It deletes the most recent version of a key, and whether older versions of the key will come back to life is undefined.
Basic Usage
SingleDelete is a new database operation. In contrast to the conventional Delete() operation, the deletion entry is removed along with the value when the two are lined up in a compaction. Therefore, similar to Delete()
method, SingleDelete()
removes the database entry for a key, but has the prerequisites that the key exists and was not overwritten. Returns OK on success, and a non-OK status on error. It is not an error if key did not exist in the database. If a key is overwritten (by calling Put()
multiple times), then the result of calling SingleDelete()
on this key is undefined. SingleDelete()
only behaves correctly if there has been only one Put()
for this key since the previous call to SingleDelete()
for this key. The following code shows how to use SingleDelete
:
std::string value;
rocksdb::Status s;
db->Put(rocksdb::WriteOptions(), "foo", "bar1");
db->SingleDelete(rocksdb::WriteOptions(), "foo");
s = db->Get(rocksdb::ReadOptions(), "foo", &value); // s.IsNotFound()==true
db->Put(rocksdb::WriteOptions(), "foo", "bar2");
db->Put(rocksdb::WriteOptions(), "foo", "bar3");
db->SingleDelete(rocksdb::ReadOptions(), "foo", &value); // Undefined result
SingleDelete
API is also available in WriteBatch
. Actually, DB::SingleDelete()
is implemented by creating a WriteBatch
with only one operation, SingleDelete
, in this batch. The following code snippet shows the basic usage of WriteBatch::SingleDelete()
:
rocksdb::WriteBatch batch;
batch.Put(key1, value);
batch.SingleDelete(key1);
s = db->Write(rocksdb::WriteOptions(), &batch);
Notes
- Callers have to ensure that
SingleDelete
only applies to a key having not been deleted usingDelete()
or written usingMerge()
. MixingSingleDelete()
operations withDelete()
andMerge()
can result in undefined behavior (other keys are not affected by this). IfCompactionIterator
sees aSingleDelete
followed by aDelete
for the same user key, an error message will be logged to the LOG file and compaction will fail. SingleDelete
is NOT compatible with cuckoo hash tables, which means you should not callSingleDelete
if you setoptions.memtable_factory
with NewHashCuckooRepFactory- Applications are not encouraged to issue multiple
SingleDelete
for the same key. However, it is possible for compaction to see multiple consecutiveSingleDelete
for the same user key in certain cases, e.g. (when write-prepared transaction is enabled). RocksDB internally is able to handle this case correctly.