mirror of https://github.com/facebook/rocksdb.git
compute compaction score once for a batch of range file deletes (#10744)
Summary: Only re-calculate compaction score once for a batch of deletions. Fix performance regression brought by https://github.com/facebook/rocksdb/pull/8434. Pull Request resolved: https://github.com/facebook/rocksdb/pull/10744 Test Plan: In one of our production cluster that recently upgraded to RocksDB 6.29, it takes more than 10 minutes to delete files in 30,000 ranges. The RocksDB instance contains approximately 80,000 files. After this patch, the duration reduces to 100+ ms, which is on par with RocksDB 6.4. Cherry-picking downstream PR: https://github.com/tikv/rocksdb/pull/316 Signed-off-by: tabokie <xy.tao@outlook.com> Reviewed By: cbi42 Differential Revision: D48002581 Pulled By: ajkr fbshipit-source-id: 7245607ee3ad79c53b648a6396c9159f166b9437
This commit is contained in:
parent
cdb11f5ce6
commit
d2b0652b32
|
@ -4447,10 +4447,12 @@ Status DBImpl::DeleteFilesInRanges(ColumnFamilyHandle* column_family,
|
|||
deleted_files.insert(level_file);
|
||||
level_file->being_compacted = true;
|
||||
}
|
||||
vstorage->ComputeCompactionScore(*cfd->ioptions(),
|
||||
*cfd->GetLatestMutableCFOptions());
|
||||
}
|
||||
}
|
||||
if (!deleted_files.empty()) {
|
||||
vstorage->ComputeCompactionScore(*cfd->ioptions(),
|
||||
*cfd->GetLatestMutableCFOptions());
|
||||
}
|
||||
if (edit.GetDeletedFiles().empty()) {
|
||||
job_context.Clean();
|
||||
return status;
|
||||
|
|
Loading…
Reference in New Issue