PERF: reindex search data without loading large post counts

app/jobs/scheduled/reindex_search.rb

@@ -10,41 +10,46 @@ module Jobs
       rebuild_problem_users
     end
 
-    def rebuild_problem_categories(limit = 10000)
+    def rebuild_problem_categories(limit = 500)
-      categories = load_problem_categories(limit)
+      category_ids = load_problem_category_ids(limit)
 
-      categories.each do |category|
+      category_ids.each do |id|
-        SearchIndexer.index(category, force: true)
+        category = Category.find_by(id: id)
+        SearchIndexer.index(category, force: true) if category
       end
     end
 
     def rebuild_problem_users(limit = 10000)
-      users = load_problem_users(limit)
+      user_ids = load_problem_user_ids(limit)
 
-      users.each do |user|
+      user_ids.each do |id|
-        SearchIndexer.index(user, force: true)
+        user = User.find_by(id: id)
+        SearchIndexer.index(user, force: true) if user
       end
     end
 
     def rebuild_problem_topics(limit = 10000)
-      topics = load_problem_topics(limit)
+      topic_ids = load_problem_topic_ids(limit)
 
-      topics.each do |topic|
+      topic_ids.each do |id|
-        SearchIndexer.index(topic, force: true)
+        topic = Topic.find_by(id: id)
+        SearchIndexer.index(topic, force: true) if topic
       end
     end
 
     def rebuild_problem_posts(limit = 10000)
-      posts = load_problem_posts(limit)
+      post_ids = load_problem_post_ids(limit)
 
-      posts.each do |post|
+      post_ids.each do |id|
-        SearchIndexer.index(post, force: true)
+        post = Post.find_by(id: id)
+        # could be deleted while iterating through batch
+        SearchIndexer.index(post, force: true) if post
       end
     end
 
     private
 
-    def load_problem_posts(limit)
+    def load_problem_post_ids(limit)
       Post.joins(:topic)
         .where('posts.id IN (
                 SELECT p2.id FROM posts p2
@@ -52,27 +57,31 @@ module Jobs
                 WHERE pd.post_id IS NULL
                 )', SiteSetting.default_locale, Search::INDEX_VERSION)
         .limit(limit)
+        .pluck(:id)
     end
 
-    def load_problem_categories(limit)
+    def load_problem_category_ids(limit)
       Category.joins(:category_search_data)
         .where('category_search_data.locale != ?
                 OR category_search_data.version != ?', SiteSetting.default_locale, Search::INDEX_VERSION)
         .limit(limit)
+        .pluck(:id)
     end
 
-    def load_problem_topics(limit)
+    def load_problem_topic_ids(limit)
       Topic.joins(:topic_search_data)
         .where('topic_search_data.locale != ?
                 OR topic_search_data.version != ?', SiteSetting.default_locale, Search::INDEX_VERSION)
         .limit(limit)
+        .pluck(:id)
     end
 
-    def load_problem_users(limit)
+    def load_problem_user_ids(limit)
       User.joins(:user_search_data)
         .where('user_search_data.locale != ?
                 OR user_search_data.version != ?', SiteSetting.default_locale, Search::INDEX_VERSION)
         .limit(limit)
+        .pluck(:id)
     end
   end
 end