You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
rocksdb/db/memtable.cc

389 lines
13 KiB

// Copyright (c) 2013, Facebook, Inc. All rights reserved.
// This source code is licensed under the BSD-style license found in the
// LICENSE file in the root directory of this source tree. An additional grant
// of patent rights can be found in the PATENTS file in the same directory.
//
// Copyright (c) 2011 The LevelDB Authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file. See the AUTHORS file for names of contributors.
#include "db/memtable.h"
#include <memory>
#include "db/dbformat.h"
#include "db/merge_context.h"
#include "rocksdb/comparator.h"
#include "rocksdb/env.h"
#include "rocksdb/iterator.h"
#include "rocksdb/merge_operator.h"
#include "util/coding.h"
#include "util/murmurhash.h"
#include "util/mutexlock.h"
#include "util/perf_context_imp.h"
#include "util/statistics_imp.h"
#include "util/stop_watch.h"
namespace std {
template <>
struct hash<rocksdb::Slice> {
size_t operator()(const rocksdb::Slice& slice) const {
return MurmurHash(slice.data(), slice.size(), 0);
}
};
}
namespace rocksdb {
MemTable::MemTable(const InternalKeyComparator& cmp,
MemTableRepFactory* table_factory,
int numlevel,
const Options& options)
: comparator_(cmp),
refs_(0),
arena_impl_(options.arena_block_size),
table_(table_factory->CreateMemTableRep(comparator_, &arena_impl_)),
flush_in_progress_(false),
flush_completed_(false),
file_number_(0),
edit_(numlevel),
first_seqno_(0),
mem_next_logfile_number_(0),
mem_logfile_number_(0),
locks_(options.inplace_update_support
? options.inplace_update_num_locks
: 0),
prefix_extractor_(options.prefix_extractor) {
if (prefix_extractor_ && options.memtable_prefix_bloom_bits > 0) {
prefix_bloom_.reset(new DynamicBloom(options.memtable_prefix_bloom_bits,
options.memtable_prefix_bloom_probes));
}
}
MemTable::~MemTable() {
assert(refs_ == 0);
}
size_t MemTable::ApproximateMemoryUsage() {
return arena_impl_.ApproximateMemoryUsage() +
table_->ApproximateMemoryUsage();
}
int MemTable::KeyComparator::operator()(const char* aptr, const char* bptr)
const {
// Internal keys are encoded as length-prefixed strings.
Slice a = GetLengthPrefixedSlice(aptr);
Slice b = GetLengthPrefixedSlice(bptr);
return comparator.Compare(a, b);
}
Slice MemTableRep::UserKey(const char* key) const {
Slice slice = GetLengthPrefixedSlice(key);
return Slice(slice.data(), slice.size() - 8);
}
// Encode a suitable internal key target for "target" and return it.
// Uses *scratch as scratch space, and the returned pointer will point
// into this scratch space.
