Вы не можете выбрать более 25 тем
Темы должны начинаться с буквы или цифры, могут содержать дефисы(-) и должны содержать не более 35 символов.
2074 строки
66 KiB
2074 строки
66 KiB
// 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/db_impl.h"
|
|
|
|
#include <algorithm>
|
|
#include <atomic>
|
|
#include <cstdint>
|
|
#include <cstdio>
|
|
#include <set>
|
|
#include <string>
|
|
#include <vector>
|
|
#include <iostream>
|
|
|
|
#include "fields.h"
|
|
#include "db/builder.h"
|
|
#include "db/db_iter.h"
|
|
#include "db/dbformat.h"
|
|
#include "db/filename.h"
|
|
#include "db/log_reader.h"
|
|
#include "db/log_writer.h"
|
|
#include "db/memtable.h"
|
|
#include "db/table_cache.h"
|
|
#include "db/version_set.h"
|
|
#include "db/write_batch_internal.h"
|
|
#include "leveldb/db.h"
|
|
#include "leveldb/env.h"
|
|
#include "leveldb/status.h"
|
|
#include "leveldb/table.h"
|
|
#include "leveldb/table_builder.h"
|
|
#include "port/port.h"
|
|
#include "table/block.h"
|
|
#include "table/merger.h"
|
|
#include "table/two_level_iterator.h"
|
|
#include "util/coding.h"
|
|
#include "util/logging.h"
|
|
#include "util/mutexlock.h"
|
|
|
|
#include "db/vlog_reader.h"
|
|
|
|
namespace leveldb {
|
|
|
|
using namespace log;
|
|
|
|
const int kNumNonTableCacheFiles = 10;
|
|
|
|
// Information kept for every waiting writer
|
|
struct DBImpl::Writer {
|
|
explicit Writer(port::Mutex* mu)
|
|
: batch(nullptr), sync(false), done(false), cv(mu) {}
|
|
|
|
Status status;
|
|
WriteBatch* batch;
|
|
bool sync;
|
|
bool done;
|
|
port::CondVar cv;
|
|
};
|
|
|
|
struct DBImpl::CompactionState {
|
|
// Files produced by compaction
|
|
struct Output {
|
|
uint64_t number;
|
|
uint64_t file_size;
|
|
InternalKey smallest, largest;
|
|
};
|
|
|
|
Output* current_output() { return &outputs[outputs.size() - 1]; }
|
|
|
|
explicit CompactionState(Compaction* c)
|
|
: compaction(c),
|
|
smallest_snapshot(0),
|
|
outfile(nullptr),
|
|
builder(nullptr),
|
|
total_bytes(0) {}
|
|
|
|
Compaction* const compaction;
|
|
|
|
// Sequence numbers < smallest_snapshot are not significant since we
|
|
// will never have to service a snapshot below smallest_snapshot.
|
|
// Therefore if we have seen a sequence number S <= smallest_snapshot,
|
|
// we can drop all entries for the same key with sequence numbers < S.
