Remove orphaned project registries

Some registries can be orphaned because of lost Sidekiq job for example

doc/administration/raketasks/geo.md

@@ -55,3 +55,20 @@ sudo gitlab-rake geo:git:housekeeping:gc
 ```bash
 sudo -u git -H bundle exec rake geo:git:housekeeping:gc RAILS_ENV=production
 ```
+
+## Remove orphaned project registries
+
+Under certain conditions your project registry can contain obsolete records, you
+can remove them using the rake task `geo:run_orphaned_project_registry_cleaner`:
+
+**Omnibus Installation**
+
+```
+sudo gitlab-rake geo:run_orphaned_project_registry_cleaner
+```
+
+**Source Installation**
+
+```bash
+sudo -u git -H bundle exec rake geo:run_orphaned_project_registry_cleaner RAILS_ENV=production
+```

ee/changelogs/unreleased/13158-orphaned-project-registries.yml

+---
+title: 'Geo: Add orphaned project registry cleaner'
+merge_request: 15021
+author:
+type: changed

ee/lib/tasks/geo.rake

@@ -166,6 +166,46 @@ namespace :geo do
     end
   end
 
+  desc 'Run orphaned project registry cleaner'
+  task run_orphaned_project_registry_cleaner: :environment do
+    abort GEO_LICENSE_ERROR_TEXT unless Gitlab::Geo.license_allows?
+
+    unless Gitlab::Geo.secondary?
+      abort 'This is not a secondary node'
+    end
+
+    from_project_id = ENV['FROM_PROJECT_ID'] || Geo::ProjectRegistry.minimum(:project_id)
+    to_project_id = ENV['TO_PROJECT_ID'] || Geo::ProjectRegistry.maximum(:project_id)
+
+    if from_project_id > to_project_id
+      abort 'FROM_PROJECT_ID can not be greater than TO_PROJECT_ID'
+    end
+
+    batch_size = 1000
+    total_count = 0
+    current_max_id = 0
+
+    until current_max_id >= to_project_id
+      current_max_id = [from_project_id + batch_size, to_project_id + 1].min
+
+      project_ids = Project
+        .where('id >= ? AND id < ?', from_project_id, current_max_id)
+        .pluck_primary_key
+
+      orphaned_registries = Geo::ProjectRegistry
+        .where('project_id NOT IN(?)', project_ids)
+        .where('project_id >= ? AND project_id < ?', from_project_id, current_max_id)
+      count = orphaned_registries.delete_all
+      total_count += count
+
+      puts "Checked project ids from #{from_project_id} to #{current_max_id} registries. Removed #{count} orphaned registries"
+
+      from_project_id = current_max_id
+    end
+
+    puts "Orphaned registries removed(total): #{total_count}"
+  end
+
   desc 'Make this node the Geo primary'
   task set_primary_node: :environment do
     abort GEO_LICENSE_ERROR_TEXT unless Gitlab::Geo.license_allows?

ee/spec/tasks/geo_rake_spec.rb

@@ -347,4 +347,51 @@ describe 'geo rake tasks', :geo do
       end
     end
   end
+
+  describe 'geo:run_orphaned_project_registry_cleaner' do
+    let!(:current_node) { create(:geo_node) }
+
+    before do
+      stub_current_geo_node(current_node)
+
+      create(:geo_project_registry)
+      create(:geo_project_registry)
+
+      @orphaned = create(:geo_project_registry)
+      @orphaned.project.delete
+      @orphaned1 = create(:geo_project_registry)
+      @orphaned1.project.delete
+
+      create(:geo_project_registry)
+    end
+
+    it 'removes orphaned registries' do
+      run_rake_task('geo:run_orphaned_project_registry_cleaner')
+
+      expect(Geo::ProjectRegistry.count).to be 3
+      expect(Geo::ProjectRegistry.find_by_id(@orphaned.id)).to be nil
+    end
+
+    it 'removes orphaned registries taking into account TO_PROJECT_ID' do
+      allow(ENV).to receive(:[]).with('FROM_PROJECT_ID').and_return(nil)
+      allow(ENV).to receive(:[]).with('TO_PROJECT_ID').and_return(@orphaned.project_id)
+
+      run_rake_task('geo:run_orphaned_project_registry_cleaner')
+
+      expect(Geo::ProjectRegistry.count).to be 4
+      expect(Geo::ProjectRegistry.find_by_id(@orphaned.id)).to be nil
+      expect(Geo::ProjectRegistry.find_by_id(@orphaned1.id)).not_to be nil
+    end
+
+    it 'removes orphaned registries taking into account FROM_PROJECT_ID' do
+      allow(ENV).to receive(:[]).with('FROM_PROJECT_ID').and_return(@orphaned1.project_id)
+      allow(ENV).to receive(:[]).with('TO_PROJECT_ID').and_return(nil)
+
+      run_rake_task('geo:run_orphaned_project_registry_cleaner')
+
+      expect(Geo::ProjectRegistry.count).to be 4
+      expect(Geo::ProjectRegistry.find_by_id(@orphaned.id)).not_to be nil
+      expect(Geo::ProjectRegistry.find_by_id(@orphaned1.id)).to be nil
+    end
+  end
 end