提交 bf66c10b 编写于 作者: M Mayank Agarwal

Use KeyMayExist for WriteBatch-Deletes

Summary:
Introduced KeyMayExist checking during writebatch-delete and removed from Outer Delete API because it uses writebatch-delete.
Added code to skip getting Table from disk if not already present in table_cache.
Some renaming of variables.
Introduced KeyMayExistImpl which allows checking since specified sequence number in GetImpl useful to check partially written writebatch.
Changed KeyMayExist to not be pure virtual and provided a default implementation.
Expanded unit-tests in db_test to check appropriately.
Ran db_stress for 1 hour with ./db_stress --max_key=100000 --ops_per_thread=10000000 --delpercent=50 --filter_deletes=1 --statistics=1.

Test Plan: db_stress;make check

Reviewers: dhruba, haobo

Reviewed By: dhruba

CC: leveldb, xjin

Differential Revision: https://reviews.facebook.net/D11745
上级 d364eea1
......@@ -328,7 +328,7 @@ static auto FLAGS_bytes_per_sync =
leveldb::Options().bytes_per_sync;
// On true, deletes use bloom-filter and drop the delete if key not present
static bool FLAGS_deletes_check_filter_first = false;
static bool FLAGS_filter_deletes = false;
namespace leveldb {
......@@ -1114,7 +1114,7 @@ unique_ptr<char []> GenerateKeyFromInt(int v, const char* suffix = "")
options.max_bytes_for_level_base = FLAGS_max_bytes_for_level_base;
options.max_bytes_for_level_multiplier =
FLAGS_max_bytes_for_level_multiplier;
options.deletes_check_filter_first = FLAGS_deletes_check_filter_first;
options.filter_deletes = FLAGS_filter_deletes;
if (FLAGS_max_bytes_for_level_multiplier_additional.size() > 0) {
if (FLAGS_max_bytes_for_level_multiplier_additional.size() !=
(unsigned int)FLAGS_num_levels) {
......@@ -2220,9 +2220,9 @@ int main(int argc, char** argv) {
FLAGS_keys_per_multiget = n;
} else if (sscanf(argv[i], "--bytes_per_sync=%ld%c", &l, &junk) == 1) {
FLAGS_bytes_per_sync = l;
} else if (sscanf(argv[i], "--deletes_check_filter_first=%d%c", &n, &junk)
} else if (sscanf(argv[i], "--filter_deletes=%d%c", &n, &junk)
== 1 && (n == 0 || n ==1 )) {
FLAGS_deletes_check_filter_first = n;
FLAGS_filter_deletes = n;
} else {
fprintf(stderr, "Invalid flag '%s'\n", argv[i]);
exit(1);
......
......@@ -691,7 +691,7 @@ Status DBImpl::RecoverLogFile(uint64_t log_number,
mem = new MemTable(internal_comparator_, NumberLevels());
mem->Ref();
}
status = WriteBatchInternal::InsertInto(&batch, mem);
status = WriteBatchInternal::InsertInto(&batch, mem, &options_);
MaybeIgnoreError(&status);
if (!status.ok()) {
break;
......@@ -2078,13 +2078,13 @@ Status DBImpl::Get(const ReadOptions& options,
return GetImpl(options, key, value);
}
// If no_IO is true, then returns Status::NotFound if key is not in memtable,
// If no_io is true, then returns Status::NotFound if key is not in memtable,
// immutable-memtable and bloom-filters can guarantee that key is not in db,
// "value" is garbage string if no_IO is true
// "value" is garbage string if no_io is true
Status DBImpl::GetImpl(const ReadOptions& options,
const Slice& key,
std::string* value,
const bool no_IO) {
const bool no_io) {
Status s;
StopWatch sw(env_, options_.statistics, DB_GET);
......@@ -2113,12 +2113,12 @@ Status DBImpl::GetImpl(const ReadOptions& options,
// s is both in/out. When in, s could either be OK or MergeInProgress.
// value will contain the current merge operand in the latter case.