|
|
SequenceNumber smallest_snapshot;
|
|
|
|
std::vector<Output> outputs;
|
|
|
|
// State kept for output being generated
|
|
WritableFile* outfile;
|
|
TableBuilder* builder;
|
|
|
|
uint64_t total_bytes;
|
|
};
|
|
|
|
// Fix user-supplied options to be reasonable
|
|
template <class T, class V>
|
|
static void ClipToRange(T* ptr, V minvalue, V maxvalue) {
|
|
if (static_cast<V>(*ptr) > maxvalue) *ptr = maxvalue;
|
|
if (static_cast<V>(*ptr) < minvalue) *ptr = minvalue;
|
|
}
|
|
Options SanitizeOptions(const std::string& dbname,
|
|
const InternalKeyComparator* icmp,
|
|
const InternalFilterPolicy* ipolicy,
|
|
const Options& src) {
|
|
Options result = src;
|
|
result.comparator = icmp;
|
|
result.filter_policy = (src.filter_policy != nullptr) ? ipolicy : nullptr;
|
|
ClipToRange(&result.max_open_files, 64 + kNumNonTableCacheFiles, 50000);
|
|
ClipToRange(&result.write_buffer_size, 64 << 10, 1 << 30);
|
|
ClipToRange(&result.max_file_size, 1 << 20, 1 << 30);
|
|
ClipToRange(&result.block_size, 1 << 10, 4 << 20);
|
|
if (result.info_log == nullptr) {
|
|
// Open a log file in the same directory as the db
|
|
src.env->CreateDir(dbname); // In case it does not exist
|
|
src.env->RenameFile(InfoLogFileName(dbname), OldInfoLogFileName(dbname));
|
|
Status s = src.env->NewLogger(InfoLogFileName(dbname), &result.info_log);
|
|
if (!s.ok()) {
|
|
// No place suitable for logging
|
|
result.info_log = nullptr;
|
|
}
|
|
}
|
|
if (result.block_cache == nullptr) {
|
|
result.block_cache = NewLRUCache(8 << 20);
|
|
}
|
|
return result;
|
|
}
|
|
|
|
static int TableCacheSize(const Options& sanitized_options) {
|
|
// Reserve ten files or so for other uses and give the rest to TableCache.
|
|
return sanitized_options.max_open_files - kNumNonTableCacheFiles;
|
|
}
|
|
|
|
DBImpl::DBImpl(const Options& raw_options, const std::string& dbname)
|
|
: env_(raw_options.env),
|
|
internal_comparator_(raw_options.comparator),
|
|
internal_filter_policy_(raw_options.filter_policy),
|
|
options_(SanitizeOptions(dbname, &internal_comparator_,
|
|
&internal_filter_policy_, raw_options)),
|
|
owns_info_log_(options_.info_log != raw_options.info_log),
|
|
owns_cache_(options_.block_cache != raw_options.block_cache),
|
|
dbname_(dbname),
|
|
table_cache_(new TableCache(dbname_, options_, TableCacheSize(options_))),
|
|
db_lock_(nullptr),
|
|
shutting_down_(false),
|
|
background_work_finished_signal_(&mutex_),
|
|
|
|
garbage_collection_work_signal_(&mutex_),
|
|
|
|
mem_(nullptr),
|
|
imm_(nullptr),
|
|
has_imm_(false),
|
|
logfile_(nullptr),
|
|
logfile_number_(0),
|
|
seed_(0),
|
|
tmp_batch_(new WriteBatch),
|
|
|
|
background_compaction_scheduled_(false),
|
|
background_GarbageCollection_scheduled_(false),
|
|
finish_back_garbage_collection_(false),
|
|
manual_compaction_(nullptr),
|
|
|
|
vlog_(nullptr),
|
|
vlog_kv_numbers_(0),
|
|
garbage_collection_management_(new SeparateManagement(raw_options.garbage_collection_threshold) ),
|
|
versions_(new VersionSet(dbname_, &options_, table_cache_,
|
|
&internal_comparator_)) {}
|
|
|
|
DBImpl::~DBImpl() {
|
|
// Wait for background work to finish.
