mirror of
https://github.com/facebook/rocksdb.git
synced 2024-11-28 15:33:54 +00:00
af2a36d2c7
Summary: This PR does two things: 1. Adds a new table property `newest_key_time` 2. Uses this property to improve TTL and temperature change compaction. ### Context The current `creation_time` table property should really be named `oldest_ancestor_time`. For flush output files, this is the oldest key time in the file. For compaction output files, this is the minimum among all oldest key times in the input files. The problem with using the oldest ancestor time for TTL compaction is that we may end up dropping files earlier than we should. What we really want is the newest (i.e. "youngest") key time. Right now we take a roundabout way to estimate this value -- we take the value of the _oldest_ key time for the _next_ (newer) SST file. This is also why the current code has checks for `index >= 1`. Our new property `newest_key_time` is set to the file creation time during flushes, and the max over all input files for compactions. There were some additional smaller changes that I had to make for testing purposes: - Refactoring the mock table reader to support specifying my own table properties - Refactoring out a test utility method `GetLevelFileMetadatas` that would otherwise be copy/pasted in 3 places Credit to cbi42 for the problem explanation and proposed solution ### Testing - Added a dedicated unit test to my `newest_key_time` logic in isolation (i.e. are we populating the property on flush and compaction) - Updated the existing unit tests (for TTL/temperate change compaction), which were comprehensive enough to break when I first made my code changes. I removed the test setup code which set the file metadata `oldest_ancestor_time`, so we know we are actually only using the new table property instead. Pull Request resolved: https://github.com/facebook/rocksdb/pull/13083 Reviewed By: cbi42 Differential Revision: D65298604 Pulled By: archang19 fbshipit-source-id: 898ef91b692ab33f5129a2a16b64ecadd4c32432
306 lines
9.8 KiB
C++
306 lines
9.8 KiB
C++
// Copyright (c) 2011-present, Facebook, Inc. All rights reserved.
|
|
// This source code is licensed under both the GPLv2 (found in the
|
|
// COPYING file in the root directory) and Apache 2.0 License
|
|
// (found in the LICENSE.Apache file in the root directory).
|
|
|
|
#include "table/mock_table.h"
|
|
|
|
#include "db/dbformat.h"
|
|
#include "env/composite_env_wrapper.h"
|
|
#include "file/random_access_file_reader.h"
|
|
#include "port/port.h"
|
|
#include "rocksdb/table_properties.h"
|
|
#include "table/get_context.h"
|
|
#include "util/coding.h"
|
|
|
|
namespace ROCKSDB_NAMESPACE::mock {
|
|
|
|
KVVector MakeMockFile(std::initializer_list<KVPair> l) { return KVVector(l); }
|
|
|
|
void SortKVVector(KVVector* kv_vector, const Comparator* ucmp) {
|
|
InternalKeyComparator icmp(ucmp);
|
|
std::sort(kv_vector->begin(), kv_vector->end(),
|
|
[icmp](KVPair a, KVPair b) -> bool {
|
|
return icmp.Compare(a.first, b.first) < 0;
|
|
});
|
|
}
|
|
|
|
class MockTableIterator : public InternalIterator {
|
|
public:
|
|
explicit MockTableIterator(const KVVector& table) : table_(table) {
|
|
itr_ = table_.end();
|
|
}
|
|
|
|
bool Valid() const override { return itr_ != table_.end(); }
|
|
|
|
void SeekToFirst() override { itr_ = table_.begin(); }
|
|
|
|
void SeekToLast() override {
|
|
itr_ = table_.end();
|
|
--itr_;
|
|
}
|
|
|
|
void Seek(const Slice& target) override {
|
|
KVPair target_pair(target.ToString(), "");
|
|
InternalKeyComparator icmp(BytewiseComparator());
|
|
itr_ = std::lower_bound(table_.begin(), table_.end(), target_pair,
|
|
[icmp](KVPair a, KVPair b) -> bool {
|
|
return icmp.Compare(a.first, b.first) < 0;
|
|
});
|
|
}
|
|
|
|
void SeekForPrev(const Slice& target) override {
|
|
KVPair target_pair(target.ToString(), "");
|
|
InternalKeyComparator icmp(BytewiseComparator());
|
|
itr_ = std::upper_bound(table_.begin(), table_.end(), target_pair,
|
|
[icmp](KVPair a, KVPair b) -> bool {
|
|
return icmp.Compare(a.first, b.first) < 0;
|
|
});
|
|
Prev();
|
|
}
|
|
|
|
void Next() override { ++itr_; }
|
|
|
|
void Prev() override {
|
|
if (itr_ == table_.begin()) {
|
|
itr_ = table_.end();
|
|
} else {
|
|
--itr_;
|
|
}
|
|
}
|
|
|
|
Slice key() const override { return Slice(itr_->first); }
|
|
|
|
Slice value() const override { return Slice(itr_->second); }
|
|
|
|
Status status() const override { return Status::OK(); }
|
|
|
|
private:
|
|
const KVVector& table_;
|
|
KVVector::const_iterator itr_;
|
|
};
|
|
|
|
class MockTableBuilder : public TableBuilder {
|
|
public:
|
|
MockTableBuilder(uint32_t id, MockTableFileSystem* file_system,
|
|
MockTableFactory::MockCorruptionMode corrupt_mode =
|
|
MockTableFactory::kCorruptNone,
|
|
size_t key_value_size = 1)
|
|
: id_(id),
|
|
file_system_(file_system),
|
|
corrupt_mode_(corrupt_mode),
|
|
key_value_size_(key_value_size) {
|
|
table_ = MakeMockFile({});
|
|
}
|
|
|
|
// REQUIRES: Either Finish() or Abandon() has been called.
