chore: remove unused settings for aggregation (#19103)

Author: dqhl76

src/query/service/src/physical_plans/physical_aggregate_final.rs

@@ -144,9 +144,7 @@ impl IPhysicalPlan for AggregateFinal {
     fn build_pipeline2(&self, builder: &mut PipelineBuilder) -> Result<()> {
         let max_block_rows = builder.settings.get_max_block_size()? as usize;
         let max_block_bytes = builder.settings.get_max_block_bytes()? as usize;
-        let enable_experimental_aggregate_hashtable = builder
-            .settings
-            .get_enable_experimental_aggregate_hashtable()?;
+
         let max_spill_io_requests = builder.settings.get_max_spill_io_requests()?;
         let max_restore_worker = builder.settings.get_max_aggregate_restore_worker()?;
         let enable_experiment_aggregate = builder.settings.get_enable_experiment_aggregate()?;
@@ -160,7 +158,6 @@ impl IPhysicalPlan for AggregateFinal {
             self.before_group_by_schema.clone(),
             &self.group_by,
             &self.agg_funcs,
-            enable_experimental_aggregate_hashtable,
             is_cluster_aggregate,
             max_spill_io_requests as usize,
             enable_experiment_aggregate,
@@ -374,9 +371,6 @@ impl PhysicalPlanBuilder {
                     group_by_shuffle_mode = "before_merge".to_string();
                 }
 
-                let enable_experimental_aggregate_hashtable =
-                    settings.get_enable_experimental_aggregate_hashtable()?;
-
                 if let Some(grouping_sets) = agg.grouping_sets.as_ref() {
                     // ignore `_grouping_id`.
                     // If the aggregation function argument if a group item,
@@ -423,7 +417,6 @@ impl PhysicalPlanBuilder {
                         AggregatePartial {
                             input: expand,
                             agg_funcs,
-                            enable_experimental_aggregate_hashtable,
                             group_by_display,
                             group_by: group_items,
                             stat_info: Some(stat_info),
@@ -436,7 +429,6 @@ impl PhysicalPlanBuilder {
                             agg_funcs,
                             rank_limit,
                             group_by_display,
-                            enable_experimental_aggregate_hashtable,
                             group_by: group_items,
                             stat_info: Some(stat_info),
                             meta: PhysicalPlanMeta::new("AggregatePartial"),
@@ -477,7 +469,6 @@ impl PhysicalPlanBuilder {
                     PhysicalPlan::new(AggregatePartial {
                         agg_funcs,
                         group_by_display,
-                        enable_experimental_aggregate_hashtable,
                         rank_limit: None,
                         group_by: group_items,
                         input: PhysicalPlan::new(expand),
@@ -488,7 +479,6 @@ impl PhysicalPlanBuilder {
                     PhysicalPlan::new(AggregatePartial {
                         input,
                         agg_funcs,
-                        enable_experimental_aggregate_hashtable,
                         group_by_display,
                         group_by: group_items,
                         stat_info: Some(stat_info),

src/query/service/src/physical_plans/physical_aggregate_partial.rs

@@ -59,7 +59,6 @@ pub struct AggregatePartial {
     pub input: PhysicalPlan,
     pub group_by: Vec<IndexType>,
     pub agg_funcs: Vec<AggregateFunctionDesc>,
-    pub enable_experimental_aggregate_hashtable: bool,
     pub group_by_display: Vec<String>,
 
     // Order by keys if keys are subset of group by key, then we can use rank to filter data in previous
@@ -163,7 +162,6 @@ impl IPhysicalPlan for AggregatePartial {
             meta: self.meta.clone(),
             group_by: self.group_by.clone(),
             agg_funcs: self.agg_funcs.clone(),
-            enable_experimental_aggregate_hashtable: self.enable_experimental_aggregate_hashtable,
             group_by_display: self.group_by_display.clone(),
             rank_limit: self.rank_limit.clone(),
             stat_info: self.stat_info.clone(),
@@ -178,17 +176,12 @@ impl IPhysicalPlan for AggregatePartial {
         let max_threads = builder.settings.get_max_threads()?;
         let max_spill_io_requests = builder.settings.get_max_spill_io_requests()?;
 
-        let enable_experimental_aggregate_hashtable = builder
-            .settings
-            .get_enable_experimental_aggregate_hashtable()?;
-
         let enable_experiment_aggregate = builder.settings.get_enable_experiment_aggregate()?;
 
         let params = PipelineBuilder::build_aggregator_params(
             self.input.output_schema()?,
             &self.group_by,
             &self.agg_funcs,
-            enable_experimental_aggregate_hashtable,
             builder.is_exchange_parent(),
             max_spill_io_requests as usize,
             enable_experiment_aggregate,

src/query/service/src/pipelines/builders/builder_aggregate.rs

@@ -34,7 +34,6 @@ impl PipelineBuilder {
         input_schema: DataSchemaRef,
         group_by: &[IndexType],
         agg_funcs: &[AggregateFunctionDesc],
-        enable_experimental_aggregate_hashtable: bool,
         cluster_aggregator: bool,
         max_spill_io_requests: usize,
         enable_experiment_aggregate: bool,
@@ -130,7 +129,6 @@ impl PipelineBuilder {
             &group_by,
             &aggs,
             &agg_args,
-            enable_experimental_aggregate_hashtable,
             cluster_aggregator,
             max_spill_io_requests,
             enable_experiment_aggregate,

src/query/service/src/pipelines/processors/transforms/aggregator/aggregator_params.rs