|
|
mutex_.Lock();
|
|
shutting_down_.store(true, std::memory_order_release);
|
|
while (background_compaction_scheduled_) {
|
|
background_work_finished_signal_.Wait();
|
|
}
|
|
while(background_GarbageCollection_scheduled_){
|
|
garbage_collection_work_signal_.Wait();
|
|
}
|
|
mutex_.Unlock();
|
|
|
|
if (db_lock_ != nullptr) {
|
|
env_->UnlockFile(db_lock_);
|
|
}
|
|
|
|
delete versions_;
|
|
if (mem_ != nullptr) mem_->Unref();
|
|
if (imm_ != nullptr) imm_->Unref();
|
|
delete tmp_batch_;
|
|
delete vlog_;
|
|
delete logfile_;
|
|
delete table_cache_;
|
|
|
|
if (owns_info_log_) {
|
|
delete options_.info_log;
|
|
}
|
|
if (owns_cache_) {
|
|
delete options_.block_cache;
|
|
}
|
|
}
|
|
|
|
Status DBImpl::NewDB() {
|
|
VersionEdit new_db;
|
|
new_db.SetComparatorName(user_comparator()->Name());
|
|
new_db.SetLogNumber(0);
|
|
new_db.SetNextFile(2);
|
|
new_db.SetLastSequence(0);
|
|
|
|
const std::string manifest = DescriptorFileName(dbname_, 1);
|
|
WritableFile* file;
|
|
Status s = env_->NewWritableFile(manifest, &file);
|
|
if (!s.ok()) {
|
|
return s;
|
|
}
|
|
{
|
|
log::Writer log(file);
|
|
std::string record;
|
|
new_db.EncodeTo(&record);
|
|
s = log.AddRecord(record);
|
|
if (s.ok()) {
|
|
s = file->Sync();
|
|
}
|
|
if (s.ok()) {
|
|
s = file->Close();
|
|
}
|
|
}
|
|
delete file;
|
|
if (s.ok()) {
|
|
// Make "CURRENT" file that points to the new manifest file.
|
|
s = SetCurrentFile(env_, dbname_, 1);
|
|
} else {
|
|
env_->RemoveFile(manifest);
|
|
}
|
|
return s;
|
|
}
|
|
|
|
Env* DBImpl::GetEnv() const {
|
|
return env_;
|
|
}
|
|
|
|
std::string DBImpl::GetDBName() const {
|
|
return dbname_;
|
|
}
|
|
|
|
void DBImpl::MaybeIgnoreError(Status* s) const {
|
|
if (s->ok() || options_.paranoid_checks) {
|
|
// No change needed
|
|
} else {
|
|
Log(options_.info_log, "Ignoring error %s", s->ToString().c_str());
|
|
*s = Status::OK();
|
|
}
|
|
}
|
|
|
|
void DBImpl::RemoveObsoleteFiles() {
|
|
mutex_.AssertHeld();
|
|
|
|
if (!bg_error_.ok()) {
|
|
// After a background error, we don't know whether a new version may
|
|
// or may not have been committed, so we cannot safely garbage collect.
|
|
return;
|
|
}
|
|
|
|
// Make a set of all of the live files
|
|
std::set<uint64_t> live = pending_outputs_;
|
|
versions_->AddLiveFiles(&live);
|
|
|
|
std::vector<std::string> filenames;
|
|
env_->GetChildren(dbname_, &filenames); // Ignoring errors on purpose
|
|
uint64_t number;
|
|
FileType type;
|
|
std::vector<std::string> files_to_delete;
|
|
for (std::string& filename : filenames) {
|
|
if (ParseFileName(filename, &number, &type)) {
|
|
bool keep = true;
|
|
switch (type) {
|
|
case kLogFile:
|
|
keep = ((number >= versions_->LogNumber()) ||
|
|
(number == versions_->PrevLogNumber()));
|
|
break;
|
|
case kDescriptorFile:
|
|
// Keep my manifest file, and any newer incarnations'
|
|
// (in case there is a race that allows other incarnations)
|
|
keep = (number >= versions_->ManifestFileNumber());
|
|
break;
|
|
case kTableFile:
|
|
keep = (live.find(number) != live.end());
|
|
break;
|
|
case kTempFile:
|
|
// Any temp files that are currently being written to must
|
|
// be recorded in pending_outputs_, which is inserted into "live"
|
|
keep = (live.find(number) != live.end());
|
|
break;
|
|
case kCurrentFile:
|
|
case kDBLockFile:
|
|
case kInfoLogFile:
|
|
keep = true;
|
|
break;
|
|
}
|
|
|
|
if (!keep) {
|
|
files_to_delete.push_back(std::move(filename));
|
|
if (type == kTableFile) {
|
|
table_cache_->Evict(number);
|
|
}
|
|
Log(options_.info_log, "Delete type=%d #%lld\n", static_cast<int>(type),
|
|
static_cast<unsigned long long>(number));
|
|
}
|
|
}
|
|
}
|
|
|
|
// While deleting all files unblock other threads. All files being deleted
|
|
// have unique names which will not collide with newly created files and
|
|
// are therefore safe to delete while allowing other threads to proceed.