LookupKey lkey(key, snapshot);
if (mem->Get(lkey, value, &s, options_, no_IO)) {
if (mem->Get(lkey, value, &s, options_, no_io)) {
// Done
} else if (imm.Get(lkey, value, &s, options_, no_IO)) {
} else if (imm.Get(lkey, value, &s, options_, no_io)) {
// Done
} else {
current->Get(options, lkey, value, &s, &stats, options_, no_IO);
current->Get(options, lkey, value, &s, &stats, options_, no_io);
have_stat_update = true;
}
mutex_.Lock();
......@@ -2209,8 +2209,17 @@ std::vector<Status> DBImpl::MultiGet(const ReadOptions& options,
}
bool DBImpl::KeyMayExist(const Slice& key) {
return KeyMayExistImpl(key, versions_->LastSequence());
}
bool DBImpl::KeyMayExistImpl(const Slice& key,
const SequenceNumber read_from_seq) {
std::string value;
const Status s = GetImpl(ReadOptions(), key, &value, true);
SnapshotImpl read_from_snapshot;
read_from_snapshot.number_ = read_from_seq;
ReadOptions ropts;
ropts.snapshot = &read_from_snapshot;
const Status s = GetImpl(ropts, key, &value, true);
return !s.IsNotFound();
}
......@@ -2249,10 +2258,6 @@ Status DBImpl::Merge(const WriteOptions& o, const Slice& key,
}
Status DBImpl::Delete(const WriteOptions& options, const Slice& key) {
if (options_.deletes_check_filter_first && !KeyMayExist(key)) {
RecordTick(options_.statistics, NUMBER_FILTERED_DELETES);
return Status::OK();
}
return DB::Delete(options, key);
}
......@@ -2311,7 +2316,8 @@ Status DBImpl::Write(const WriteOptions& options, WriteBatch* my_batch) {
}
}
if (status.ok()) {
status = WriteBatchInternal::InsertInto(updates, mem_);
status = WriteBatchInternal::InsertInto(updates, mem_, &options_, this,
options_.filter_deletes);
if (!status.ok()) {
// Panic for in-memory corruptions
// Note that existing logic was not sound. Any partial failure writing
......
......@@ -103,6 +103,15 @@ class DBImpl : public DB {
// Trigger's a background call for testing.
void TEST_PurgeObsoleteteWAL();
// KeyMayExist's internal function, but can be called internally from rocksdb
// to check memtable from sequence_number=read_from_seq. This is useful to
// check presence of key in db when key's existence is to be also checked in
// an incompletely written WriteBatch in memtable. eg. Database doesn't have
// key A and WriteBatch=[PutA,B; DelA]. A KeyMayExist called from DelA also
// needs to check itself for any PutA to be sure to not drop the delete.
bool KeyMayExistImpl(const Slice& key,
const SequenceNumber read_from_seq);
protected:
Env* const env_;
const std::string dbname_;
......@@ -399,11 +408,11 @@ class DBImpl : public DB {
std::vector<SequenceNumber>& snapshots,
SequenceNumber* prev_snapshot);
// Function that Get and KeyMayExist call with no_IO true or false
// Function that Get and KeyMayExistImpl call with no_io true or false
Status GetImpl(const ReadOptions& options,
const Slice& key,
std::string* value,
const bool no_IO = false);
const bool no_io = false);
};
// Sanitize db options. The caller should delete result.info_log if
......
......@@ -291,7 +291,7 @@ class DBTest {
// TODO -- test more options
break;
case kDeletesFilterFirst:
options.deletes_check_filter_first = true;
options.filter_deletes = true;
break;
default:
break;
......@@ -805,6 +805,44 @@ TEST(DBTest, KeyMayExist) {
} while (ChangeOptions());
}
// A delete is skipped for key if KeyMayExist(key) returns False
// Tests Writebatch consistency and proper delete behaviour
TEST(DBTest, FilterDeletes) {
Options options = CurrentOptions();
options.filter_policy = NewBloomFilterPolicy(20);
options.filter_deletes = true;
Reopen(&options);
WriteBatch batch;
batch.Delete("a");
dbfull()->Write(WriteOptions(), &batch);
ASSERT_EQ(AllEntriesFor("a"), "[ ]"); // Delete skipped
batch.Clear();
batch.Put("a", "b");
batch.Delete("a");
dbfull()->Write(WriteOptions(), &batch);
ASSERT_EQ(Get("a"), "NOT_FOUND");
ASSERT_EQ(AllEntriesFor("a"), "[ DEL, b ]"); // Delete issued
batch.Clear();
batch.Delete("c");
batch.Put("c", "d");
dbfull()->Write(WriteOptions(), &batch);
ASSERT_EQ(Get("c"), "d");
ASSERT_EQ(AllEntriesFor("c"), "[ d ]"); // Delete skipped
batch.Clear();
dbfull()->Flush(FlushOptions()); // A stray Flush
batch.Delete("c");
dbfull()->Write(WriteOptions(), &batch);
ASSERT_EQ(AllEntriesFor("c"), "[ DEL, d ]"); // Delete issued
batch.Clear();
delete options.filter_policy;
}
TEST(DBTest, IterEmpty) {
Iterator* iter = db_->NewIterator(ReadOptions());
......@@ -3192,6 +3230,9 @@ static bool CompareIterators(int step,
TEST(DBTest, Randomized) {
Random rnd(test::RandomSeed());
do {
if (CurrentOptions().filter_deletes) {
ChangeOptions(); // DBTest.Randomized not suited for filter_deletes
}
ModelDB model(CurrentOptions());
const int N = 10000;
const Snapshot* model_snap = nullptr;
......
