Merge branch...

Merge branch '10286-geo-sql-statement-times-out-whilst-determining-lfs-objects-to-sync' into 'master'

Geo - Rewrite query to determine LFS objects to sync with selective sync

Closes #10286

See merge request gitlab-org/gitlab!19051

ee/app/models/geo/fdw/geo_node.rb

@@ -35,7 +35,20 @@ module Geo
       def lfs_objects
         return Geo::Fdw::LfsObject.all unless selective_sync?
 
-        Geo::Fdw::LfsObject.project_id_in(projects)
+        query = Geo::Fdw::LfsObjectsProject.project_id_in(projects).select(:lfs_object_id)
+        cte = Gitlab::SQL::CTE.new(:restricted_lfs_objects, query)
+        fdw_lfs_object_table = Geo::Fdw::LfsObject.arel_table
+
+        inner_join_restricted_lfs_objects =
+          cte.table
+            .join(fdw_lfs_object_table, Arel::Nodes::InnerJoin)
+            .on(cte.table[:lfs_object_id].eq(fdw_lfs_object_table[:id]))
+            .join_sources
+
+        Geo::Fdw::LfsObject
+          .with(cte.to_arel)
+          .from(cte.table)
+          .joins(inner_join_restricted_lfs_objects)
       end
 
       def lfs_object_registries

ee/app/models/geo/fdw/lfs_objects_project.rb

@@ -7,6 +7,8 @@ module Geo
 
       belongs_to :lfs_object, class_name: 'Geo::Fdw::LfsObject', inverse_of: :projects
       belongs_to :project, class_name: 'Geo::Fdw::Project', inverse_of: :lfs_objects
+
+      scope :project_id_in, ->(ids) { joins(:project).merge(Geo::Fdw::Project.id_in(ids)) }
     end
   end
 end

ee/app/workers/geo/file_download_dispatch_worker/attachment_job_finder.rb

@@ -15,7 +15,7 @@ module Geo
       # rubocop: disable CodeReuse/ActiveRecord
       def convert_resource_relation_to_job_args(relation)
         relation.pluck(:id, :uploader)
-                .map { |id, uploader| [uploader.sub(/Uploader\z/, '').underscore, id] }
+                .map! { |id, uploader| [uploader.sub(/Uploader\z/, '').underscore, id] }
       end
       # rubocop: enable CodeReuse/ActiveRecord
 

ee/app/workers/geo/file_download_dispatch_worker/job_finder.rb

@@ -50,13 +50,13 @@ module Geo
 
       # rubocop: disable CodeReuse/ActiveRecord
       def convert_resource_relation_to_job_args(relation)
-        relation.pluck(:id).map { |id| [self.class::FILE_SERVICE_OBJECT_TYPE.to_s, id] }
+        relation.pluck(relation.model.arel_table[:id]).map! { |id| [self.class::FILE_SERVICE_OBJECT_TYPE.to_s, id] }
       end
       # rubocop: enable CodeReuse/ActiveRecord
 
       # rubocop: disable CodeReuse/ActiveRecord
       def convert_registry_relation_to_job_args(relation)
-        relation.pluck(self.class::RESOURCE_ID_KEY).map { |id| [self.class::FILE_SERVICE_OBJECT_TYPE.to_s, id] }
+        relation.pluck(self.class::RESOURCE_ID_KEY).map! { |id| [self.class::FILE_SERVICE_OBJECT_TYPE.to_s, id] }
       end
       # rubocop: enable CodeReuse/ActiveRecord
     end

ee/changelogs/unreleased/10286-geo-sql-statement-times-out-whilst-determining-lfs-objects-to-sy.yml

+---
+title: Geo - Improve query performance to determine LFS objects to sync when selective
+  sync is enabled
+merge_request: 19051
+author:
+type: performance