|
|
mutex_.Unlock();
|
|
for (const std::string& filename : files_to_delete) {
|
|
env_->RemoveFile(dbname_ + "/" + filename);
|
|
}
|
|
mutex_.Lock();
|
|
}
|
|
|
|
Status DBImpl::Recover(VersionEdit* edit, bool* save_manifest) {
|
|
mutex_.AssertHeld();
|
|
|
|
// Ignore error from CreateDir since the creation of the DB is
|
|
// committed only when the descriptor is created, and this directory
|
|
// may already exist from a previous failed creation attempt.
|
|
env_->CreateDir(dbname_);
|
|
assert(db_lock_ == nullptr);
|
|
Status s = env_->LockFile(LockFileName(dbname_), &db_lock_);
|
|
if (!s.ok()) {
|
|
return s;
|
|
}
|
|
|
|
if (!env_->FileExists(CurrentFileName(dbname_))) {
|
|
if (options_.create_if_missing) {
|
|
Log(options_.info_log, "Creating DB %s since it was missing.",
|
|
dbname_.c_str());
|
|
s = NewDB();
|
|
if (!s.ok()) {
|
|
return s;
|
|
}
|
|
} else {
|
|
return Status::InvalidArgument(
|
|
dbname_, "does not exist (create_if_missing is false)");
|
|
}
|
|
} else {
|
|
if (options_.error_if_exists) {
|
|
return Status::InvalidArgument(dbname_,
|
|
"exists (error_if_exists is true)");
|
|
}
|
|
}
|
|
|
|
s = versions_->Recover(save_manifest);
|
|
if (!s.ok()) {
|
|
return s;
|
|
}
|
|
SequenceNumber max_sequence(0);
|
|
|
|
// Recover from all newer log files than the ones named in the
|
|
// descriptor (new log files may have been added by the previous
|
|
// incarnation without registering them in the descriptor).
|
|
//
|
|
// Note that PrevLogNumber() is no longer used, but we pay
|
|
// attention to it in case we are recovering a database
|
|
// produced by an older version of leveldb.
|
|
const uint64_t min_log = versions_->LogNumber();
|
|
const uint64_t prev_log = versions_->PrevLogNumber();
|
|
std::vector<std::string> filenames;
|
|
s = env_->GetChildren(dbname_, &filenames);
|
|
if (!s.ok()) {
|
|
return s;
|
|
}
|
|
std::set<uint64_t> expected;
|
|
versions_->AddLiveFiles(&expected);
|
|
uint64_t number;
|
|
FileType type;
|
|
std::vector<uint64_t> logs;
|
|
uint64_t max_number = 0;
|
|
for (size_t i = 0; i < filenames.size(); i++) {
|
|
if (ParseFileName(filenames[i], &number, &type)) {
|
|
// begin 注释： max_number 为要恢复的最新的文件编号
|
|
if (number > max_number) max_number = number;
|
|
// expected 里的文件现在依然存在，可以删除
|
|
expected.erase(number);
|
|
// 保存当前已有的 vlog 文件，基于它们进行恢复
|
|
if (type == kLogFile)
|
|
logs.push_back(number);
|
|
// end
|
|
}
|
|
}
|
|
if (!expected.empty()) {
|
|
char buf[50];
|
|
std::snprintf(buf, sizeof(buf), "%d missing files; e.g.",
|
|
static_cast<int>(expected.size()));
|
|
return Status::Corruption(buf, TableFileName(dbname_, *(expected.begin())));
|
|
}
|
|
|
|
// Recover in the order in which the logs were generated
|
|
std::sort(logs.begin(), logs.end());
|
|
assert( logs.size() == 0 || logs[logs.size() - 1] >= versions_->ImmLogFileNumber() );
|
|
// for (size_t i = 0; i < logs.size(); i++) {
|
|
// s = RecoverLogFile(logs[i], (i == logs.size() - 1), save_manifest, edit,
|
|
// &max_sequence);
|
|
// if (!s.ok()) {
|
|
// return s;
|
|
// }
|
|
|
|
// // The previous incarnation may not have written any MANIFEST
|
|
// // records after allocating this log number. So we manually
|
|
// // update the file number allocation counter in VersionSet.