const char* EncodeKey(std::string* scratch, const Slice& target) {
scratch->clear();
PutVarint32(scratch, target.size());
scratch->append(target.data(), target.size());
return scratch->data();
}
class MemTableIterator: public Iterator {
public:
MemTableIterator(const MemTable& mem, const ReadOptions& options)
: mem_(mem), iter_(), dynamic_prefix_seek_(false), valid_(false) {
if (options.prefix) {
iter_ = mem_.table_->GetPrefixIterator(*options.prefix);
} else if (options.prefix_seek) {
dynamic_prefix_seek_ = true;
iter_ = mem_.table_->GetDynamicPrefixIterator();
} else {
iter_ = mem_.table_->GetIterator();
}
}
virtual bool Valid() const { return valid_; }
virtual void Seek(const Slice& k) {
if (dynamic_prefix_seek_ && mem_.prefix_bloom_ &&
!mem_.prefix_bloom_->MayContain(
mem_.prefix_extractor_->Transform(ExtractUserKey(k)))) {
valid_ = false;
return;
}
iter_->Seek(k, nullptr);
valid_ = iter_->Valid();
}
virtual void SeekToFirst() {
iter_->SeekToFirst();
valid_ = iter_->Valid();
}
virtual void SeekToLast() {
iter_->SeekToLast();
valid_ = iter_->Valid();
}
virtual void Next() {
assert(Valid());
iter_->Next();
valid_ = iter_->Valid();
}
virtual void Prev() {
assert(Valid());
iter_->Prev();
valid_ = iter_->Valid();
}
virtual Slice key() const {
assert(Valid());
return GetLengthPrefixedSlice(iter_->key());
}
virtual Slice value() const {
assert(Valid());
Slice key_slice = GetLengthPrefixedSlice(iter_->key());
return GetLengthPrefixedSlice(key_slice.data() + key_slice.size());
}
virtual Status status() const { return Status::OK(); }
private:
const MemTable& mem_;
std::shared_ptr<MemTableRep::Iterator> iter_;
bool dynamic_prefix_seek_;
bool valid_;
// No copying allowed
MemTableIterator(const MemTableIterator&);
void operator=(const MemTableIterator&);
};
Iterator* MemTable::NewIterator(const ReadOptions& options) {
return new MemTableIterator(*this, options);
}
port::RWMutex* MemTable::GetLock(const Slice& key) {
return &locks_[std::hash<Slice>()(key) % locks_.size()];
}
void MemTable::Add(SequenceNumber s, ValueType type,
const Slice& key, /* user key */
const Slice& value) {
// Format of an entry is concatenation of:
// key_size : varint32 of internal_key.size()
// key bytes : char[internal_key.size()]
// value_size : varint32 of value.size()
// value bytes : char[value.size()]
size_t key_size = key.size();
size_t val_size = value.size();
size_t internal_key_size = key_size + 8;
const size_t encoded_len =
VarintLength(internal_key_size) + internal_key_size +
VarintLength(val_size) + val_size;
char* buf = arena_impl_.Allocate(encoded_len);
char* p = EncodeVarint32(buf, internal_key_size);
memcpy(p, key.data(), key_size);
p += key_size;
EncodeFixed64(p, (s << 8) | type);
p += 8;
p = EncodeVarint32(p, val_size);
memcpy(p, value.data(), val_size);
assert((p + val_size) - buf == (unsigned)encoded_len);
table_->Insert(buf);
if (prefix_bloom_) {
assert(prefix_extractor_);
prefix_bloom_->Add(prefix_extractor_->Transform(key));
}
// The first sequence number inserted into the memtable
assert(first_seqno_ == 0 || s > first_seqno_);
if (first_seqno_ == 0) {
first_seqno_ = s;
}
}
bool MemTable::Get(const LookupKey& key, std::string* value, Status* s,
MergeContext& merge_context, const Options& options) {
StopWatchNano memtable_get_timer(options.env, false);
StartPerfTimer(&memtable_get_timer);
Slice mem_key = key.memtable_key();
Slice user_key = key.user_key();
std::shared_ptr<MemTableRep::Iterator> iter;
if (prefix_bloom_ &&
!prefix_bloom_->MayContain(prefix_extractor_->Transform(user_key))) {
// iter is null if prefix bloom says the key does not exist
} else {
iter = table_->GetIterator(user_key);
iter->Seek(user_key, mem_key.data());
}
