Struct CompactionJobStats 

pub struct CompactionJobStats<'a> { /* private fields */ }

What one compaction job did, borrowed from the event that reported it.

Read from a compaction job info or a subcompaction job info. Counters that are not applicable to the compaction, or that RocksDB was not asked to collect, read back as 0.

Implementations

impl<'a> CompactionJobStats<'a>

pub fn elapsed_micros(&self) -> u64

Wall clock time this compaction took, in microseconds.

pub fn cpu_micros(&self) -> u64

CPU time this compaction took, in microseconds.

pub fn has_accurate_num_input_records(&self) -> bool

Whether Self::num_input_records is accurate across all subcompactions.

pub fn num_input_records(&self) -> u64

Number of compaction input records. Only trustworthy when Self::has_accurate_num_input_records is true.

pub fn num_blobs_read(&self) -> u64

Number of blobs read from blob files.

pub fn num_input_files(&self) -> usize

Number of compaction input files, counting table files only.

pub fn num_input_files_at_output_level(&self) -> usize

Number of compaction input table files that were already at the output level.

pub fn num_filtered_input_files(&self) -> usize

Number of compaction input files filtered out by compaction optimizations.

pub fn num_filtered_input_files_at_output_level(&self) -> usize

Number of compaction input files at the output level that were filtered out by compaction optimizations.

pub fn num_output_records(&self) -> u64

Number of compaction output records.

pub fn num_output_files(&self) -> usize

Number of compaction output table files.

pub fn num_output_files_blob(&self) -> usize

Number of compaction output blob files.

pub fn is_full_compaction(&self) -> bool

Whether this was a full compaction, meaning every live SST file was an input.

pub fn is_manual_compaction(&self) -> bool

Whether this was a manual compaction.

pub fn is_remote_compaction(&self) -> bool

Whether the compaction ran in a remote worker.

Only the compaction completed event carries the truth. On the compaction begin event RocksDB sets this to true whenever a compaction_service is configured, before it knows whether the job will really be scheduled remotely or fall back to local.

pub fn total_input_bytes(&self) -> u64

Total size of the table files in the compaction input.

pub fn total_skipped_input_bytes(&self) -> u64

Total size of the input table files that were skipped because compaction optimizations filtered them out.

pub fn total_blob_bytes_read(&self) -> u64

Total size of the blobs read from blob files.

pub fn total_output_bytes(&self) -> u64

Total size of the table files in the compaction output.

pub fn total_output_bytes_blob(&self) -> u64

Total size of the blob files in the compaction output.

pub fn num_input_files_trivially_moved(&self) -> usize

Number of input files that were trivially moved rather than rewritten.

pub fn num_records_replaced(&self) -> u64

Number of records superseded by a newer record for the same key. Counts both updates and deletions.

pub fn total_input_raw_key_bytes(&self) -> u64

Sum of the uncompressed input keys, in bytes.

pub fn total_input_raw_value_bytes(&self) -> u64

Sum of the uncompressed input values, in bytes.

pub fn num_input_deletion_records(&self) -> u64

Number of deletion entries before the compaction. Deletion entries can disappear during compaction because they expired.

pub fn num_expired_deletion_records(&self) -> u64

Number of deletion records dropped as obsolete because every deletion they could still cause has already happened.

pub fn num_corrupt_keys(&self) -> u64

Number of corrupt keys encountered and written out, meaning keys that failed to parse as internal keys.

pub fn file_write_nanos(&self) -> u64

Time spent in file Append calls, in nanoseconds.

Only populated when report_bg_io_stats is on.

pub fn file_range_sync_nanos(&self) -> u64

Time spent syncing file ranges, in nanoseconds.

Only populated when report_bg_io_stats is on.

pub fn file_fsync_nanos(&self) -> u64

Time spent in file fsync, in nanoseconds.

Only populated when report_bg_io_stats is on.

pub fn file_prepare_write_nanos(&self) -> u64

Time spent preparing file writes, such as fallocate, in nanoseconds.

Only populated when report_bg_io_stats is on.

pub fn smallest_output_key_prefix(&self) -> &'a [u8]

First 8 bytes of the smallest user key in the output, or fewer if the key is shorter.

Empty when the compaction wrote no table files.

pub fn largest_output_key_prefix(&self) -> &'a [u8]

First 8 bytes of the largest user key in the output, or fewer if the key is shorter.

Empty when the compaction wrote no table files.

pub fn num_single_del_fallthru(&self) -> u64

Number of single deletes that did not meet a put.

pub fn num_single_del_mismatch(&self) -> u64

Number of single deletes that met something other than a put.