       before do
-        allow(Gitlab::Geo).to receive(:license_allows?).and_return(true)
+        stub_licensed_features(geo: true)
       end
 
       it 'flags registry for recheck' do
@@ -135,7 +135,7 @@ describe Admin::Geo::ProjectsController, :geo do
 
     context 'with a valid license' do
       before do
-        allow(Gitlab::Geo).to receive(:license_allows?).and_return(true)
+        stub_licensed_features(geo: true)
       end
 
       it 'flags registry for resync' do
@@ -146,6 +146,56 @@ describe Admin::Geo::ProjectsController, :geo do
     end
   end
 
+  describe '#recheck_all' do
+    subject { post :recheck_all }
+
+    it_behaves_like 'license required'
+
+    context 'with a valid license' do
+      before do
+        stub_licensed_features(geo: true)
+      end
+
+      it 'schedules a batch job' do
+        Sidekiq::Testing.fake! do
+          expect { subject }.to change(Geo::ProjectRegistryBatchWorker.jobs, :size).by(1)
+          expect(Geo::ProjectRegistryBatchWorker.jobs.last['args']).to include('recheck_repositories')
+        end
+      end
+
+      it 'redirects back and display confirmation' do
+        Sidekiq::Testing.inline! do
+          expect(subject).to redirect_to(admin_geo_projects_path)
+          expect(flash[:notice]).to include('All projects are being scheduled for re-check')
+        end
+      end
+    end
+  end
+
+  describe '#resync_all' do
+    subject { post :resync_all }
+
+    it_behaves_like 'license required'
+
+    context 'with a valid license' do
+      before do
+        stub_licensed_features(geo: true)
+      end
+
+      it 'schedules a batch job' do
+        Sidekiq::Testing.fake! do
+          expect { subject }.to change(Geo::ProjectRegistryBatchWorker.jobs, :size).by(1)
+          expect(Geo::ProjectRegistryBatchWorker.jobs.last['args']).to include('resync_repositories')
+        end
+      end
+
+      it 'redirects back and display confirmation' do
+        expect(subject).to redirect_to(admin_geo_projects_path)
+        expect(flash[:notice]).to include('All projects are being scheduled for re-sync')
+      end
+    end
+  end
+
   describe '#force_redownload' do
     subject { post :force_redownload, id: synced_registry }
 
@@ -153,7 +203,7 @@ describe Admin::Geo::ProjectsController, :geo do
 
     context 'with a valid license' do
       before do
-        allow(Gitlab::Geo).to receive(:license_allows?).and_return(true)
+        stub_licensed_features(geo: true)
       end
 
       it 'flags registry for re-download' do

ee/spec/lib/ee/gitlab/middleware/read_only_spec.rb

@@ -41,46 +41,30 @@ describe Gitlab::Middleware::ReadOnly do
       allow(Gitlab::Database).to receive(:read_only?) { true }
     end
 
-    context 'whitelisted requests' do
-      it 'expects a PATCH request to geo_nodes update URL to be allowed' do
+    shared_examples 'whitelisted request' do |request_type, request_url|
+      it "expects a #{request_type.upcase} request to #{request_url} to be allowed" do
         expect(Rails.application.routes).to receive(:recognize_path).and_call_original
-        response = request.patch('/admin/geo/nodes/1')
+        response = request.send(request_type, request_url)
 
         expect(response).not_to be_redirect
         expect(subject).not_to disallow_request
       end
-
-      it 'expects a DELETE request to geo projects delete URL to be allowed' do
-        expect(Rails.application.routes).to receive(:recognize_path).and_call_original
-        response = request.delete('/admin/geo/projects/1')
-
-        expect(response).not_to be_redirect
-        expect(subject).not_to disallow_request
     end
 
-      it 'expects a POST request to geo projects resync URL to be allowed' do
-        expect(Rails.application.routes).to receive(:recognize_path).and_call_original
-        response = request.post('/admin/geo/projects/1/resync')
+    context 'whitelisted requests' do
+      it_behaves_like 'whitelisted request', :patch, '/admin/geo/nodes/1'
 
-        expect(response).not_to be_redirect
-        expect(subject).not_to disallow_request
-      end
+      it_behaves_like 'whitelisted request', :delete, '/admin/geo/projects/1'
 
-      it 'expects a POST request to geo projects recheck URL to be allowed' do
-        expect(Rails.application.routes).to receive(:recognize_path).and_call_original
-        response = request.post('/admin/geo/projects/1/recheck')
+      it_behaves_like 'whitelisted request', :post, '/admin/geo/projects/1/resync'
 
-        expect(response).not_to be_redirect
-        expect(subject).not_to disallow_request
-      end
+      it_behaves_like 'whitelisted request', :post, '/admin/geo/projects/1/recheck'
 