......@@ -39,15 +39,19 @@ TableCache::~TableCache() {
Status TableCache::FindTable(const EnvOptions& toptions,
uint64_t file_number, uint64_t file_size,
Cache::Handle** handle, bool* tableIO) {
Cache::Handle** handle, bool* table_io,
const bool no_io) {
Status s;
char buf[sizeof(file_number)];
EncodeFixed64(buf, file_number);
Slice key(buf, sizeof(buf));
*handle = cache_->Lookup(key);
if (*handle == nullptr) {
if (tableIO != nullptr) {
*tableIO = true; // we had to do IO from storage
if (no_io) { // Dont do IO and return a not-found status
return Status::NotFound("Table not found in table_cache, no_io is set");
}
if (table_io != nullptr) {
*table_io = true; // we had to do IO from storage
}
std::string fname = TableFileName(dbname_, file_number);
unique_ptr<RandomAccessFile> file;
......@@ -112,17 +116,21 @@ Status TableCache::Get(const ReadOptions& options,
const Slice& k,
void* arg,
bool (*saver)(void*, const Slice&, const Slice&, bool),
bool* tableIO,
bool* table_io,
void (*mark_key_may_exist)(void*),
const bool no_IO) {
const bool no_io) {
Cache::Handle* handle = nullptr;
Status s = FindTable(storage_options_, file_number, file_size,
&handle, tableIO);
&handle, table_io, no_io);
if (s.ok()) {
Table* t =
reinterpret_cast<Table*>(cache_->Value(handle));
s = t->InternalGet(options, k, arg, saver, mark_key_may_exist, no_IO);
s = t->InternalGet(options, k, arg, saver, mark_key_may_exist, no_io);
cache_->Release(handle);
} else if (no_io && s.IsNotFound()) {
// Couldnt find Table in cache but treat as kFound if no_io set
(*mark_key_may_exist)(arg);
return Status::OK();
}
return s;
}
......
......@@ -48,9 +48,9 @@ class TableCache {
const Slice& k,
void* arg,
bool (*handle_result)(void*, const Slice&, const Slice&, bool),
bool* tableIO,
bool* table_io,
void (*mark_key_may_exist)(void*) = nullptr,
const bool no_IO = false);
const bool no_io = false);
// Evict any entry for the specified file number
void Evict(uint64_t file_number);
......@@ -62,9 +62,9 @@ class TableCache {
const EnvOptions& storage_options_;
std::shared_ptr<Cache> cache_;
Status FindTable(const EnvOptions& toptions,
uint64_t file_number, uint64_t file_size, Cache::Handle**,
bool* tableIO = nullptr);
Status FindTable(const EnvOptions& toptions, uint64_t file_number,
uint64_t file_size, Cache::Handle**, bool* table_io=nullptr,
const bool no_io = false);
};
} // namespace leveldb
......
