skip unsupported data types

Signed-off-by: Murphy <[email protected]>

be/src/storage/rowset/data_sample.cpp

@@ -22,6 +22,7 @@
 #include "storage/olap_common.h"
 #include "storage/types.h"
 #include "storage/zone_map_detail.h"
+#include "types/logical_type.h"
 #include "util/runtime_profile.h"
 
 namespace starrocks {
@@ -86,6 +87,28 @@ double SortableZoneMap::width(const ZoneMapDetail& zone) {
     return width(zone.min_value(), zone.max_value());
 }
 
+bool SortableZoneMap::is_support_data_type(LogicalType type) {
+    // TODO: support decimal type
+    switch (type) {
+    case TYPE_TINYINT:
+    case TYPE_SMALLINT:
+    case TYPE_INT:
+    case TYPE_BIGINT:
+    case TYPE_LARGEINT:
+    case TYPE_UNSIGNED_TINYINT:
+    case TYPE_UNSIGNED_SMALLINT:
+    case TYPE_UNSIGNED_INT:
+    case TYPE_UNSIGNED_BIGINT:
+    case TYPE_FLOAT:
+    case TYPE_DOUBLE:
+    case TYPE_DATE:
+    case TYPE_DATETIME:
+        return true;
+    default:
+        return false;
+    }
+}
+
 double SortableZoneMap::width(const Datum& lhs, const Datum& rhs) {
     if (lhs.is_null() || rhs.is_null()) {
         return 0;
@@ -113,11 +136,6 @@ double SortableZoneMap::width(const Datum& lhs, const Datum& rhs) {
         return rhs.get_float() - lhs.get_float();
     case TYPE_DOUBLE:
         return rhs.get_double() - lhs.get_double();
-    case TYPE_DECIMAL:
-    case TYPE_DECIMAL32:
-    case TYPE_DECIMAL64:
-    case TYPE_DECIMAL128:
-        return rhs.get_decimal() - lhs.get_decimal();
     case TYPE_DATE:
         return rhs.get_date().julian() - lhs.get_date().julian();
     case TYPE_DATETIME:

be/src/storage/rowset/data_sample.h

@@ -82,6 +82,8 @@ struct SortableZoneMap {
     std::vector<size_t> page_indices;        // Keep zonemap immutable, sort by this page_indices
     std::vector<PagesWithZoneMap> histogram; // Only exist if the zonemap is diverse enough
 
+    static bool is_support_data_type(LogicalType type);
+
     void sort();
     bool is_diverse();
     void build_histogram(size_t buckets);
@@ -107,6 +109,7 @@ class PageDataSample final : public DataSample {
     void with_zonemap(std::shared_ptr<SortableZoneMap> zonemap) { _zonemap = std::move(zonemap); }
 
 private:
+    bool _is_histogram_supported_type(LogicalType type) const;
     void _prepare_histogram(OlapReaderStatistics* stats);
     bool _has_histogram() const;
     StatusOr<RowIdSparseRange> _bernoulli_sample(OlapReaderStatistics* stats);

be/src/storage/rowset/segment_iterator.cpp

@@ -2089,7 +2089,7 @@ StatusOr<RowIdSparseRange> SegmentIterator::_sample_by_page() {
     int64_t random_seed = _opts.sample_options.random_seed;
     auto sampler = DataSample::make_page_sample(probability_percent, random_seed, num_data_pages, page_indexer);
 
-    if (column_reader->has_zone_map()) {
+    if (column_reader->has_zone_map() && SortableZoneMap::is_support_data_type(column_reader->column_type())) {
         IndexReadOptions opts = _index_read_options(cid);
         ASSIGN_OR_RETURN(auto zonemap, column_reader->get_raw_zone_map(opts));
         auto sorted = std::make_shared<SortableZoneMap>(column_reader->column_type(), std::move(zonemap));

tools/benchmark/table_sample.sh

@@ -6,9 +6,10 @@
 mysql_user=${MYSQL_USER:-"root"}
 mysql_host=${MYSQL_HOST:-"127.0.0.1"}
 mysql_port=${MYSQL_PORT:-"9030"}
-mysql_database=ssb_100g
+mysql_database=tpcds_100g
 
-query_columns="select table_name, column_name from information_schema.columns where table_schema='ssb_100g' and table_name='lineorder'"
+query_columns="select table_name, column_name from information_schema.columns where table_schema='tpcds_100g' \
+  and table_name in ('catalog_sales', 'store_returns', 'store_sales', 'web_sales') "
 
 
 percents=(
@@ -33,6 +34,10 @@ execute_sql() {
   local query="$2"
 
   result=$(mysql -u ${mysql_user} -P ${mysql_port} -h ${mysql_host} ${mysql_database} -vvv -e "$query" 2>&1)
+  if [ $? -ne 0 ]; then
+    echo "MySQL execution failed: ${result}"
+    exit 1;
+  fi
 
   execution_time=$(echo "$result" | grep -oP '\(\K[0-9]+\.[0-9]+(?= sec\))')
   count=$(echo "$result" | grep -A2 "count" | tail -n1 | awk '{print $2}')
@@ -44,7 +49,7 @@ execute_sql() {
 
 
 while read table col; do
-    full_column="table=$table.$col"
+    full_column="column=$table.$col"
     method="method=full"
     percent="percent=100"