chore: adjust table data cache population queue size (#12773)

* chore: temporarily enable disk cache for ci-benchmark

* adjust queue size (by default num of cpu cores)

* fix typo

* adjust default cache population queue size

* clean up

---------

Co-authored-by: BohuTANG <overred.shuttler@gmail.com>

Author: dantengsky

Cargo.lock

@@ -2298,10 +2298,10 @@ pub struct CacheConfig {
     /// - please monitor the 'population_overflow_count' metric
     ///   if it keeps increasing, and disk cache hits rate is not as expected. please consider
     ///   increase this value.
-    #[clap(
-        long = "cache-data-cache-population-queue-size",
-        default_value = "65536"
-    )]
+    ///
+    /// default value is 0, which means queue size will be adjusted automatically based on
+    /// number of CPU cores.
+    #[clap(long = "cache-data-cache-population-queue-size", default_value = "0")]
     pub table_data_cache_population_queue_size: u32,
 
     /// Storage that hold the data caches

src/query/config/src/config.rs

@@ -597,7 +597,7 @@ impl Default for CacheConfig {
             table_bloom_index_filter_size: 2147483648,
             table_prune_partitions_count: 256,
             data_cache_storage: Default::default(),
-            table_data_cache_population_queue_size: 65536,
+            table_data_cache_population_queue_size: 0,
             disk_cache_config: Default::default(),
             table_data_deserialized_data_bytes: 0,
         }

src/query/config/src/inner.rs

@@ -12,7 +12,7 @@ DB.Table: 'system'.'configs', Table: configs-table_id:1, ver:0, Engine: SystemCo
 | 'cache'   | 'table_bloom_index_filter_count'           | '0'                                                            | ''       |
 | 'cache'   | 'table_bloom_index_filter_size'            | '2147483648'                                                   | ''       |
 | 'cache'   | 'table_bloom_index_meta_count'             | '3000'                                                         | ''       |
-| 'cache'   | 'table_data_cache_population_queue_size'   | '65536'                                                        | ''       |
+| 'cache'   | 'table_data_cache_population_queue_size'   | '0'                                                            | ''       |
 | 'cache'   | 'table_data_deserialized_data_bytes'       | '0'                                                            | ''       |
 | 'cache'   | 'table_meta_segment_bytes'                 | '1073741824'                                                   | ''       |
 | 'cache'   | 'table_meta_segment_count'                 | 'null'                                                         | ''       |

src/query/service/tests/it/storages/testdata/configs_table_basic.txt

@@ -18,3 +18,5 @@ common-exception = { path = "../../../../common/exception" }
 storages-common-cache = { path = "../../common/cache" }
 storages-common-index = { path = "../../common/index" }
 storages-common-table-meta = { path = "../../common/table-meta" }
+
+log = { workspace = true }

src/query/storages/common/cache-manager/Cargo.toml

@@ -21,6 +21,7 @@ use common_cache::DefaultHashBuilder;
 use common_config::CacheConfig;
 use common_config::CacheStorageTypeInnerConfig;
 use common_exception::Result;
+use log::info;
 use storages_common_cache::InMemoryCacheBuilder;
 use storages_common_cache::InMemoryItemCacheHolder;
 use storages_common_cache::Named;
@@ -66,9 +67,26 @@ impl CacheManager {
                     let real_disk_cache_root = PathBuf::from(&config.disk_cache_config.path)
                         .join(tenant_id.into())
                         .join("v1");
+
+                    let queue_size: u32 = if config.table_data_cache_population_queue_size > 0 {
+                        config.table_data_cache_population_queue_size
+                    } else {
+                        std::cmp::max(
+                            1,
+                            std::thread::available_parallelism()
+                                .expect("Cannot get thread count")
+                                .get() as u32,
+                        )
+                    };
+
+                    info!(
+                        "disk cache enabled, cache population queue size {}",
+                        queue_size
+                    );
+
                     Self::new_block_data_cache(
                         &real_disk_cache_root,
-                        config.table_data_cache_population_queue_size,
+                        queue_size,
                         config.disk_cache_config.max_bytes,
                     )?
                 }