|
|
// versions_->MarkFileNumberUsed(logs[i]);
|
|
// }
|
|
//注释：逐个恢复日志的内容
|
|
bool found_sequence_pos = false;
|
|
for(int i = 0; i < logs.size(); ++i){
|
|
if( logs[i] < versions_->ImmLogFileNumber() ) {
|
|
continue;
|
|
}
|
|
Log(options_.info_log, "RecoverLogFile old log: %06llu \n", static_cast<unsigned long long>(logs[i]));
|
|
// 重做日志操作
|
|
s = RecoverLogFile(logs[i], (i == logs.size() - 1), save_manifest, edit,
|
|
&max_sequence, found_sequence_pos);
|
|
if (!s.ok()) {
|
|
return s;
|
|
}
|
|
}
|
|
versions_->MarkFileNumberUsed(max_number);
|
|
|
|
if (versions_->LastSequence() < max_sequence) {
|
|
versions_->SetLastSequence(max_sequence);
|
|
}
|
|
|
|
return Status::OK();
|
|
}
|
|
|
|
/* 恢复内存中的数据，将 VLog 中记录的操作读取出来，重新写入到 memtable */
|
|
Status DBImpl::RecoverLogFile(uint64_t log_number, bool last_log,
|
|
bool* save_manifest, VersionEdit* edit,
|
|
SequenceNumber* max_sequence,
|
|
bool& found_sequence_pos) {
|
|
struct LogReporter : public log::Reader::Reporter {
|
|
Env* env;
|
|
Logger* info_log;
|
|
const char* fname;
|
|
Status* status; // null if options_.paranoid_checks==false
|
|
void Corruption(size_t bytes, const Status& s) override {
|
|
Log(info_log, "%s%s: dropping %d bytes; %s",
|
|
(this->status == nullptr ? "(ignoring error) " : ""), fname,
|
|
static_cast<int>(bytes), s.ToString().c_str());
|
|
if (this->status != nullptr && this->status->ok()) *this->status = s;
|
|
}
|
|
};
|
|
|
|
mutex_.AssertHeld();
|
|
|
|
// Open the log file
|
|
std::string fname = LogFileName(dbname_, log_number);
|
|
SequentialFile* file;
|
|
Status status = env_->NewSequentialFile(fname, &file);
|
|
if (!status.ok()) {
|
|
MaybeIgnoreError(&status);
|
|
return status;
|
|
}
|
|
|
|
// Create the log reader.
|
|
LogReporter reporter;
|
|
reporter.env = env_;
|
|
reporter.info_log = options_.info_log;
|
|
reporter.fname = fname.c_str();
|
|
reporter.status = (options_.paranoid_checks ? &status : nullptr);
|
|
// We intentionally make log::Reader do checksumming even if
|
|
// paranoid_checks==false so that corruptions cause entire commits
|
|
// to be skipped instead of propagating bad information (like overly
|
|
// large sequence numbers).