[RocksDB] [MergeOperator] The new Merge Interface! Uses merge sequences. Summary: Here are the major changes to the Merge Interface. It has been expanded to handle cases where the MergeOperator is not associative. It does so by stacking up merge operations while scanning through the key history (i.e.: during Get() or Compaction), until a valid Put/Delete/end-of-history is encountered; it then applies all of the merge operations in the correct sequence starting with the base/sentinel value. I have also introduced an "AssociativeMerge" function which allows the user to take advantage of associative merge operations (such as in the case of counters). The implementation will always attempt to merge the operations/operands themselves together when they are encountered, and will resort to the "stacking" method if and only if the "associative-merge" fails. This implementation is conjectured to allow MergeOperator to handle the general case, while still providing the user with the ability to take advantage of certain efficiencies in their own merge-operator / data-structure. NOTE: This is a preliminary diff. This must still go through a lot of review, revision, and testing. Feedback welcome! Test Plan: -This is a preliminary diff. I have only just begun testing/debugging it. -I will be testing this with the existing MergeOperator use-cases and unit-tests (counters, string-append, and redis-lists) -I will be "desk-checking" and walking through the code with the help gdb. -I will find a way of stress-testing the new interface / implementation using db_bench, db_test, merge_test, and/or db_stress. -I will ensure that my tests cover all cases: Get-Memtable, Get-Immutable-Memtable, Get-from-Disk, Iterator-Range-Scan, Flush-Memtable-to-L0, Compaction-L0-L1, Compaction-Ln-L(n+1), Put/Delete found, Put/Delete not-found, end-of-history, end-of-file, etc. -A lot of feedback from the reviewers. Reviewers: haobo, dhruba, zshao, emayanke Reviewed By: haobo CC: leveldb Differential Revision: https://reviews.facebook.net/D11499
11 years ago
bool merge_in_progress = s->IsMergeInProgress();
auto merge_operator = options.merge_operator.get();
auto logger = options.info_log;
[RocksDB] [MergeOperator] The new Merge Interface! Uses merge sequences. Summary: Here are the major changes to the Merge Interface. It has been expanded to handle cases where the MergeOperator is not associative. It does so by stacking up merge operations while scanning through the key history (i.e.: during Get() or Compaction), until a valid Put/Delete/end-of-history is encountered; it then applies all of the merge operations in the correct sequence starting with the base/sentinel value. I have also introduced an "AssociativeMerge" function which allows the user to take advantage of associative merge operations (such as in the case of counters). The implementation will always attempt to merge the operations/operands themselves together when they are encountered, and will resort to the "stacking" method if and only if the "associative-merge" fails. This implementation is conjectured to allow MergeOperator to handle the general case, while still providing the user with the ability to take advantage of certain efficiencies in their own merge-operator / data-structure. NOTE: This is a preliminary diff. This must still go through a lot of review, revision, and testing. Feedback welcome! Test Plan: -This is a preliminary diff. I have only just begun testing/debugging it. -I will be testing this with the existing MergeOperator use-cases and unit-tests (counters, string-append, and redis-lists) -I will be "desk-checking" and walking through the code with the help gdb. -I will find a way of stress-testing the new interface / implementation using db_bench, db_test, merge_test, and/or db_stress. -I will ensure that my tests cover all cases: Get-Memtable, Get-Immutable-Memtable, Get-from-Disk, Iterator-Range-Scan, Flush-Memtable-to-L0, Compaction-L0-L1, Compaction-Ln-L(n+1), Put/Delete found, Put/Delete not-found, end-of-history, end-of-file, etc. -A lot of feedback from the reviewers. Reviewers: haobo, dhruba, zshao, emayanke Reviewed By: haobo CC: leveldb Differential Revision: https://reviews.facebook.net/D11499
11 years ago
std::string merge_result;
bool found_final_value = false;
for (; !found_final_value && iter && iter->Valid(); iter->Next()) {
// entry format is:
// klength varint32
// userkey char[klength-8]
// tag uint64
// vlength varint32
// value char[vlength]
// Check that it belongs to same user key. We do not check the
// sequence number since the Seek() call above should have skipped
// all entries with overly large sequence numbers.