@@ -37,7 +37,6 @@ pub struct AggregatorParams {
     // If there is no aggregate function, layout is None
     pub states_layout: Option<StatesLayout>,
 
-    pub enable_experimental_aggregate_hashtable: bool,
     pub cluster_aggregator: bool,
     pub max_spill_io_requests: usize,
 
@@ -55,7 +54,6 @@ impl AggregatorParams {
         group_columns: &[usize],
         agg_funcs: &[AggregateFunctionRef],
         agg_args: &[Vec<usize>],
-        enable_experimental_aggregate_hashtable: bool,
         cluster_aggregator: bool,
         max_spill_io_requests: usize,
         enable_experiment_aggregate: bool,
@@ -75,7 +73,6 @@ impl AggregatorParams {
             aggregate_functions: agg_funcs.to_vec(),
             aggregate_functions_arguments: agg_args.to_vec(),
             states_layout,
-            enable_experimental_aggregate_hashtable,
             cluster_aggregator,
             max_block_rows,
             max_block_bytes,

src/query/settings/src/settings_default.rs

@@ -327,13 +327,6 @@ impl DefaultSettings {
                     scope: SettingScope::Both,
                     range: Some(SettingRange::String(all_timezones)),
                 }),
-                ("group_by_two_level_threshold", DefaultSettingValue {
-                    value: UserSettingValue::UInt64(20000),
-                    desc: "Sets the number of keys in a GROUP BY operation that will trigger a two-level aggregation.",
-                    mode: SettingMode::Both,
-                    scope: SettingScope::Both,
-                    range: Some(SettingRange::Numeric(0..=u64::MAX)),
-                }),
                 ("max_inlist_to_or", DefaultSettingValue {
                     value: UserSettingValue::UInt64(3),
                     desc: "Sets the maximum number of values that can be included in an IN expression to be converted to an OR operator.",
@@ -727,13 +720,6 @@ impl DefaultSettings {
                     scope: SettingScope::Both,
                     range: Some(SettingRange::String(vec!["before_partial".into(), "before_merge".into()])),
                 }),
-                ("efficiently_memory_group_by", DefaultSettingValue {
-                    value: UserSettingValue::UInt64(0),
-                    desc: "Memory is used efficiently, but this may cause performance degradation.",
-                    mode: SettingMode::Both,
-                    scope: SettingScope::Both,
-                    range: Some(SettingRange::Numeric(0..=1)),
-                }),
                 ("lazy_read_threshold", DefaultSettingValue {
                     value: UserSettingValue::UInt64(1000),
                     desc: "Sets the maximum LIMIT in a query to enable lazy read optimization. Setting it to 0 disables the optimization.",
@@ -1005,13 +991,6 @@ impl DefaultSettings {
                     scope: SettingScope::Both,
                     range: Some(SettingRange::Numeric(0..=1)),
                 }),
-                ("enable_experimental_aggregate_hashtable", DefaultSettingValue {
-                    value: UserSettingValue::UInt64(1),
-                    desc: "Enables experimental aggregate hashtable",
-                    mode: SettingMode::Both,
-                    scope: SettingScope::Both,
-                    range: Some(SettingRange::Numeric(0..=1)),
-                }),
                 ("numeric_cast_option", DefaultSettingValue {
                     value: UserSettingValue::String("rounding".to_string()),
                     desc: "Set numeric cast mode as \"rounding\" or \"truncating\".",

src/query/settings/src/settings_getter_setter.rs

@@ -305,11 +305,6 @@ impl Settings {
         self.try_get_string("timezone")
     }
 
-    // Get group by two level threshold
-    pub fn get_group_by_two_level_threshold(&self) -> Result<u64> {
-        self.try_get_u64("group_by_two_level_threshold")
-    }
-
     pub fn get_max_inlist_to_or(&self) -> Result<u64> {
         self.try_get_u64("max_inlist_to_or")
     }
@@ -537,14 +532,6 @@ impl Settings {
         Ok(self.try_get_u64("grouping_sets_to_union")? == 1)
     }
 
-    pub fn get_efficiently_memory_group_by(&self) -> Result<bool> {
-        Ok(self.try_get_u64("efficiently_memory_group_by")? == 1)
-    }
-
-    pub fn get_enable_experimental_aggregate_hashtable(&self) -> Result<bool> {
-        Ok(self.try_get_u64("enable_experimental_aggregate_hashtable")? == 1)
-    }
-
     pub fn get_lazy_read_threshold(&self) -> Result<u64> {
         self.try_get_u64("lazy_read_threshold")
     }

tests/sqllogictests/suites/base/03_common/03_0003_select_group_by.test

@@ -114,23 +114,8 @@ SELECT number, count(*) FROM numbers_mt(10000) group by number order by number l
 3 1
 4 1
 
-statement ok
-set group_by_two_level_threshold=10
-
-query II
-SELECT number, count(*) FROM numbers_mt(1000) group by number order by number limit 5
-----
-0 1
-1 1
-2 1
-3 1
-4 1
-
-statement ok
-set group_by_two_level_threshold=1000000000
-
 query II
-SELECT number, count(*) FROM numbers_mt(1000) group by number order by number limit 5
+SELECT number, count(*) FROM numbers_mt(100000) group by number order by number limit 5
 ----
 0 1
 1 1