PbCompactionConfig

Type Alias PbCompactionConfig 

Source
pub type PbCompactionConfig = CompactionConfig;

Aliased Type§

pub struct PbCompactionConfig {
Show 32 fields pub max_bytes_for_level_base: u64, pub max_level: u64, pub max_bytes_for_level_multiplier: u64, pub max_compaction_bytes: u64, pub sub_level_max_compaction_bytes: u64, pub level0_tier_compact_file_number: u64, pub compaction_mode: i32, pub compression_algorithm: Vec<String>, pub target_file_size_base: u64, pub compaction_filter_mask: u32, pub max_sub_compaction: u32, pub max_space_reclaim_bytes: u64, pub split_by_state_table: bool, pub split_weight_by_vnode: u32, pub level0_stop_write_threshold_sub_level_number: u64, pub level0_max_compact_file_number: u64, pub level0_sub_level_compact_level_count: u32, pub level0_overlapping_sub_level_compact_level_count: u32, pub tombstone_reclaim_ratio: u32, pub enable_emergency_picker: bool, pub max_l0_compact_level_count: Option<u32>, pub sst_allowed_trivial_move_min_size: Option<u64>, pub disable_auto_group_scheduling: Option<bool>, pub max_overlapping_level_size: Option<u64>, pub emergency_level0_sst_file_count: Option<u32>, pub emergency_level0_sub_level_partition: Option<u32>, pub level0_stop_write_threshold_max_sst_count: Option<u32>, pub level0_stop_write_threshold_max_size: Option<u64>, pub sst_allowed_trivial_move_max_count: Option<u32>, pub enable_optimize_l0_interval_selection: Option<bool>, pub vnode_aligned_level_size_threshold: Option<u64>, pub max_kv_count_for_xor16: Option<u64>,
}

Fields§

§max_bytes_for_level_base: u64§max_level: u64§max_bytes_for_level_multiplier: u64§max_compaction_bytes: u64§sub_level_max_compaction_bytes: u64§level0_tier_compact_file_number: u64§compaction_mode: i32§compression_algorithm: Vec<String>§target_file_size_base: u64§compaction_filter_mask: u32§max_sub_compaction: u32§max_space_reclaim_bytes: u64§split_by_state_table: bool§split_weight_by_vnode: u32

Compaction needs to cut the state table every time 1/weight of vnodes in the table have been processed.

§level0_stop_write_threshold_sub_level_number: u64

soft limit for max number of sub level number

§level0_max_compact_file_number: u64§level0_sub_level_compact_level_count: u32§level0_overlapping_sub_level_compact_level_count: u32

for tier compaction pick overlapping level

§tombstone_reclaim_ratio: u32§enable_emergency_picker: bool§max_l0_compact_level_count: Option<u32>

The limitation of the level count of l0 compaction

§sst_allowed_trivial_move_min_size: Option<u64>

The limitation of base level trivial move sst size

§disable_auto_group_scheduling: Option<bool>

The limitation of auto group scheduling

§max_overlapping_level_size: Option<u64>

The limitation of the max size of the overlapping-level for the compaction hummock will reorg the commit-sstables to the multi overlapping-level if the size of the commit-sstables is larger than max_overlapping_level_size

§emergency_level0_sst_file_count: Option<u32>

The emergency compaction limitations for the level0 sstables file count

§emergency_level0_sub_level_partition: Option<u32>

The emergency compaction limitations for the level0 sub level partition

§level0_stop_write_threshold_max_sst_count: Option<u32>

The limitation of the max sst count of the level0 to trigger the write stop

§level0_stop_write_threshold_max_size: Option<u64>

The limitation of the max sst size of the level0 to trigger the write stop

§sst_allowed_trivial_move_max_count: Option<u32>

The limitation of the max sst count of the trivial move task

§enable_optimize_l0_interval_selection: Option<bool>

Enable the optimization of the interval selection for the level0 compaction

§vnode_aligned_level_size_threshold: Option<u64>

Enable strict one-vnode-per-SST for levels that meet the size threshold. Only effective when both conditions are met:

  1. target_level > base_level (deeper levels only)
  2. level.total_file_size >= this threshold

This prevents generating too many small SSTs when level size is small. Each SST will contain exactly one vnode, enabling precise query pruning.

Typical value: 50GB.

§max_kv_count_for_xor16: Option<u64>

The maximum key-value count threshold for using Xor16 filter instead of BlockedXor16 filter. When the number of keys in a compaction task or SST exceeds this threshold, a BlockedXor16 filter will be used instead of a single Xor16 filter.

Typical value: 256 * 1024 (256K keys).