......@@ -339,7 +339,7 @@ void Version::Get(const ReadOptions& options,
Status *status,
GetStats* stats,
const Options& db_options,
const bool no_IO) {
const bool no_io) {
Slice ikey = k.internal_key();
Slice user_key = k.user_key();
const Comparator* ucmp = vset_->icmp_.user_comparator();
......@@ -348,7 +348,7 @@ void Version::Get(const ReadOptions& options,
auto logger = db_options.info_log;
assert(status->ok() || status->IsMergeInProgress());
if (no_IO) {
if (no_io) {
assert(status->ok());
}
Saver saver;
......@@ -419,7 +419,7 @@ void Version::Get(const ReadOptions& options,
bool tableIO = false;
*status = vset_->table_cache_->Get(options, f->number, f->file_size,
ikey, &saver, SaveValue, &tableIO,
MarkKeyMayExist, no_IO);
MarkKeyMayExist, no_io);
// TODO: examine the behavior for corrupted key
if (!status->ok()) {
return;
......
......@@ -75,7 +75,7 @@ class Version {
};
void Get(const ReadOptions&, const LookupKey& key, std::string* val,
Status* status, GetStats* stats, const Options& db_option,
const bool no_IO = false);
const bool no_io = false);
// Adds "stats" into the current state. Returns true if a new
// compaction may need to be triggered, false otherwise.
......
......@@ -16,8 +16,9 @@
#include "leveldb/write_batch.h"
#include "leveldb/db.h"
#include "leveldb/statistics.h"
#include "db/dbformat.h"
#include "db/db_impl.h"
#include "db/memtable.h"
#include "db/write_batch_internal.h"
#include "util/coding.h"
......@@ -139,6 +140,23 @@ class MemTableInserter : public WriteBatch::Handler {
public:
SequenceNumber sequence_;
MemTable* mem_;
const Options* options_;
DBImpl* db_;
const bool filter_deletes_;
MemTableInserter(SequenceNumber sequence, MemTable* mem, const Options* opts,
DB* db, const bool filter_deletes)
: sequence_(sequence),
mem_(mem),
options_(opts),
db_(reinterpret_cast<DBImpl*>(db)),
filter_deletes_(filter_deletes) {
assert(mem_);
if (filter_deletes_) {
assert(options_);
assert(db_);
}
}
virtual void Put(const Slice& key, const Slice& value) {
mem_->Add(sequence_, kTypeValue, key, value);
......@@ -149,17 +167,21 @@ class MemTableInserter : public WriteBatch::Handler {
sequence_++;
}
virtual void Delete(const Slice& key) {
if (filter_deletes_ && !db_->KeyMayExistImpl(key, sequence_)) {
RecordTick(options_->statistics, NUMBER_FILTERED_DELETES);
return;
}
mem_->Add(sequence_, kTypeDeletion, key, Slice());
sequence_++;
}
};
} // namespace
Status WriteBatchInternal::InsertInto(const WriteBatch* b,
MemTable* memtable) {
MemTableInserter inserter;
inserter.sequence_ = WriteBatchInternal::Sequence(b);
inserter.mem_ = memtable;
Status WriteBatchInternal::InsertInto(const WriteBatch* b, MemTable* mem,
const Options* opts, DB* db,
const bool filter_deletes) {
MemTableInserter inserter(WriteBatchInternal::Sequence(b), mem, opts, db,
filter_deletes);
return b->Iterate(&inserter);
}
......
......@@ -7,6 +7,8 @@
#include "leveldb/types.h"
#include "leveldb/write_batch.h"
#include "leveldb/db.h"
#include "leveldb/options.h"
namespace leveldb {
......@@ -39,7 +41,12 @@ class WriteBatchInternal {
static void SetContents(WriteBatch* batch, const Slice& contents);
static Status InsertInto(const WriteBatch* batch, MemTable* memtable);
// Inserts batch entries into memtable
// Drops deletes in batch if filter_del is set to true and
// db->KeyMayExist returns false
static Status InsertInto(const WriteBatch* batch, MemTable* memtable,
const Options* opts = nullptr, DB* db = nullptr,
const bool filter_del = false);
static void Append(WriteBatch* dst, const WriteBatch* src);
};
......
......@@ -122,8 +122,12 @@ class DB {
// If the key definitely does not exist in the database, then this method
// returns false. Otherwise return true. This check is potentially
// lighter-weight than invoking DB::Get(). No IO is performed
virtual bool KeyMayExist(const Slice& key) = 0;
// lighter-weight than invoking DB::Get(). One way to make this lighter weight
// is to avoid doing any IOs
// Default implementation here returns true
virtual bool KeyMayExist(const Slice& key) {
return true;
}
// Return a heap-allocated iterator over the contents of the database.
// The result of NewIterator() is initially invalid (caller must
......