const char* entry = iter->key();
uint32_t key_length;
const char* key_ptr = GetVarint32Ptr(entry, entry + 5, &key_length);
if (comparator_.comparator.user_comparator()->Compare(
Slice(key_ptr, key_length - 8), key.user_key()) == 0) {
// Correct user key
const uint64_t tag = DecodeFixed64(key_ptr + key_length - 8);
switch (static_cast<ValueType>(tag & 0xff)) {
case kTypeValue: {
if (options.inplace_update_support) {
GetLock(key.user_key())->ReadLock();
}
Slice v = GetLengthPrefixedSlice(key_ptr + key_length);
[RocksDB] [MergeOperator] The new Merge Interface! Uses merge sequences. Summary: Here are the major changes to the Merge Interface. It has been expanded to handle cases where the MergeOperator is not associative. It does so by stacking up merge operations while scanning through the key history (i.e.: during Get() or Compaction), until a valid Put/Delete/end-of-history is encountered; it then applies all of the merge operations in the correct sequence starting with the base/sentinel value. I have also introduced an "AssociativeMerge" function which allows the user to take advantage of associative merge operations (such as in the case of counters). The implementation will always attempt to merge the operations/operands themselves together when they are encountered, and will resort to the "stacking" method if and only if the "associative-merge" fails. This implementation is conjectured to allow MergeOperator to handle the general case, while still providing the user with the ability to take advantage of certain efficiencies in their own merge-operator / data-structure. NOTE: This is a preliminary diff. This must still go through a lot of review, revision, and testing. Feedback welcome! Test Plan: -This is a preliminary diff. I have only just begun testing/debugging it. -I will be testing this with the existing MergeOperator use-cases and unit-tests (counters, string-append, and redis-lists) -I will be "desk-checking" and walking through the code with the help gdb. -I will find a way of stress-testing the new interface / implementation using db_bench, db_test, merge_test, and/or db_stress. -I will ensure that my tests cover all cases: Get-Memtable, Get-Immutable-Memtable, Get-from-Disk, Iterator-Range-Scan, Flush-Memtable-to-L0, Compaction-L0-L1, Compaction-Ln-L(n+1), Put/Delete found, Put/Delete not-found, end-of-history, end-of-file, etc. -A lot of feedback from the reviewers. Reviewers: haobo, dhruba, zshao, emayanke Reviewed By: haobo CC: leveldb Differential Revision: https://reviews.facebook.net/D11499
11 years ago
*s = Status::OK();
if (merge_in_progress) {
[RocksDB] [MergeOperator] The new Merge Interface! Uses merge sequences. Summary: Here are the major changes to the Merge Interface. It has been expanded to handle cases where the MergeOperator is not associative. It does so by stacking up merge operations while scanning through the key history (i.e.: during Get() or Compaction), until a valid Put/Delete/end-of-history is encountered; it then applies all of the merge operations in the correct sequence starting with the base/sentinel value. I have also introduced an "AssociativeMerge" function which allows the user to take advantage of associative merge operations (such as in the case of counters). The implementation will always attempt to merge the operations/operands themselves together when they are encountered, and will resort to the "stacking" method if and only if the "associative-merge" fails. This implementation is conjectured to allow MergeOperator to handle the general case, while still providing the user with the ability to take advantage of certain efficiencies in their own merge-operator / data-structure. NOTE: This is a preliminary diff. This must still go through a lot of review, revision, and testing. Feedback welcome! Test Plan: -This is a preliminary diff. I have only just begun testing/debugging it. -I will be testing this with the existing MergeOperator use-cases and unit-tests (counters, string-append, and redis-lists) -I will be "desk-checking" and walking through the code with the help gdb. -I will find a way of stress-testing the new interface / implementation using db_bench, db_test, merge_test, and/or db_stress. -I will ensure that my tests cover all cases: Get-Memtable, Get-Immutable-Memtable, Get-from-Disk, Iterator-Range-Scan, Flush-Memtable-to-L0, Compaction-L0-L1, Compaction-Ln-L(n+1), Put/Delete found, Put/Delete not-found, end-of-history, end-of-file, etc. -A lot of feedback from the reviewers. Reviewers: haobo, dhruba, zshao, emayanke Reviewed By: haobo CC: leveldb Differential Revision: https://reviews.facebook.net/D11499
