chore: do not drop database first for bench (#15041)

Author: everpcpc

.github/workflows/reuse.benchmark.yml

@@ -168,7 +168,7 @@ jobs:
         env:
           BENDSQL_DSN: "databend://${{ secrets.BENCHMARK_CLOUD_USER }}:${{ secrets.BENCHMARK_CLOUD_PASSWORD }}@${{ secrets.BENCHMARK_CLOUD_GATEWAY }}:443/?warehouse=default"
         run: |
-          echo "DROP DATABASE IF EXISTS 'load_test_${{ inputs.run_id }}';" | bendsql
+          echo "DROP DATABASE IF EXISTS load_test_${{ inputs.run_id }};" | bendsql
           echo "DROP WAREHOUSE IF EXISTS 'benchmark-${{ inputs.run_id }}';" | bendsql
 
   comment:

benchmark/clickbench/benchmark_cloud.sh

@@ -75,7 +75,6 @@ export BENDSQL_DSN="databend://${CLOUD_USER}:${CLOUD_PASSWORD}@${CLOUD_GATEWAY}:
 
 if [[ "${BENCHMARK_DATASET}" == "load" ]]; then
     echo "Creating database..."
-    echo "DROP DATABASE IF EXISTS ${BENCHMARK_DATABASE};" | bendsql --database default
     echo "CREATE DATABASE ${BENCHMARK_DATABASE};" | bendsql --database default
 fi
 

benchmark/clickbench/load/queries/00.sql

@@ -1,5 +1,5 @@
-DROP TABLE IF EXISTS hits_csv;
-CREATE TABLE hits_csv (
+DROP TABLE IF EXISTS hits_parquet;
+CREATE TABLE hits_parquet (
     WatchID BIGINT NOT NULL,
     JavaEnable SMALLINT NOT NULL,
     Title TEXT,
@@ -106,5 +106,5 @@ CREATE TABLE hits_csv (
     URLHash BIGINT NOT NULL,
     CLID INTEGER NOT NULL
 ) CLUSTER BY (CounterID, EventDate, UserID, EventTime, WatchID);
-COPY INTO hits_csv
-FROM 's3://databend-datasets/hits_compatible/hits.csv.gz' CONNECTION = (CONNECTION_NAME = 'repo') FILE_FORMAT = (TYPE = 'CSV', COMPRESSION = AUTO);
+COPY INTO hits_parquet
+FROM 's3://databend-datasets/hits_compatible/hits.parquet' CONNECTION = (CONNECTION_NAME = 'repo') FILE_FORMAT = (TYPE = 'PARQUET');

benchmark/clickbench/load/queries/01.sql

@@ -1,5 +1,5 @@
-DROP TABLE IF EXISTS hits_tsv;
-CREATE TABLE hits_tsv (
+DROP TABLE IF EXISTS hits_csv;
+CREATE TABLE hits_csv (
     WatchID BIGINT NOT NULL,
     JavaEnable SMALLINT NOT NULL,
     Title TEXT,
@@ -106,5 +106,5 @@ CREATE TABLE hits_tsv (
     URLHash BIGINT NOT NULL,
     CLID INTEGER NOT NULL
 ) CLUSTER BY (CounterID, EventDate, UserID, EventTime, WatchID);
-COPY INTO hits_tsv
-FROM 's3://databend-datasets/hits_compatible/hits.tsv.gz' CONNECTION = (CONNECTION_NAME = 'repo') FILE_FORMAT = (TYPE = 'TSV', COMPRESSION = AUTO);
+COPY INTO hits_csv
+FROM 's3://databend-datasets/hits_compatible/hits.csv.gz' CONNECTION = (CONNECTION_NAME = 'repo') FILE_FORMAT = (TYPE = 'CSV', COMPRESSION = AUTO);

benchmark/clickbench/load/queries/02.sql

@@ -1,5 +1,5 @@
-DROP TABLE IF EXISTS hits_json;
-CREATE TABLE hits_json (
+DROP TABLE IF EXISTS hits_tsv;
+CREATE TABLE hits_tsv (
     WatchID BIGINT NOT NULL,
     JavaEnable SMALLINT NOT NULL,
     Title TEXT,
@@ -106,5 +106,5 @@ CREATE TABLE hits_json (
     URLHash BIGINT NOT NULL,
     CLID INTEGER NOT NULL
 ) CLUSTER BY (CounterID, EventDate, UserID, EventTime, WatchID);
-COPY INTO hits_json
-FROM 's3://databend-datasets/hits_compatible/hits.json.gz' CONNECTION = (CONNECTION_NAME = 'repo') FILE_FORMAT = (TYPE = 'NDJSON', COMPRESSION = AUTO);
+COPY INTO hits_tsv
+FROM 's3://databend-datasets/hits_compatible/hits.tsv.gz' CONNECTION = (CONNECTION_NAME = 'repo') FILE_FORMAT = (TYPE = 'TSV', COMPRESSION = AUTO);