|
|
log::Reader reader(file, &reporter, true /*checksum*/, 0 /*initial_offset*/);
|
|
Log(options_.info_log, "Recovering log #%llu",
|
|
(unsigned long long)log_number);
|
|
|
|
// Read all the records and add to a memtable
|
|
std::string scratch;
|
|
Slice record;
|
|
WriteBatch batch;
|
|
uint64_t record_offset = 0;
|
|
int compactions = 0;
|
|
MemTable* mem = nullptr;
|
|
//注释：设置 imm_last_sequence
|
|
uint64_t imm_last_sequence = versions_->ImmLastSequence();
|
|
while (reader.ReadRecord(&record, &scratch) && status.ok()) {
|
|
// if (record.size() < 12) {
|
|
if (record.size() < 20) {
|
|
reporter.Corruption(record.size(),
|
|
Status::Corruption("log record too small"));
|
|
continue;
|
|
}
|
|
// begin 如果 imm_last_sequence == 0 的话，
|
|
// 那么整个说明没有进行一次 imm 转 sst的情况，所有的 log 文件都需要进行回收
|
|
// 回收编号最大的 log 文件即可
|
|
if( !found_sequence_pos && imm_last_sequence != 0 ){
|
|
Slice tmp = record;
|
|
tmp.remove_prefix(8);
|
|
uint64_t seq = DecodeFixed64(tmp.data());
|
|
tmp.remove_prefix(8);
|
|
uint64_t kv_numbers = DecodeFixed32(tmp.data());
|
|
|
|
// 解析出来的 seq 不符合要求跳过。恢复时定位 seq 位置一定要大于等于 versions_->LastSequence()
|
|
if( ( seq + kv_numbers - 1 ) < imm_last_sequence ) {
|
|
record_offset += record.size();
|
|
continue;
|
|
}else if( ( seq + kv_numbers - 1 ) == imm_last_sequence ){
|
|
// open db 落盘过 sst，再一次打开 db
|
|
found_sequence_pos = true;
|
|
record_offset += record.size();
|
|
continue;
|
|
} else { // open db 之后没有落盘过 sst，然后关闭 db，第二次恢复的时候
|
|
found_sequence_pos = true;
|
|
}
|
|
}
|
|
// 去除头部信息 crc 和length
|
|
record.remove_prefix(log::vHeaderSize);
|
|
// end
|
|
WriteBatchInternal::SetContents(&batch, record);
|
|
|
|
if (mem == nullptr) {
|
|
mem = new MemTable(internal_comparator_);
|
|
mem->Ref();
|
|
}
|
|
// status = WriteBatchInternal::InsertInto(&batch, mem);
|
|
status = WriteBatchInternal::InsertInto(&batch, mem,log_number,record_offset + 4);
|
|
MaybeIgnoreError(&status);
|
|
if (!status.ok()) {
|
|
break;
|
|
}
|
|
const SequenceNumber last_seq = WriteBatchInternal::Sequence(&batch) +
|
|
WriteBatchInternal::Count(&batch) - 1;
|
|
if (last_seq > *max_sequence) {
|
|
*max_sequence = last_seq;
|
|
}
|
|
|
|
if (mem->ApproximateMemoryUsage() > options_.write_buffer_size) {
|
|
compactions++;
|
|
*save_manifest = true;
|
|
|
|
// 注释：mem 落盘修改 imm_last_sequence，版本恢复
|
|
versions_->SetImmLastSequence(mem->GetTailSequence());
|
|
versions_->SetImmLogFileNumber(log_number);
|
|
|
|
status = WriteLevel0Table(mem, edit, nullptr);
|
|
mem->Unref();
|
|
mem = nullptr;
|
|
if (!status.ok()) {
|
|
// Reflect errors immediately so that conditions like full
|
|
// file-systems cause the DB::Open() to fail.
