Add methods to find container registries to sync

ee/app/finders/geo/container_repository_registry_finder.rb

@@ -30,6 +30,42 @@ module Geo
       [untracked_ids, unused_tracked_ids]
     end
 
+    # Returns Geo::ContainerRepositoryRegistry records that have never been synced.
+    #
+    # Does not care about selective sync, because it considers the Registry
+    # table to be the single source of truth. The contract is that other
+    # processes need to ensure that the table only contains records that should
+    # be synced.
+    #
+    # Any registries that have ever been synced that currently need to be
+    # resynced will be handled by other find methods (like
+    # #find_retryable_dirty_registries)
+    #
+    # You can pass a list with `except_ids:` so you can exclude items you
+    # already scheduled but haven't finished and aren't persisted to the database yet
+    #
+    # @param [Integer] batch_size used to limit the results returned
+    # @param [Array<Integer>] except_ids ids that will be ignored from the query
+    # rubocop:disable CodeReuse/ActiveRecord
+    def find_never_synced_registries(batch_size:, except_ids: [])
+      Geo::ContainerRepositoryRegistry
+        .never_synced
+        .model_id_not_in(except_ids)
+        .limit(batch_size)
+    end
+    # rubocop:enable CodeReuse/ActiveRecord
+
+    # rubocop:disable CodeReuse/ActiveRecord
+    def find_retryable_dirty_registries(batch_size:, except_ids: [])
+      Geo::ContainerRepositoryRegistry
+        .failed
+        .retry_due
+        .model_id_not_in(except_ids)
+        .order(Gitlab::Database.nulls_first_order(:last_synced_at))
+        .limit(batch_size)
+    end
+    # rubocop:enable CodeReuse/ActiveRecord
+
     # Find limited amount of non replicated container repositories.
     #
     # You can pass a list with `except_repository_ids:` so you can exclude items you
@@ -37,16 +73,16 @@ module Geo
     #
     # @param [Integer] batch_size used to limit the results returned
     # @param [Array<Integer>] except_repository_ids ids that will be ignored from the query
-    # rubocop: disable CodeReuse/ActiveRecord
+    # rubocop:disable CodeReuse/ActiveRecord
     def find_unsynced(batch_size:, except_repository_ids: [])
       container_repositories
         .missing_container_repository_registry
         .id_not_in(except_repository_ids)
         .limit(batch_size)
     end
-    # rubocop: enable CodeReuse/ActiveRecord
+    # rubocop:enable CodeReuse/ActiveRecord
 
-    # rubocop: disable CodeReuse/ActiveRecord
+    # rubocop:disable CodeReuse/ActiveRecord
     def find_retryable_failed_ids(batch_size:, except_repository_ids: [])
       Geo::ContainerRepositoryRegistry
         .failed
@@ -55,7 +91,7 @@ module Geo
         .limit(batch_size)
         .pluck_container_repository_key
     end
-    # rubocop: enable CodeReuse/ActiveRecord
+    # rubocop:enable CodeReuse/ActiveRecord
 
     private
 

ee/app/models/geo/container_repository_registry.rb

@@ -8,6 +8,7 @@ class Geo::ContainerRepositoryRegistry < Geo::BaseRegistry
 
   belongs_to :container_repository
 
+  scope :never_synced, -> { with_state(:pending).where(last_synced_at: nil) }
   scope :failed, -> { with_state(:failed) }
   scope :synced, -> { with_state(:synced) }
   scope :retry_due, -> { where(arel_table[:retry_at].eq(nil).or(arel_table[:retry_at].lt(Time.current))) }