Merge branch...

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

Geo - Improve query to determine job artifacts to sync when selective sync is enabled

Closes #10286

See merge request gitlab-org/gitlab!19583

ee/app/models/geo/fdw/ci/job_artifact.rb

@@ -13,7 +13,7 @@ module Geo
         belongs_to :project, class_name: 'Geo::Fdw::Project', inverse_of: :job_artifacts
 
         scope :not_expired, -> { where('expire_at IS NULL OR expire_at > ?', Time.current) }
-        scope :project_id_in, ->(ids) { joins(:project).merge(Geo::Fdw::Project.id_in(ids)) }
+        scope :project_id_in, ->(ids) { where(project_id: ids) }
 
         class << self
           def inner_join_job_artifact_registry

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

@@ -29,7 +29,20 @@ module Geo
       def job_artifacts
         Geo::Fdw::Ci::JobArtifact.all unless selective_sync?
 
-        Geo::Fdw::Ci::JobArtifact.project_id_in(projects)
+        query = Geo::Fdw::Ci::JobArtifact.project_id_in(projects).select(:id)
+        cte = Gitlab::SQL::CTE.new(:restricted_job_artifacts, query)
+        fdw_job_artifact_table = Geo::Fdw::Ci::JobArtifact.arel_table
+
+        inner_join_restricted_job_artifacts =
+          cte.table
+            .join(fdw_job_artifact_table, Arel::Nodes::InnerJoin)
+            .on(cte.table[:id].eq(fdw_job_artifact_table[:id]))
+            .join_sources
+
+        Geo::Fdw::Ci::JobArtifact
+          .with(cte.to_arel)
+          .from(cte.table)
+          .joins(inner_join_restricted_job_artifacts)
       end
 
       def lfs_objects

ee/app/workers/geo/migrated_local_files_clean_up_worker.rb

@@ -69,7 +69,7 @@ module Geo
       return [] unless job_artifacts_object_store_enabled?
 
       job_artifacts_finder.find_migrated_local(batch_size: batch_size, except_artifact_ids: scheduled_file_ids(:job_artifact))
-                          .pluck(:id)
+                          .pluck(Geo::Fdw::Ci::JobArtifact.arel_table[:id])
                           .map { |id| ['job_artifact', id] }
     end
     # rubocop: enable CodeReuse/ActiveRecord

ee/changelogs/unreleased/10286-geo-sql-statement-times-out-whilst-determining-job-artifacts-to-.yml

+---
+title: Geo - Improve query performance to determine job artifacts to sync when selective
+  sync is enabled
+merge_request: 19583
+author:
+type: performance