From f93d15f78124b25e70fb2f61a837c878965a66b6 Mon Sep 17 00:00:00 2001 From: Bodobolero Date: Thu, 21 Dec 2023 13:34:31 +0100 Subject: [PATCH] add comment to run vacuum for clickbench (#6212) ## Problem This is a comment only change. To ensure that our benchmarking results are fair we need to have correct stats in catalog. Otherwise optimizer chooses seq scan instead of index only scan for some queries. Added comment to run vacuum after data prep. --- test_runner/performance/test_perf_olap.py | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/test_runner/performance/test_perf_olap.py b/test_runner/performance/test_perf_olap.py index 1de7e95bbe..1e6e9a0174 100644 --- a/test_runner/performance/test_perf_olap.py +++ b/test_runner/performance/test_perf_olap.py @@ -151,7 +151,9 @@ def test_clickbench(query: LabelledQuery, remote_compare: RemoteCompare, scale: An OLAP-style ClickHouse benchmark Based on https://github.com/ClickHouse/ClickBench/tree/c00135ca5b6a0d86fedcdbf998fdaa8ed85c1c3b/aurora-postgresql - The DB prepared manually in advance + The DB prepared manually in advance. + Important: after intial data load, run `VACUUM (DISABLE_PAGE_SKIPPING, FREEZE, ANALYZE) hits;` + to ensure that Postgres optimizer chooses the same plans as RDS and Aurora. """ explain: bool = os.getenv("TEST_OLAP_COLLECT_EXPLAIN", "false").lower() == "true"