FIX: Split backfill into separate migrations to use independent transactions (#1063)

db/migrate/20241230153300_new_embeddings_tables.rb

@@ -65,22 +65,6 @@ class NewEmbeddingsTables < ActiveRecord::Migration[7.2]
         WHERE model_id = #{model_id} AND strategy_id = 1;
       SQL
     end
-
-    # Copy data from old tables to new tables
-    execute <<~SQL
-      INSERT INTO ai_topics_embeddings (topic_id, model_id, model_version, strategy_id, strategy_version, digest, embeddings, created_at, updated_at)
-      SELECT * FROM ai_topic_embeddings;
-    SQL
-
-    execute <<~SQL
-      INSERT INTO ai_posts_embeddings (post_id, model_id, model_version, strategy_id, strategy_version, digest, embeddings, created_at, updated_at)
-      SELECT * FROM ai_post_embeddings;
-    SQL
-
-    execute <<~SQL
-      INSERT INTO ai_document_fragments_embeddings (rag_document_fragment_id, model_id, model_version, strategy_id, strategy_version, digest, embeddings, created_at, updated_at)
-      SELECT * FROM ai_document_fragment_embeddings;
-    SQL
   end
 
   def down

db/migrate/20250114160417_backfill_topic_embeddings.rb

@@ -0,0 +1,18 @@
+# frozen_string_literal: true
+class BackfillTopicEmbeddings < ActiveRecord::Migration[7.2]
+  def up
+    not_backfilled = DB.query_single("SELECT COUNT(*) FROM ai_topics_embeddings").first.to_i == 0
+
+    if not_backfilled
+      # Copy data from old tables to new tables
+      execute <<~SQL
+        INSERT INTO ai_topics_embeddings (topic_id, model_id, model_version, strategy_id, strategy_version, digest, embeddings, created_at, updated_at)
+        SELECT * FROM ai_topic_embeddings;
+      SQL
+    end
+  end
+
+  def down
+    raise ActiveRecord::IrreversibleMigration
+  end
+end

db/migrate/20250114160446_backfill_post_embeddings.rb

@@ -0,0 +1,18 @@
+# frozen_string_literal: true
+class BackfillPostEmbeddings < ActiveRecord::Migration[7.2]
+  def up
+    not_backfilled = DB.query_single("SELECT COUNT(*) FROM ai_posts_embeddings").first.to_i == 0
+
+    if not_backfilled
+      # Copy data from old tables to new tables
+      execute <<~SQL
+        INSERT INTO ai_posts_embeddings (post_id, model_id, model_version, strategy_id, strategy_version, digest, embeddings, created_at, updated_at)
+        SELECT * FROM ai_post_embeddings;
+      SQL
+    end
+  end
+
+  def down
+    raise ActiveRecord::IrreversibleMigration
+  end
+end

db/migrate/20250114160500_backfill_rag_embeddings.rb

@@ -0,0 +1,19 @@
+# frozen_string_literal: true
+class BackfillRagEmbeddings < ActiveRecord::Migration[7.2]
+  def up
+    not_backfilled =
+      DB.query_single("SELECT COUNT(*) FROM ai_document_fragments_embeddings").first.to_i == 0
+
+    if not_backfilled
+      # Copy data from old tables to new tables
+      execute <<~SQL
+        INSERT INTO ai_document_fragments_embeddings (rag_document_fragment_id, model_id, model_version, strategy_id, strategy_version, digest, embeddings, created_at, updated_at)
+        SELECT * FROM ai_document_fragment_embeddings;
+      SQL
+    end
+  end
+
+  def down
+    raise ActiveRecord::IrreversibleMigration
+  end
+end

db/post_migrate/20250113171444_drop_old_embedding_tables.rb

@@ -2,48 +2,7 @@
 class DropOldEmbeddingTables < ActiveRecord::Migration[7.2]
   def up
     # Copy rag embeddings created during deploy.
-    execute <<~SQL
-      INSERT INTO ai_document_fragments_embeddings (rag_document_fragment_id, model_id, model_version, strategy_id, strategy_version, digest, embeddings, created_at, updated_at)
-      (
-        SELECT  ai_document_fragment_embeddings.*
-        FROM ai_document_fragment_embeddings 
-        LEFT OUTER JOIN ai_document_fragments_embeddings ON ai_document_fragment_embeddings.rag_document_fragment_id = ai_document_fragments_embeddings.rag_document_fragment_id
-        WHERE ai_document_fragments_embeddings.rag_document_fragment_id IS NULL
-      )
-    SQL
-
-    execute <<~SQL
-      DROP INDEX IF EXISTS ai_topic_embeddings_1_1_search_bit;
-      DROP INDEX IF EXISTS ai_topic_embeddings_2_1_search_bit;
-      DROP INDEX IF EXISTS ai_topic_embeddings_3_1_search_bit;
-      DROP INDEX IF EXISTS ai_topic_embeddings_4_1_search_bit;
-      DROP INDEX IF EXISTS ai_topic_embeddings_5_1_search_bit;
-      DROP INDEX IF EXISTS ai_topic_embeddings_6_1_search_bit;
-      DROP INDEX IF EXISTS ai_topic_embeddings_7_1_search_bit;
-      DROP INDEX IF EXISTS ai_topic_embeddings_8_1_search_bit;
-
-      DROP INDEX IF EXISTS ai_post_embeddings_1_1_search_bit;
-      DROP INDEX IF EXISTS ai_post_embeddings_2_1_search_bit;
-      DROP INDEX IF EXISTS ai_post_embeddings_3_1_search_bit;
-      DROP INDEX IF EXISTS ai_post_embeddings_4_1_search_bit;
-      DROP INDEX IF EXISTS ai_post_embeddings_5_1_search_bit;
-      DROP INDEX IF EXISTS ai_post_embeddings_6_1_search_bit;
-      DROP INDEX IF EXISTS ai_post_embeddings_7_1_search_bit;
-      DROP INDEX IF EXISTS ai_post_embeddings_8_1_search_bit;
-
-      DROP INDEX IF EXISTS ai_document_fragment_embeddings_1_1_search_bit;
-      DROP INDEX IF EXISTS ai_document_fragment_embeddings_2_1_search_bit;
-      DROP INDEX IF EXISTS ai_document_fragment_embeddings_3_1_search_bit;
-      DROP INDEX IF EXISTS ai_document_fragment_embeddings_4_1_search_bit;
-      DROP INDEX IF EXISTS ai_document_fragment_embeddings_5_1_search_bit;
-      DROP INDEX IF EXISTS ai_document_fragment_embeddings_6_1_search_bit;
-      DROP INDEX IF EXISTS ai_document_fragment_embeddings_7_1_search_bit;
-      DROP INDEX IF EXISTS ai_document_fragment_embeddings_8_1_search_bit;
-    SQL
-
-    drop_table :ai_topic_embeddings
-    drop_table :ai_post_embeddings
-    drop_table :ai_document_fragment_embeddings
+    # noop. TODO(roman): Will follow-up with a new migration to drop these tables.
   end
 
   def down