Merge branch '322788-speed-up-backfill-mr-permissions-elasticsearch-migration' into 'master'

Speed up Elasticsearch AddNewDataToMergeRequestsDocuments migration

See merge request gitlab-org/gitlab!60563

ee/changelogs/unreleased/322788-speed-up-backfill-mr-permissions-elasticsearch-migration.yml

+---
+title: Speed up Elastic AddNewDataToMergeRequestsDocuments migration
+merge_request: 60563
+author:
+type: changed

ee/elastic/migrate/20210421140400_add_new_data_to_merge_requests_documents.rb

@@ -2,9 +2,9 @@
 
 class AddNewDataToMergeRequestsDocuments < Elastic::Migration
   batched!
-  throttle_delay 5.minutes
+  throttle_delay 3.minutes
 
-  QUERY_BATCH_SIZE = 5000
+  QUERY_BATCH_SIZE = 6000
   UPDATE_BATCH_SIZE = 100
 
   def migrate
@@ -38,7 +38,7 @@ class AddNewDataToMergeRequestsDocuments < Elastic::Migration
     end
 
     document_references.each_slice(UPDATE_BATCH_SIZE) do |refs|
-      Elastic::ProcessBookkeepingService.track!(*refs)
+      Elastic::ProcessInitialBookkeepingService.track!(*refs)
     end
 
     log "Adding visibility_level field to merge_requests documents is completed for batch of #{document_references.size} documents"

ee/spec/elastic/migrate/20210421140400_add_new_data_to_merge_requests_documents_spec.rb

@@ -20,7 +20,7 @@ RSpec.describe AddNewDataToMergeRequestsDocuments, :elastic, :sidekiq_inline do
   describe 'migration_options' do
     it 'has migration options set', :aggregate_failures do
       expect(migration.batched?).to be_truthy
-      expect(migration.throttle_delay).to eq(5.minutes)
+      expect(migration.throttle_delay).to eq(3.minutes)
     end
   end
 
@@ -33,7 +33,7 @@ RSpec.describe AddNewDataToMergeRequestsDocuments, :elastic, :sidekiq_inline do
       end
 
       it 'does not modify data', :aggregate_failures do
-        expect(::Elastic::ProcessBookkeepingService).not_to receive(:track!)
+        expect(::Elastic::ProcessInitialBookkeepingService).not_to receive(:track!)
 
         subject
       end
@@ -46,7 +46,7 @@ RSpec.describe AddNewDataToMergeRequestsDocuments, :elastic, :sidekiq_inline do
 
       it 'updates all merge_request documents' do
         # track calls are batched in groups of 100
-        expect(::Elastic::ProcessBookkeepingService).to receive(:track!).once do |*tracked_refs|
+        expect(::Elastic::ProcessInitialBookkeepingService).to receive(:track!).once do |*tracked_refs|
           expect(tracked_refs.count).to eq(3)
         end
 
@@ -58,7 +58,7 @@ RSpec.describe AddNewDataToMergeRequestsDocuments, :elastic, :sidekiq_inline do
         add_visibility_level_for_merge_requests(merge_requests[1..-1])
 
         expected = [Gitlab::Elastic::DocumentReference.new(MergeRequest, merge_request.id, merge_request.es_id, merge_request.es_parent)]
-        expect(::Elastic::ProcessBookkeepingService).to receive(:track!).with(*expected).once
+        expect(::Elastic::ProcessInitialBookkeepingService).to receive(:track!).with(*expected).once
 
         subject
       end
@@ -67,7 +67,7 @@ RSpec.describe AddNewDataToMergeRequestsDocuments, :elastic, :sidekiq_inline do
         stub_const("#{described_class}::QUERY_BATCH_SIZE", 2)
         stub_const("#{described_class}::UPDATE_BATCH_SIZE", 1)
 
-        expect(::Elastic::ProcessBookkeepingService).to receive(:track!).exactly(3).times.and_call_original
+        expect(::Elastic::ProcessInitialBookkeepingService).to receive(:track!).exactly(3).times.and_call_original
 
         # cannot use subject in spec because it is memoized
         migration.migrate