11 years ago
assert(merge_operator);
if (!merge_operator->FullMerge(key.user_key(), &v,
merge_context.GetOperands(), value,
logger.get())) {
RecordTick(options.statistics.get(), NUMBER_MERGE_FAILURES);
[RocksDB] [MergeOperator] The new Merge Interface! Uses merge sequences. Summary: Here are the major changes to the Merge Interface. It has been expanded to handle cases where the MergeOperator is not associative. It does so by stacking up merge operations while scanning through the key history (i.e.: during Get() or Compaction), until a valid Put/Delete/end-of-history is encountered; it then applies all of the merge operations in the correct sequence starting with the base/sentinel value. I have also introduced an "AssociativeMerge" function which allows the user to take advantage of associative merge operations (such as in the case of counters). The implementation will always attempt to merge the operations/operands themselves together when they are encountered, and will resort to the "stacking" method if and only if the "associative-merge" fails. This implementation is conjectured to allow MergeOperator to handle the general case, while still providing the user with the ability to take advantage of certain efficiencies in their own merge-operator / data-structure. NOTE: This is a preliminary diff. This must still go through a lot of review, revision, and testing. Feedback welcome! Test Plan: -This is a preliminary diff. I have only just begun testing/debugging it. -I will be testing this with the existing MergeOperator use-cases and unit-tests (counters, string-append, and redis-lists) -I will be "desk-checking" and walking through the code with the help gdb. -I will find a way of stress-testing the new interface / implementation using db_bench, db_test, merge_test, and/or db_stress. -I will ensure that my tests cover all cases: Get-Memtable, Get-Immutable-Memtable, Get-from-Disk, Iterator-Range-Scan, Flush-Memtable-to-L0, Compaction-L0-L1, Compaction-Ln-L(n+1), Put/Delete found, Put/Delete not-found, end-of-history, end-of-file, etc. -A lot of feedback from the reviewers. Reviewers: haobo, dhruba, zshao, emayanke Reviewed By: haobo CC: leveldb Differential Revision: https://reviews.facebook.net/D11499
11 years ago
*s = Status::Corruption("Error: Could not perform merge.");
}
} else {
value->assign(v.data(), v.size());
}
if (options.inplace_update_support) {
GetLock(key.user_key())->Unlock();
}
found_final_value = true;
break;
}
case kTypeDeletion: {
if (merge_in_progress) {
[RocksDB] [MergeOperator] The new Merge Interface! Uses merge sequences. Summary: Here are the major changes to the Merge Interface. It has been expanded to handle cases where the MergeOperator is not associative. It does so by stacking up merge operations while scanning through the key history (i.e.: during Get() or Compaction), until a valid Put/Delete/end-of-history is encountered; it then applies all of the merge operations in the correct sequence starting with the base/sentinel value. I have also introduced an "AssociativeMerge" function which allows the user to take advantage of associative merge operations (such as in the case of counters). The implementation will always attempt to merge the operations/operands themselves together when they are encountered, and will resort to the "stacking" method if and only if the "associative-merge" fails. This implementation is conjectured to allow MergeOperator to handle the general case, while still providing the user with the ability to take advantage of certain efficiencies in their own merge-operator / data-structure. NOTE: This is a preliminary diff. This must still go through a lot of review, revision, and testing. Feedback welcome! Test Plan: -This is a preliminary diff. I have only just begun testing/debugging it. -I will be testing this with the existing MergeOperator use-cases and unit-tests (counters, string-append, and redis-lists) -I will be "desk-checking" and walking through the code with the help gdb. -I will find a way of stress-testing the new interface / implementation using db_bench, db_test, merge_test, and/or db_stress. -I will ensure that my tests cover all cases: Get-Memtable, Get-Immutable-Memtable, Get-from-Disk, Iterator-Range-Scan, Flush-Memtable-to-L0, Compaction-L0-L1, Compaction-Ln-L(n+1), Put/Delete found, Put/Delete not-found, end-of-history, end-of-file, etc. -A lot of feedback from the reviewers. Reviewers: haobo, dhruba, zshao, emayanke Reviewed By: haobo CC: leveldb Differential Revision: https://reviews.facebook.net/D11499