|
|
break;
|
|
}
|
|
}
|
|
// 前面已经移除了一个头部了，所以偏移位置要个头部
|
|
record_offset += record.size() + log::vHeaderSize ;
|
|
}
|
|
|
|
delete file;
|
|
|
|
// See if we should keep reusing the last log file.
|
|
if (status.ok() && options_.reuse_logs && last_log && compactions == 0) {
|
|
assert(logfile_ == nullptr);
|
|
assert(vlog_ == nullptr);
|
|
assert(mem_ == nullptr);
|
|
uint64_t lfile_size;
|
|
if (env_->GetFileSize(fname, &lfile_size).ok() &&
|
|
env_->NewAppendableFile(fname, &logfile_).ok()) {
|
|
Log(options_.info_log, "Reusing old log %s \n", fname.c_str());
|
|
vlog_ = new log::VlogWriter(logfile_, lfile_size);
|
|
logfile_number_ = log_number;
|
|
if (mem != nullptr) {
|
|
mem_ = mem;
|
|
mem = nullptr;
|
|
} else {
|
|
// mem can be nullptr if lognum exists but was empty.
|
|
mem_ = new MemTable(internal_comparator_);
|
|
mem_->Ref();
|
|
}
|
|
}
|
|
}
|
|
|
|
if (mem != nullptr) {
|
|
// mem did not get reused; compact it.
|
|
if (status.ok()) {
|
|
|
|
//注释： mem 落盘修改 imm_last_sequence，版本恢复
|
|
versions_->SetImmLastSequence(mem->GetTailSequence());
|
|
versions_->SetImmLogFileNumber(log_number);
|
|
*save_manifest = true;
|
|
status = WriteLevel0Table(mem, edit, nullptr);
|
|
}
|
|
mem->Unref();
|
|
}
|
|
|
|
return status;
|
|
}
|
|
|
|
Status DBImpl::WriteLevel0Table(MemTable* mem, VersionEdit* edit,
|
|
Version* base) {
|
|
mutex_.AssertHeld();
|
|
const uint64_t start_micros = env_->NowMicros();
|
|
FileMetaData meta;
|
|
meta.number = versions_->NewFileNumber();
|
|
pending_outputs_.insert(meta.number);
|
|
Iterator* iter = mem->NewIterator();
|
|
Log(options_.info_log, "Level-0 table #%llu: started",
|
|
(unsigned long long)meta.number);
|
|
|
|
Status s;
|
|
{
|
|
mutex_.Unlock();
|
|
s = BuildTable(dbname_, env_, options_, table_cache_, iter, &meta);
|
|
mutex_.Lock();
|
|
}
|
|
|
|
Log(options_.info_log, "Level-0 table #%llu: %lld bytes %s",
|
|
(unsigned long long)meta.number, (unsigned long long)meta.file_size,
|
|
s.ToString().c_str());
|
|
delete iter;
|
|
pending_outputs_.erase(meta.number);
|
|
|
|
// Note that if file_size is zero, the file has been deleted and
|
|
// should not be added to the manifest.
