use sample bloom filter

src/query/config/src/config.rs

@@ -2805,13 +2805,6 @@ pub struct CacheConfig {
     )]
     pub table_prune_partitions_count: u64,
 
-    #[clap(
-        long = "cache-table-prune-bloom-filter-invalid-keys-count",
-        value_name = "VALUE",
-        default_value = "256"
-    )]
-    pub table_prune_bloom_filter_invalid_keys_count: u64,
-
     /// Type of data cache storage
     #[clap(
         long = "cache-data-cache-storage",
@@ -3027,8 +3020,6 @@ mod cache_config_converters {
                 inverted_index_filter_size: value.inverted_index_filter_size,
                 inverted_index_filter_memory_ratio: value.inverted_index_filter_memory_ratio,
                 table_prune_partitions_count: value.table_prune_partitions_count,
-                table_prune_bloom_filter_invalid_keys_count: value
-                    .table_prune_bloom_filter_invalid_keys_count,
                 data_cache_storage: value.data_cache_storage.try_into()?,
                 table_data_cache_population_queue_size: value
                     .table_data_cache_population_queue_size,
@@ -3055,8 +3046,6 @@ mod cache_config_converters {
                 inverted_index_filter_size: value.inverted_index_filter_size,
                 inverted_index_filter_memory_ratio: value.inverted_index_filter_memory_ratio,
                 table_prune_partitions_count: value.table_prune_partitions_count,
-                table_prune_bloom_filter_invalid_keys_count: value
-                    .table_prune_bloom_filter_invalid_keys_count,
                 data_cache_storage: value.data_cache_storage.into(),
                 data_cache_key_reload_policy: value.data_cache_key_reload_policy.into(),
                 table_data_cache_population_queue_size: value

src/query/config/src/inner.rs

@@ -530,9 +530,6 @@ pub struct CacheConfig {
     /// Max number of cached prune partitions objects. Set it to 0 to disable it.
     pub table_prune_partitions_count: u64,
 
-    /// Max number of cached prune bloom filter invalid keys objects. Set it to 0 to disable it.
-    pub table_prune_bloom_filter_invalid_keys_count: u64,
-
     /// Max number of cached bloom index filters. Set it to 0 to disable it.
     // One bloom index filter per column of data block being indexed will be generated if necessary.
     //
@@ -668,7 +665,6 @@ impl Default for CacheConfig {
             inverted_index_filter_size: 2147483648,
             inverted_index_filter_memory_ratio: 0,
             table_prune_partitions_count: 256,
-            table_prune_bloom_filter_invalid_keys_count: 256,
             data_cache_storage: Default::default(),
             table_data_cache_population_queue_size: 0,
             disk_cache_config: Default::default(),

src/query/expression/src/function.rs

@@ -114,7 +114,7 @@ pub struct FunctionContext {
 
     pub geometry_output_format: GeometryDataType,
     pub parse_datetime_ignore_remainder: bool,
-    pub bloom_filter_ignore_invalid_key_ratio: Option<f32>,
+    pub enable_bloom_filter_ignore_invalid_key: bool,
 }
 
 #[derive(Clone)]

src/query/service/src/sessions/query_ctx.rs

@@ -698,13 +698,9 @@ impl TableContext for QueryContext {
         let disable_variant_check = settings.get_disable_variant_check()?;
         let geometry_output_format = settings.get_geometry_output_format()?;
         let parse_datetime_ignore_remainder = settings.get_parse_datetime_ignore_remainder()?;
-        let invalid_key_ratio_value = settings.get_bloom_filter_ignore_invalid_key_ratio()?;
-        // invalid key ratio is 100 means don't need cache invalid filter keys
-        let bloom_filter_ignore_invalid_key_ratio = if invalid_key_ratio_value < 100 {
-            Some(invalid_key_ratio_value as f32 / 100.0)
-        } else {
-            None
-        };
+        let enable_bloom_filter_ignore_invalid_key =
+            settings.get_enable_bloom_filter_ignore_invalid_key()?;
+
         let query_config = &GlobalConfig::instance().query;
 
         Ok(FunctionContext {
@@ -725,7 +721,7 @@ impl TableContext for QueryContext {
             external_server_request_batch_rows,
             geometry_output_format,
             parse_datetime_ignore_remainder,
-            bloom_filter_ignore_invalid_key_ratio,
+            enable_bloom_filter_ignore_invalid_key,
         })
     }
 

src/query/service/tests/it/storages/fuse/operations/mutation/recluster_mutator.rs

@@ -128,12 +128,10 @@ async fn test_recluster_mutator_block_select() -> Result<()> {
     // unused snapshot.
     let snapshot = TableSnapshot::new_empty_snapshot(schema.as_ref().clone(), None);
 
-    let table_id = 1;
     let ctx: Arc<dyn TableContext> = ctx.clone();
     let segment_locations = create_segment_location_vector(test_segment_locations, None);
     let compact_segments = FuseTable::segment_pruning(
         &ctx,
-        table_id,
         schema.clone(),
         data_accessor.clone(),
         &None,
@@ -259,12 +257,10 @@ async fn test_safety_for_recluster() -> Result<()> {
             }
         }
 
-        let table_id = 1;
         let ctx: Arc<dyn TableContext> = ctx.clone();
         let segment_locations = create_segment_location_vector(locations.clone(), None);
         let compact_segments = FuseTable::segment_pruning(
             &ctx,
-            table_id,
             schema.clone(),
             data_accessor.clone(),
             &None,