11 years ago
assert(merge_operator);
*s = Status::OK();
if (!merge_operator->FullMerge(key.user_key(), nullptr,
merge_context.GetOperands(), value,
logger.get())) {
RecordTick(options.statistics.get(), NUMBER_MERGE_FAILURES);
[RocksDB] [MergeOperator] The new Merge Interface! Uses merge sequences. Summary: Here are the major changes to the Merge Interface. It has been expanded to handle cases where the MergeOperator is not associative. It does so by stacking up merge operations while scanning through the key history (i.e.: during Get() or Compaction), until a valid Put/Delete/end-of-history is encountered; it then applies all of the merge operations in the correct sequence starting with the base/sentinel value. I have also introduced an "AssociativeMerge" function which allows the user to take advantage of associative merge operations (such as in the case of counters). The implementation will always attempt to merge the operations/operands themselves together when they are encountered, and will resort to the "stacking" method if and only if the "associative-merge" fails. This implementation is conjectured to allow MergeOperator to handle the general case, while still providing the user with the ability to take advantage of certain efficiencies in their own merge-operator / data-structure. NOTE: This is a preliminary diff. This must still go through a lot of review, revision, and testing. Feedback welcome! Test Plan: -This is a preliminary diff. I have only just begun testing/debugging it. -I will be testing this with the existing MergeOperator use-cases and unit-tests (counters, string-append, and redis-lists) -I will be "desk-checking" and walking through the code with the help gdb. -I will find a way of stress-testing the new interface / implementation using db_bench, db_test, merge_test, and/or db_stress. -I will ensure that my tests cover all cases: Get-Memtable, Get-Immutable-Memtable, Get-from-Disk, Iterator-Range-Scan, Flush-Memtable-to-L0, Compaction-L0-L1, Compaction-Ln-L(n+1), Put/Delete found, Put/Delete not-found, end-of-history, end-of-file, etc. -A lot of feedback from the reviewers. Reviewers: haobo, dhruba, zshao, emayanke Reviewed By: haobo CC: leveldb Differential Revision: https://reviews.facebook.net/D11499
11 years ago
*s = Status::Corruption("Error: Could not perform merge.");
}
} else {
*s = Status::NotFound(Slice());
}
found_final_value = true;
break;
}
[RocksDB] [MergeOperator] The new Merge Interface! Uses merge sequences. Summary: Here are the major changes to the Merge Interface. It has been expanded to handle cases where the MergeOperator is not associative. It does so by stacking up merge operations while scanning through the key history (i.e.: during Get() or Compaction), until a valid Put/Delete/end-of-history is encountered; it then applies all of the merge operations in the correct sequence starting with the base/sentinel value. I have also introduced an "AssociativeMerge" function which allows the user to take advantage of associative merge operations (such as in the case of counters). The implementation will always attempt to merge the operations/operands themselves together when they are encountered, and will resort to the "stacking" method if and only if the "associative-merge" fails. This implementation is conjectured to allow MergeOperator to handle the general case, while still providing the user with the ability to take advantage of certain efficiencies in their own merge-operator / data-structure. NOTE: This is a preliminary diff. This must still go through a lot of review, revision, and testing. Feedback welcome! Test Plan: -This is a preliminary diff. I have only just begun testing/debugging it. -I will be testing this with the existing MergeOperator use-cases and unit-tests (counters, string-append, and redis-lists) -I will be "desk-checking" and walking through the code with the help gdb. -I will find a way of stress-testing the new interface / implementation using db_bench, db_test, merge_test, and/or db_stress. -I will ensure that my tests cover all cases: Get-Memtable, Get-Immutable-Memtable, Get-from-Disk, Iterator-Range-Scan, Flush-Memtable-to-L0, Compaction-L0-L1, Compaction-Ln-L(n+1), Put/Delete found, Put/Delete not-found, end-of-history, end-of-file, etc. -A lot of feedback from the reviewers. Reviewers: haobo, dhruba, zshao, emayanke Reviewed By: haobo CC: leveldb Differential Revision: https://reviews.facebook.net/D11499