|
|
~MockTableBuilder() = default;
|
|
|
|
// Add key,value to the table being constructed.
|
|
// REQUIRES: key is after any previously added key according to comparator.
|
|
// REQUIRES: Finish(), Abandon() have not been called
|
|
void Add(const Slice& key, const Slice& value) override {
|
|
if (corrupt_mode_ == MockTableFactory::kCorruptValue) {
|
|
// Corrupt the value
|
|
table_.push_back({key.ToString(), value.ToString() + " "});
|
|
corrupt_mode_ = MockTableFactory::kCorruptNone;
|
|
} else if (corrupt_mode_ == MockTableFactory::kCorruptKey) {
|
|
table_.push_back({key.ToString() + " ", value.ToString()});
|
|
corrupt_mode_ = MockTableFactory::kCorruptNone;
|
|
} else if (corrupt_mode_ == MockTableFactory::kCorruptReorderKey) {
|
|
if (prev_key_.empty()) {
|
|
prev_key_ = key.ToString();
|
|
prev_value_ = value.ToString();
|
|
} else {
|
|
table_.push_back({key.ToString(), value.ToString()});
|
|
table_.push_back({prev_key_, prev_value_});
|
|
corrupt_mode_ = MockTableFactory::kCorruptNone;
|
|
}
|
|
} else {
|
|
table_.push_back({key.ToString(), value.ToString()});
|
|
}
|
|
}
|
|
|
|
// Return non-ok iff some error has been detected.
|
|
Status status() const override { return Status::OK(); }
|
|
|
|
// Return non-ok iff some error happens during IO.
|
|
IOStatus io_status() const override { return IOStatus::OK(); }
|
|
|
|
Status Finish() override {
|
|
MutexLock lock_guard(&file_system_->mutex);
|
|
file_system_->files.insert({id_, table_});
|
|
return Status::OK();
|
|
}
|
|
|
|
void Abandon() override {}
|
|
|
|
uint64_t NumEntries() const override { return table_.size(); }
|
|
|
|
uint64_t FileSize() const override { return table_.size() * key_value_size_; }
|
|
|
|
TableProperties GetTableProperties() const override {
|
|
return TableProperties();
|
|
}
|
|
|
|
// Get file checksum
|
|
std::string GetFileChecksum() const override { return kUnknownFileChecksum; }
|
|
// Get file checksum function name
|
|
const char* GetFileChecksumFuncName() const override {
|
|
return kUnknownFileChecksumFuncName;
|
|
}
|
|
|
|
private:
|
|
uint32_t id_;
|
|
std::string prev_key_;
|
|
std::string prev_value_;
|
|
MockTableFileSystem* file_system_;
|
|
int corrupt_mode_;
|
|
KVVector table_;
|
|
size_t key_value_size_;
|
|
};
|
|
|
|
InternalIterator* MockTableReader::NewIterator(
|
|
const ReadOptions&, const SliceTransform* /* prefix_extractor */,
|
|
Arena* /*arena*/, bool /*skip_filters*/, TableReaderCaller /*caller*/,
|
|
size_t /*compaction_readahead_size*/, bool /* allow_unprepared_value */) {
|
|
return new MockTableIterator(table_);
|
|
}
|
|
|
|
Status MockTableReader::Get(const ReadOptions&, const Slice& key,
|
|
GetContext* get_context,
|
|
const SliceTransform* /*prefix_extractor*/,
|
|
bool /*skip_filters*/) {
|
|
std::unique_ptr<MockTableIterator> iter(new MockTableIterator(table_));
|
|
for (iter->Seek(key); iter->Valid(); iter->Next()) {
|
|
ParsedInternalKey parsed_key;
|
|
Status pik_status =
|
|
ParseInternalKey(iter->key(), &parsed_key, true /* log_err_key */);
|
|
if (!pik_status.ok()) {
|
|
return pik_status;
|
|
}
|
|
|
|
bool dont_care __attribute__((__unused__));
|
|
Status read_status;
|
|
bool ret = get_context->SaveValue(parsed_key, iter->value(), &dont_care,
|
|
&read_status);
|
|
if (!read_status.ok()) {
|
|
return read_status;
|
|
}
|
|
if (!ret) {
|
|
break;
|
|
}
|
|
}
|
|
return Status::OK();
|
|
}
|
|
|
|
MockTableFactory::MockTableFactory()
|
|
: next_id_(1), corrupt_mode_(MockTableFactory::kCorruptNone) {}
