Merge branch '53966-make-hashed-storage-migration-safer-and-more-inviting' into 'master'

Hashed Storage rollback mechanism

See merge request gitlab-org/gitlab-ce!23955

3 files changed, 93 insertions, 5 deletions

diff --git a/lib/gitlab/hashed_storage/migrator.rb b/lib/gitlab/hashed_storage/migrator.rb
index bf463077dcc..7046b4e2a43 100644
--- a/lib/gitlab/hashed_storage/migrator.rb
+++ b/lib/gitlab/hashed_storage/migrator.rb
@@ -13,10 +13,18 @@ module Gitlab
       #
       # @param [Integer] start first project id for the range
       # @param [Integer] finish last project id for the range
-      def bulk_schedule(start:, finish:)
+      def bulk_schedule_migration(start:, finish:)
         ::HashedStorage::MigratorWorker.perform_async(start, finish)
       end
 
+      # Schedule a range of projects to be bulk rolledback with #bulk_rollback asynchronously
+      #
+      # @param [Integer] start first project id for the range
+      # @param [Integer] finish last project id for the range
+      def bulk_schedule_rollback(start:, finish:)
+        ::HashedStorage::RollbackerWorker.perform_async(start, finish)
+      end
+
       # Start migration of projects from specified range
       #
       # Flagging a project to be migrated is a synchronous action
@@ -34,6 +42,23 @@ module Gitlab
       end
       # rubocop: enable CodeReuse/ActiveRecord
 
+      # Start rollback of projects from specified range
+      #
+      # Flagging a project to be rolled back is a synchronous action
+      # but the rollback runs through async jobs
+      #
+      # @param [Integer] start first project id for the range
+      # @param [Integer] finish last project id for the range
+      # rubocop: disable CodeReuse/ActiveRecord
+      def bulk_rollback(start:, finish:)
+        projects = build_relation(start, finish)
+
+        projects.with_route.find_each(batch_size: BATCH_SIZE) do |project|
+          rollback(project)
+        end
+      end
+      # rubocop: enable CodeReuse/ActiveRecord
+
       # Flag a project to be migrated to Hashed Storage
       #
       # @param [Project] project that will be migrated
@@ -45,8 +70,15 @@ module Gitlab
         Rails.logger.error("#{err.message} migrating storage of #{project.full_path} (ID=#{project.id}), trace - #{err.backtrace}")
       end
 
+      # Flag a project to be rolled-back to Legacy Storage
+      #
+      # @param [Project] project that will be rolled-back
       def rollback(project)
-        # TODO: implement rollback strategy
+        Rails.logger.info "Starting storage rollback of #{project.full_path} (ID=#{project.id})..."
+
+        project.rollback_to_legacy_storage!
+      rescue => err
+        Rails.logger.error("#{err.message} rolling-back storage of #{project.full_path} (ID=#{project.id}), trace - #{err.backtrace}")
       end
 
       private
diff --git a/lib/gitlab/hashed_storage/rake_helper.rb b/lib/gitlab/hashed_storage/rake_helper.rb
index 38f552fab03..87a31a37e3f 100644
--- a/lib/gitlab/hashed_storage/rake_helper.rb
+++ b/lib/gitlab/hashed_storage/rake_helper.rb
@@ -24,7 +24,7 @@ module Gitlab
       end
 
       # rubocop: disable CodeReuse/ActiveRecord
-      def self.project_id_batches(&block)
+      def self.project_id_batches_migration(&block)
         Project.with_unmigrated_storage.in_batches(of: batch_size, start: range_from, finish: range_to) do |relation| # rubocop: disable Cop/InBatches
           ids = relation.pluck(:id)
 
@@ -34,6 +34,16 @@ module Gitlab
       # rubocop: enable CodeReuse/ActiveRecord
 
       # rubocop: disable CodeReuse/ActiveRecord
+      def self.project_id_batches_rollback(&block)
+        Project.with_storage_feature(:repository).in_batches(of: batch_size, start: range_from, finish: range_to) do |relation| # rubocop: disable Cop/InBatches
+          ids = relation.pluck(:id)
+
+          yield ids.min, ids.max
+        end
+      end
+      # rubocop: enable CodeReuse/ActiveRecord
+
+      # rubocop: disable CodeReuse/ActiveRecord
       def self.legacy_attachments_relation
         Upload.joins(<<~SQL).where('projects.storage_version < :version OR projects.storage_version IS NULL', version: Project::HASHED_STORAGE_FEATURES[:attachments])
           JOIN projects
diff --git a/lib/tasks/gitlab/storage.rake b/lib/tasks/gitlab/storage.rake
index f9ce3e1d338..a2136ce1b92 100644
--- a/lib/tasks/gitlab/storage.rake
+++ b/lib/tasks/gitlab/storage.rake
@@ -36,8 +36,54 @@ namespace :gitlab do
 
       print "Enqueuing migration of #{legacy_projects_count} projects in batches of #{helper.batch_size}"
 
-      helper.project_id_batches do |start, finish|
-        storage_migrator.bulk_schedule(start: start, finish: finish)
+      helper.project_id_batches_migration do |start, finish|
+        storage_migrator.bulk_schedule_migration(start: start, finish: finish)
+
+        print '.'
+      end
+
+      puts ' Done!'
+    end
+
+    desc 'GitLab | Storage | Rollback existing projects to Legacy Storage'
+    task rollback_to_legacy: :environment do
+      if Gitlab::Database.read_only?
+        warn 'This task requires database write access. Exiting.'
+
+        next
+      end
+
+      storage_migrator = Gitlab::HashedStorage::Migrator.new
+      helper = Gitlab::HashedStorage::RakeHelper
+
+      if helper.range_single_item?
+        project = Project.with_storage_feature(:repository).find_by(id: helper.range_from)
+
+        unless project
+          warn "There are no projects that can be rolledback with ID=#{helper.range_from}"
+
+          next
+        end
+
+        puts "Enqueueing storage rollback of #{project.full_path} (ID=#{project.id})..."
+        storage_migrator.rollback(project)
+
+        next
+      end
+
+      hashed_projects_count = Project.with_storage_feature(:repository).count
+
+      if hashed_projects_count == 0
+        warn 'There are no projects that can have storage rolledback. Nothing to do!'
+
+        next
+      end
+
+      print "Enqueuing rollback of #{hashed_projects_count} projects in batches of #{helper.batch_size}"
+
+      helper.project_id_batches_rollback do |start, finish|
+        puts "Start: #{start} FINISH: #{finish}"
+        storage_migrator.bulk_schedule_rollback(start: start, finish: finish)
 
         print '.'
       end