11 years ago
case kTypeMerge: {
Slice v = GetLengthPrefixedSlice(key_ptr + key_length);
merge_in_progress = true;
merge_context.PushOperand(v);
while(merge_context.GetNumOperands() >= 2) {
[RocksDB] [MergeOperator] The new Merge Interface! Uses merge sequences. Summary: Here are the major changes to the Merge Interface. It has been expanded to handle cases where the MergeOperator is not associative. It does so by stacking up merge operations while scanning through the key history (i.e.: during Get() or Compaction), until a valid Put/Delete/end-of-history is encountered; it then applies all of the merge operations in the correct sequence starting with the base/sentinel value. I have also introduced an "AssociativeMerge" function which allows the user to take advantage of associative merge operations (such as in the case of counters). The implementation will always attempt to merge the operations/operands themselves together when they are encountered, and will resort to the "stacking" method if and only if the "associative-merge" fails. This implementation is conjectured to allow MergeOperator to handle the general case, while still providing the user with the ability to take advantage of certain efficiencies in their own merge-operator / data-structure. NOTE: This is a preliminary diff. This must still go through a lot of review, revision, and testing. Feedback welcome! Test Plan: -This is a preliminary diff. I have only just begun testing/debugging it. -I will be testing this with the existing MergeOperator use-cases and unit-tests (counters, string-append, and redis-lists) -I will be "desk-checking" and walking through the code with the help gdb. -I will find a way of stress-testing the new interface / implementation using db_bench, db_test, merge_test, and/or db_stress. -I will ensure that my tests cover all cases: Get-Memtable, Get-Immutable-Memtable, Get-from-Disk, Iterator-Range-Scan, Flush-Memtable-to-L0, Compaction-L0-L1, Compaction-Ln-L(n+1), Put/Delete found, Put/Delete not-found, end-of-history, end-of-file, etc. -A lot of feedback from the reviewers. Reviewers: haobo, dhruba, zshao, emayanke Reviewed By: haobo CC: leveldb Differential Revision: https://reviews.facebook.net/D11499
11 years ago
// Attempt to associative merge. (Returns true if successful)
if (merge_operator->PartialMerge(key.user_key(),
merge_context.GetOperand(0),
merge_context.GetOperand(1),
&merge_result, logger.get())) {
merge_context.PushPartialMergeResult(merge_result);
[RocksDB] [MergeOperator] The new Merge Interface! Uses merge sequences. Summary: Here are the major changes to the Merge Interface. It has been expanded to handle cases where the MergeOperator is not associative. It does so by stacking up merge operations while scanning through the key history (i.e.: during Get() or Compaction), until a valid Put/Delete/end-of-history is encountered; it then applies all of the merge operations in the correct sequence starting with the base/sentinel value. I have also introduced an "AssociativeMerge" function which allows the user to take advantage of associative merge operations (such as in the case of counters). The implementation will always attempt to merge the operations/operands themselves together when they are encountered, and will resort to the "stacking" method if and only if the "associative-merge" fails. This implementation is conjectured to allow MergeOperator to handle the general case, while still providing the user with the ability to take advantage of certain efficiencies in their own merge-operator / data-structure. NOTE: This is a preliminary diff. This must still go through a lot of review, revision, and testing. Feedback welcome! Test Plan: -This is a preliminary diff. I have only just begun testing/debugging it. -I will be testing this with the existing MergeOperator use-cases and unit-tests (counters, string-append, and redis-lists) -I will be "desk-checking" and walking through the code with the help gdb. -I will find a way of stress-testing the new interface / implementation using db_bench, db_test, merge_test, and/or db_stress. -I will ensure that my tests cover all cases: Get-Memtable, Get-Immutable-Memtable, Get-from-Disk, Iterator-Range-Scan, Flush-Memtable-to-L0, Compaction-L0-L1, Compaction-Ln-L(n+1), Put/Delete found, Put/Delete not-found, end-of-history, end-of-file, etc. -A lot of feedback from the reviewers. Reviewers: haobo, dhruba, zshao, emayanke Reviewed By: haobo CC: leveldb Differential Revision: https://reviews.facebook.net/D11499
