Convert migration to populate latest merge request ID into a background migration

This is to smear updates over a few hours to avoid causing excessive
replication lag as seen in https://gitlab.com/gitlab-com/infrastructure/issues/3235.

1 files changed, 30 insertions, 0 deletions

diff --git a/lib/gitlab/background_migration/populate_merge_requests_latest_merge_request_diff_id.rb b/lib/gitlab/background_migration/populate_merge_requests_latest_merge_request_diff_id.rb
new file mode 100644
index 00000000000..7e109e96e73
--- /dev/null
+++ b/lib/gitlab/background_migration/populate_merge_requests_latest_merge_request_diff_id.rb
@@ -0,0 +1,30 @@
+module Gitlab
+  module BackgroundMigration
+    class PopulateMergeRequestsLatestMergeRequestDiffId
+      BATCH_SIZE = 1_000
+
+      class MergeRequest < ActiveRecord::Base
+        self.table_name = 'merge_requests'
+
+        include ::EachBatch
+      end
+
+      def perform(start_id, stop_id)
+        update = '
+          latest_merge_request_diff_id = (
+            SELECT MAX(id)
+            FROM merge_request_diffs
+            WHERE merge_requests.id = merge_request_diffs.merge_request_id
+          )'.squish
+
+        MergeRequest
+          .where(id: start_id..stop_id)
+          .where(latest_merge_request_diff_id: nil)
+          .each_batch(of: BATCH_SIZE) do |relation|
+
+          relation.update_all(update)
+        end
+      end
+    end
+  end
+end