1 files changed, 44 insertions, 24 deletions

diff --git a/lib/tasks/gitlab/pages.rake b/lib/tasks/gitlab/pages.rake
index e15cbb4e32e..b598dab901d 100644
--- a/lib/tasks/gitlab/pages.rake
+++ b/lib/tasks/gitlab/pages.rake
@@ -1,35 +1,55 @@
+# frozen_string_literal: true
+
 require 'logger'
 
 namespace :gitlab do
   namespace :pages do
     desc "GitLab | Pages | Migrate legacy storage to zip format"
     task migrate_legacy_storage: :gitlab_environment do
-      logger = Logger.new(STDOUT)
       logger.info('Starting to migrate legacy pages storage to zip deployments')
-      processed_projects = 0
-
-      ProjectPagesMetadatum.only_on_legacy_storage.each_batch(of: 10) do |batch|
-        batch.preload(project: [:namespace, :route, pages_metadatum: :pages_deployment]).each do |metadatum|
-          project = metadatum.project
-
-          result = nil
-          time = Benchmark.realtime do
-            result = ::Pages::MigrateLegacyStorageToDeploymentService.new(project).execute
-          end
-          processed_projects += 1
-
-          if result[:status] == :success
-            logger.info("project_id: #{project.id} #{project.pages_path} has been migrated in #{time} seconds")
-          else
-            logger.error("project_id: #{project.id} #{project.pages_path} failed to be migrated in #{time} seconds: #{result[:message]}")
-          end
-        rescue => e
-          logger.error("#{e.message} project_id: #{project&.id}")
-          Gitlab::ErrorTracking.track_exception(e, project_id: project&.id)
-        end
-
-        logger.info("#{processed_projects} pages projects are processed")
+
+      result = ::Pages::MigrateFromLegacyStorageService.new(logger,
+                                                            migration_threads: migration_threads,
+                                                            batch_size: batch_size,
+                                                            ignore_invalid_entries: ignore_invalid_entries).execute
+
+      logger.info("A total of #{result[:migrated] + result[:errored]} projects were processed.")
+      logger.info("- The #{result[:migrated]} projects migrated successfully")
+      logger.info("- The #{result[:errored]} projects failed to be migrated")
+    end
+
+    desc "GitLab | Pages | DANGER: Removes data which was migrated from legacy storage on zip storage. Can be used if some bugs in migration are discovered and migration needs to be restarted from scratch."
+    task clean_migrated_zip_storage: :gitlab_environment do
+      destroyed_deployments = 0
+
+      logger.info("Starting to delete migrated pages deployments")
+
+      ::PagesDeployment.migrated_from_legacy_storage.each_batch(of: batch_size) do |batch|
+        destroyed_deployments += batch.count
+
+        # we need to destroy associated files, so can't use delete_all
+        batch.destroy_all # rubocop: disable Cop/DestroyAll
+
+        logger.info("#{destroyed_deployments} deployments were deleted")
       end
     end
+
+    def logger
+      @logger ||= Logger.new(STDOUT)
+    end
+
+    def migration_threads
+      ENV.fetch('PAGES_MIGRATION_THREADS', '3').to_i
+    end
+
+    def batch_size
+      ENV.fetch('PAGES_MIGRATION_BATCH_SIZE', '10').to_i
+    end
+
+    def ignore_invalid_entries
+      Gitlab::Utils.to_boolean(
+        ENV.fetch('PAGES_MIGRATION_IGNORE_INVALID_ENTRIES', 'false')
+      )
+    end
   end
 end