11 years ago
} else {
// Stack them because user can't associative merge
break;
}
}
break;
}
case kTypeLogData:
assert(false);
break;
}
} else {
// exit loop if user key does not match
break;
}
}
[RocksDB] [MergeOperator] The new Merge Interface! Uses merge sequences. Summary: Here are the major changes to the Merge Interface. It has been expanded to handle cases where the MergeOperator is not associative. It does so by stacking up merge operations while scanning through the key history (i.e.: during Get() or Compaction), until a valid Put/Delete/end-of-history is encountered; it then applies all of the merge operations in the correct sequence starting with the base/sentinel value. I have also introduced an "AssociativeMerge" function which allows the user to take advantage of associative merge operations (such as in the case of counters). The implementation will always attempt to merge the operations/operands themselves together when they are encountered, and will resort to the "stacking" method if and only if the "associative-merge" fails. This implementation is conjectured to allow MergeOperator to handle the general case, while still providing the user with the ability to take advantage of certain efficiencies in their own merge-operator / data-structure. NOTE: This is a preliminary diff. This must still go through a lot of review, revision, and testing. Feedback welcome! Test Plan: -This is a preliminary diff. I have only just begun testing/debugging it. -I will be testing this with the existing MergeOperator use-cases and unit-tests (counters, string-append, and redis-lists) -I will be "desk-checking" and walking through the code with the help gdb. -I will find a way of stress-testing the new interface / implementation using db_bench, db_test, merge_test, and/or db_stress. -I will ensure that my tests cover all cases: Get-Memtable, Get-Immutable-Memtable, Get-from-Disk, Iterator-Range-Scan, Flush-Memtable-to-L0, Compaction-L0-L1, Compaction-Ln-L(n+1), Put/Delete found, Put/Delete not-found, end-of-history, end-of-file, etc. -A lot of feedback from the reviewers. Reviewers: haobo, dhruba, zshao, emayanke Reviewed By: haobo CC: leveldb Differential Revision: https://reviews.facebook.net/D11499
11 years ago
// No change to value, since we have not yet found a Put/Delete
if (!found_final_value && merge_in_progress) {
*s = Status::MergeInProgress("");
}
BumpPerfTime(&perf_context.get_from_memtable_time, &memtable_get_timer);
BumpPerfCount(&perf_context.get_from_memtable_count);
return found_final_value;
}
bool MemTable::Update(SequenceNumber seq, ValueType type,
const Slice& key,
const Slice& value) {
LookupKey lkey(key, seq);
Slice mem_key = lkey.memtable_key();
std::shared_ptr<MemTableRep::Iterator> iter(
table_->GetIterator(lkey.user_key()));
iter->Seek(key, mem_key.data());
if (iter->Valid()) {
// entry format is:
// klength varint32
// userkey char[klength-8]
// tag uint64
// vlength varint32
// value char[vlength]
// Check that it belongs to same user key. We do not check the
// sequence number since the Seek() call above should have skipped
// all entries with overly large sequence numbers.
const char* entry = iter->key();
uint32_t key_length;
const char* key_ptr = GetVarint32Ptr(entry, entry + 5, &key_length);
if (comparator_.comparator.user_comparator()->Compare(
Slice(key_ptr, key_length - 8), lkey.user_key()) == 0) {
// Correct user key
const uint64_t tag = DecodeFixed64(key_ptr + key_length - 8);
switch (static_cast<ValueType>(tag & 0xff)) {
case kTypeValue: {
uint32_t vlength;
GetVarint32Ptr(key_ptr + key_length,
key_ptr + key_length+5, &vlength);
// Update value, if newValue size <= curValue size
if (value.size() <= vlength) {
char* p = EncodeVarint32(const_cast<char*>(key_ptr) + key_length,
value.size());
WriteLock wl(GetLock(lkey.user_key()));
memcpy(p, value.data(), value.size());
assert(
(p + value.size()) - entry ==
(unsigned) (VarintLength(key_length) +
key_length +
VarintLength(value.size()) +
value.size())
);
// no need to update bloom, as user key does not change.
return true;
}
}
default:
// If the latest value is kTypeDeletion, kTypeMerge or kTypeLogData
// then we probably don't have enough space to update in-place
// Maybe do something later
// Return false, and do normal Add()
return false;
}
}
}
// Key doesn't exist
return false;
}
} // namespace rocksdb