-      it 'expects a POST request to geo projects force redownload URL to be allowed' do
-        expect(Rails.application.routes).to receive(:recognize_path).and_call_original
-        response = request.post('/admin/geo/projects/1/force_redownload')
+      it_behaves_like 'whitelisted request', :post, '/admin/geo/projects/recheck_all'
 
-        expect(response).not_to be_redirect
-        expect(subject).not_to disallow_request
-      end
+      it_behaves_like 'whitelisted request', :post, '/admin/geo/projects/resync_all'
+
+      it_behaves_like 'whitelisted request', :post, '/admin/geo/projects/1/force_redownload'
     end
   end
 end

ee/spec/models/geo/project_registry_spec.rb

@@ -175,6 +175,39 @@ describe Geo::ProjectRegistry do
     end
   end
 
+  describe '.flag_repositories_for_recheck!' do
+    it 'modified record to a recheck state' do
+      registry = create(:geo_project_registry, :repository_verified)
+
+      described_class.flag_repositories_for_recheck!
+
+      expect(registry.reload).to have_attributes(
+        repository_verification_checksum_sha: nil,
+        last_repository_verification_failure: nil,
+        repository_checksum_mismatch: false
+      )
+    end
+  end
+
+  describe '.flag_repositories_for_resync!' do
+    it 'modified record to a resync state' do
+      registry = create(:geo_project_registry, :synced)
+
+      described_class.flag_repositories_for_resync!
+
+      expect(registry.reload).to have_attributes(
+        resync_repository: true,
+        repository_verification_checksum_sha: nil,
+        last_repository_verification_failure: nil,
+        repository_checksum_mismatch: false,
+        repository_verification_retry_count: nil,
+        repository_retry_count: nil,
+        repository_retry_at: nil
+
+      )
+    end
+  end
+
   describe '#repository_sync_due?' do
     where(:last_synced_at, :resync, :retry_at, :expected) do
       now = Time.now

ee/spec/workers/geo/project_registry_batch_worker_spec.rb

+# frozen_string_literal: true
+
+require 'rails_helper'
+
+RSpec.describe Geo::ProjectRegistryBatchWorker do
+  include ExclusiveLeaseHelpers
+  include ::EE::GeoHelpers
+
+  set(:secondary) { create(:geo_node) }
+
+  before do
+    stub_current_geo_node(secondary)
+    stub_exclusive_lease(renew: true)
+  end
+
+  describe '#perform' do
+    context 'when operation is :recheck_repositories' do
+      let!(:registry) { create(:geo_project_registry, :repository_verified) }
+
+      it 'flags repositories for recheck' do
+        Sidekiq::Testing.inline! do
+          subject.perform(:recheck_repositories)
+        end
+
+        expect(registry.reload.repository_verification_pending?).to be_truthy
+      end
+
+      it 'does nothing if exclusive lease is already acquired' do
+        stub_exclusive_lease_taken('geo/project_registry_batch_worker', timeout: 20)
+
+        Sidekiq::Testing.inline! do
+          subject.perform(:recheck_repositories)
+        end
+
+        expect(registry).to have_attributes(registry.reload.attributes)
+      end
+    end
+
+    context 'when operation is :resync_repositories' do
+      let!(:registry) { create(:geo_project_registry, :synced) }
+
+      it 'flags repositories for resync' do
+        Sidekiq::Testing.inline! do
+          subject.perform(:resync_repositories)
+        end
+
+        expect(registry.reload.resync_repository?).to be_truthy
+      end
+
+      it 'does nothing if exclusive lease is already acquired' do
+        stub_exclusive_lease_taken('geo/project_registry_batch_worker', timeout: 20)
+
+        Sidekiq::Testing.inline! do
+          subject.perform(:recheck_repositories)
+        end
+
+        expect(registry).to have_attributes(registry.reload.attributes)
+      end
+    end
+
+    context 'when informed operation is unknown/invalid' do
+      it 'fails with ArgumentError' do
+        expect { subject.perform(:unknown_operation) }.to raise_error(ArgumentError)
+      end
+    end
+  end
+end

locale/gitlab.pot

@@ -3721,6 +3721,15 @@ msgstr ""
 msgid "Geo|All projects"
 msgstr ""
 
+msgid "Geo|All projects are being scheduled for re-check"
+msgstr ""
+
+msgid "Geo|All projects are being scheduled for re-sync"
+msgstr ""
+
+msgid "Geo|Batch operations"
+msgstr ""
+
 msgid "Geo|Could not remove tracking entry for an existing project."
 msgstr ""
 
@@ -3778,6 +3787,9 @@ msgstr ""
 msgid "Geo|Recheck"
 msgstr ""
 
+msgid "Geo|Recheck all projects"
+msgstr ""
+
 msgid "Geo|Redownload"
 msgstr ""
 
@@ -3790,6 +3802,9 @@ msgstr ""
 msgid "Geo|Resync"
 msgstr ""
 
+msgid "Geo|Resync all projects"
+msgstr ""
+
 msgid "Geo|Retry count"
 msgstr ""