From 61f8502f730a40c8f37c424afe1f301ff482db10 Mon Sep 17 00:00:00 2001 From: Peter Larsen Date: Fri, 21 Aug 2026 20:14:17 -0400 Subject: [PATCH] cluster spec sheet: OPTIMIZE TABLE after seeding the MySQL source table The QA table now carries a PRIMARY KEY, and the seeding INSERT emits customer_id in scattered order; inserted out of key order, the clustered index page-splits to ~2.5x its packed size (11.5 GiB instead of 4.6 GiB for 50M rows), which is what doubled the source_ingestion hydration measurement from ~4m to ~8m after the July 24 re-seed. OPTIMIZE TABLE rebuilds it packed. Co-Authored-By: Claude Fable 5 --- test/cluster-spec-sheet/mzcompose.py | 1 + 1 file changed, 1 insertion(+) diff --git a/test/cluster-spec-sheet/mzcompose.py b/test/cluster-spec-sheet/mzcompose.py index c02e246a7af7b..40e6bc0ee148f 100644 --- a/test/cluster-spec-sheet/mzcompose.py +++ b/test/cluster-spec-sheet/mzcompose.py @@ -2161,6 +2161,7 @@ def setup(self) -> list[str]: # UNION ALL SELECT 5 UNION ALL SELECT 6 UNION ALL SELECT 7 UNION ALL SELECT 8 UNION ALL SELECT 9) h # ) nums # WHERE n BETWEEN 1 AND 50000000; + # OPTIMIZE TABLE tbl; # Kafka (Confluent) # node /usr/local/bin/datagen -f avro -n 50000000 -w 0 -p qa_cluster_spec_sheet -s table.json