......@@ -472,7 +472,7 @@ struct Options {
// only incurs in-memory look up. This optimization avoids writing the delete
// to storage when appropriate.
// Default: false
bool deletes_check_filter_first;
bool filter_deletes;
};
......
......@@ -324,7 +324,7 @@ Status Table::InternalGet(const ReadOptions& options, const Slice& k,
bool (*saver)(void*, const Slice&, const Slice&,
bool),
void (*mark_key_may_exist)(void*),
const bool no_IO) {
const bool no_io) {
Status s;
Iterator* iiter = rep_->index_block->NewIterator(rep_->options.comparator);
bool done = false;
......@@ -340,9 +340,9 @@ Status Table::InternalGet(const ReadOptions& options, const Slice& k,
// cross one data block, we should be fine.
RecordTick(rep_->options.statistics, BLOOM_FILTER_USEFUL);
break;
} else if (no_IO) {
} else if (no_io) {
// Update Saver.state to Found because we are only looking for whether
// bloom-filter can guarantee the key is not there when "no_IO"
// bloom-filter can guarantee the key is not there when "no_io"
(*mark_key_may_exist)(arg);
done = true;
} else {
......
......@@ -88,7 +88,7 @@ class Table {
void* arg,
bool (*handle_result)(void* arg, const Slice& k, const Slice& v, bool),
void (*mark_key_may_exist)(void*) = nullptr,
const bool no_IO = false);
const bool no_io = false);
void ReadMeta(const Footer& footer);
......
......@@ -181,7 +181,7 @@ static uint32_t FLAGS_log2_keys_per_lock = 2; // implies 2^2 keys per lock
static uint32_t FLAGS_purge_redundant_percent = 50;
// On true, deletes use bloom-filter and drop the delete if key not present
static bool FLAGS_deletes_check_filter_first = false;
static bool FLAGS_filter_deletes = false;
// Level0 compaction start trigger
static int FLAGS_level0_file_num_compaction_trigger = 0;
......@@ -904,7 +904,7 @@ class StressTest {
fprintf(stdout, "Purge redundant %% : %d\n",
FLAGS_purge_redundant_percent);
fprintf(stdout, "Deletes use filter : %d\n",
FLAGS_deletes_check_filter_first);
FLAGS_filter_deletes);
fprintf(stdout, "Num keys per lock : %d\n",
1 << FLAGS_log2_keys_per_lock);
......@@ -960,7 +960,7 @@ class StressTest {
options.delete_obsolete_files_period_micros =
FLAGS_delete_obsolete_files_period_micros;
options.max_manifest_file_size = 1024;
options.deletes_check_filter_first = FLAGS_deletes_check_filter_first;
options.filter_deletes = FLAGS_filter_deletes;
static Random purge_percent(1000); // no benefit from non-determinism here
if (purge_percent.Uniform(100) < FLAGS_purge_redundant_percent - 1) {
options.purge_redundant_kvs_while_flush = false;
......@@ -1160,9 +1160,9 @@ int main(int argc, char** argv) {
} else if (sscanf(argv[i], "--purge_redundant_percent=%d%c", &n, &junk) == 1
&& (n >= 0 && n <= 100)) {
FLAGS_purge_redundant_percent = n;
} else if (sscanf(argv[i], "--deletes_check_filter_first=%d%c", &n, &junk)
} else if (sscanf(argv[i], "--filter_deletes=%d%c", &n, &junk)
== 1 && (n == 0 || n == 1)) {
FLAGS_deletes_check_filter_first = n;
FLAGS_filter_deletes = n;
} else {
fprintf(stderr, "Invalid flag '%s'\n", argv[i]);
exit(1);
......
......@@ -75,7 +75,7 @@ Options::Options()
access_hint_on_compaction_start(NORMAL),
use_adaptive_mutex(false),
bytes_per_sync(0),
deletes_check_filter_first(false) {
filter_deletes(false) {
}
static const char* const access_hints[] = {
......@@ -209,8 +209,8 @@ Options::Dump(Logger* log) const
use_adaptive_mutex);
Log(log," Options.bytes_per_sync: %ld",
bytes_per_sync);
Log(log," Options.deletes_check_filter_first: %d",
deletes_check_filter_first);
Log(log," Options.filter_deletes: %d",
filter_deletes);
} // Options::Dump
//
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册