|
|
|
|
Status MockTableFactory::NewTableReader(
|
|
const ReadOptions& /*ro*/,
|
|
const TableReaderOptions& /*table_reader_options*/,
|
|
std::unique_ptr<RandomAccessFileReader>&& file, uint64_t /*file_size*/,
|
|
std::unique_ptr<TableReader>* table_reader,
|
|
bool /*prefetch_index_and_filter_in_cache*/) const {
|
|
uint32_t id;
|
|
Status s = GetIDFromFile(file.get(), &id);
|
|
if (!s.ok()) {
|
|
return s;
|
|
}
|
|
|
|
MutexLock lock_guard(&file_system_.mutex);
|
|
|
|
auto it = file_system_.files.find(id);
|
|
if (it == file_system_.files.end()) {
|
|
return Status::IOError("Mock file not found");
|
|
}
|
|
|
|
table_reader->reset(new MockTableReader(it->second));
|
|
|
|
return Status::OK();
|
|
}
|
|
|
|
TableBuilder* MockTableFactory::NewTableBuilder(
|
|
const TableBuilderOptions& /*table_builder_options*/,
|
|
WritableFileWriter* file) const {
|
|
uint32_t id;
|
|
Status s = GetAndWriteNextID(file, &id);
|
|
assert(s.ok());
|
|
|
|
return new MockTableBuilder(id, &file_system_, corrupt_mode_,
|
|
key_value_size_);
|
|
}
|
|
|
|
Status MockTableFactory::CreateMockTable(Env* env, const std::string& fname,
|
|
KVVector file_contents) {
|
|
std::unique_ptr<WritableFileWriter> file_writer;
|
|
Status s = WritableFileWriter::Create(env->GetFileSystem(), fname,
|
|
FileOptions(), &file_writer, nullptr);
|
|
if (!s.ok()) {
|
|
return s;
|
|
}
|
|
uint32_t id;
|
|
s = GetAndWriteNextID(file_writer.get(), &id);
|
|
if (s.ok()) {
|
|
file_system_.files.insert({id, std::move(file_contents)});
|
|
}
|
|
return s;
|
|
}
|
|
|
|
Status MockTableFactory::GetAndWriteNextID(WritableFileWriter* file,
|
|
uint32_t* next_id) const {
|
|
*next_id = next_id_.fetch_add(1);
|
|
char buf[4];
|
|
EncodeFixed32(buf, *next_id);
|
|
return file->Append(IOOptions(), Slice(buf, 4));
|
|
}
|
|
|
|
Status MockTableFactory::GetIDFromFile(RandomAccessFileReader* file,
|
|
uint32_t* id) const {
|
|
char buf[4];
|
|
Slice result;
|
|
Status s = file->Read(IOOptions(), 0, 4, &result, buf, nullptr);
|
|
assert(result.size() == 4);
|
|
*id = DecodeFixed32(buf);
|
|
return s;
|
|
}
|
|
|
|
void MockTableFactory::AssertSingleFile(const KVVector& file_contents) {
|
|
ASSERT_EQ(file_system_.files.size(), 1U);
|
|
ASSERT_EQ(file_contents, file_system_.files.begin()->second);
|
|
}
|
|
|
|
void MockTableFactory::AssertLatestFiles(
|
|
const std::vector<KVVector>& files_contents) {
|
|
ASSERT_GE(file_system_.files.size(), files_contents.size());
|
|
auto it = file_system_.files.rbegin();
|
|
for (auto expect = files_contents.rbegin(); expect != files_contents.rend();
|
|
expect++, it++) {
|
|
ASSERT_TRUE(it != file_system_.files.rend());
|
|
if (*expect != it->second) {
|
|
std::cout << "Wrong content! Content of file, expect:" << std::endl;
|
|
for (const auto& kv : *expect) {
|
|
ParsedInternalKey ikey;
|
|
std::string key, value;
|
|
std::tie(key, value) = kv;
|
|
ASSERT_OK(ParseInternalKey(Slice(key), &ikey, true /* log_err_key */));
|
|
std::cout << ikey.DebugString(true, false) << " -> " << value
|
|
<< std::endl;
|
|
}
|
|
std::cout << "actual:" << std::endl;
|
|
for (const auto& kv : it->second) {
|
|
ParsedInternalKey ikey;
|
|
std::string key, value;
|
|
std::tie(key, value) = kv;
|
|
ASSERT_OK(ParseInternalKey(Slice(key), &ikey, true /* log_err_key */));
|
|
std::cout << ikey.DebugString(true, false) << " -> " << value
|
|
<< std::endl;
|
|
}
|
|
FAIL();
|
|
}
|
|
}
|
|
}
|
|
|
|
} // namespace ROCKSDB_NAMESPACE::mock
|