|
|
int level = 0;
|
|
if (s.ok() && meta.file_size > 0) {
|
|
const Slice min_user_key = meta.smallest.user_key();
|
|
const Slice max_user_key = meta.largest.user_key();
|
|
if (base != nullptr) {
|
|
level = base->PickLevelForMemTableOutput(min_user_key, max_user_key);
|
|
}
|
|
edit->AddFile(level, meta.number, meta.file_size, meta.smallest,
|
|
meta.largest);
|
|
}
|
|
|
|
CompactionStats stats;
|
|
stats.micros = env_->NowMicros() - start_micros;
|
|
stats.bytes_written = meta.file_size;
|
|
stats_[level].Add(stats);
|
|
return s;
|
|
}
|
|
|
|
void DBImpl::CompactMemTable() {
|
|
mutex_.AssertHeld();
|
|
assert(imm_ != nullptr);
|
|
|
|
// Save the contents of the memtable as a new Table
|
|
VersionEdit edit;
|
|
Version* base = versions_->current();
|
|
base->Ref();
|
|
Status s = WriteLevel0Table(imm_, &edit, base);
|
|
base->Unref();
|
|
|
|
if (s.ok() && shutting_down_.load(std::memory_order_acquire)) {
|
|
s = Status::IOError("Deleting DB during memtable compaction");
|
|
}
|
|
|
|
// Replace immutable memtable with the generated Table
|
|
if (s.ok()) {
|
|
edit.SetPrevLogNumber(0);
|
|
edit.SetLogNumber(logfile_number_); // Earlier logs no longer needed
|
|
// s = versions_->LogAndApply(&edit, &mutex_);
|
|
//注释： 构建新版本，并将其加入到 version_当中
|
|
versions_->StartImmLastSequence(true);
|
|
versions_->SetImmLastSequence(imm_->GetTailSequence());
|
|
versions_->SetImmLogFileNumber(imm_->GetLogFileNumber());
|
|
s = versions_->LogAndApply(&edit, &mutex_);
|
|
versions_->StartImmLastSequence(false);
|
|
}
|
|
|
|
if (s.ok()) {
|
|
// Commit to the new state
|
|
imm_->Unref();
|
|
imm_ = nullptr;
|
|
has_imm_.store(false, std::memory_order_release);
|
|
RemoveObsoleteFiles();
|
|
} else {
|
|
RecordBackgroundError(s);
|
|
}
|
|
}
|
|
|
|
void DBImpl::CompactRange(const Slice* begin, const Slice* end) {
|
|
int max_level_with_files = 1;
|
|
{
|
|
MutexLock l(&mutex_);
|
|
Version* base = versions_->current();
|
|
for (int level = 1; level < config::kNumLevels; level++) {
|
|
if (base->OverlapInLevel(level, begin, end)) {
|
|
max_level_with_files = level;
|
|
}
|
|
}
|
|
}
|
|
TEST_CompactMemTable(); // TODO(sanjay): Skip if memtable does not overlap
|
|
for (int level = 0; level < max_level_with_files; level++) {
|
|
TEST_CompactRange(level, begin, end);
|
|
}
|
|
}
|
|
|
|
void DBImpl::TEST_CompactRange(int level, const Slice* begin,
|
|
const Slice* end) {
|
|
assert(level >= 0);
|
|
assert(level + 1 < config::kNumLevels);
|
|
|
|
InternalKey begin_storage, end_storage;
|
|
|
|
ManualCompaction manual;
|
|
manual.level = level;
|
|
manual.done = false;
|
|
if (begin == nullptr) {
|
|
manual.begin = nullptr;
|
|
} else {
|
|
begin_storage = InternalKey(*begin, kMaxSequenceNumber, kValueTypeForSeek);
|
|
manual.begin = &begin_storage;
|
|
}
|
|
if (end == nullptr) {
|
|
manual.end = nullptr;
|
|
} else {
|
|
end_storage = InternalKey(*end, 0, static_cast<ValueType>(0));
|
|
manual.end = &end_storage;
|
|
}
|
|
|
|
MutexLock l(&mutex_);
|
|
while (!manual.done && !shutting_down_.load(std::memory_order_acquire) &&
|
|
bg_error_.ok()) {
|
|
if (manual_compaction_ == nullptr) { // Idle
|
|
manual_compaction_ = &manual;
|
|
MaybeScheduleCompaction();
|
|
} else { // Running either my compaction or another compaction.
|
|
background_work_finished_signal_.Wait();
|
|
}
|
|
}
|
|
// Finish current background compaction in the case where
|
|
// `background_work_finished_signal_` was signalled due to an error.
|
|
while (background_compaction_scheduled_) {
|
|
background_work_finished_signal_.Wait();
|
|
}
|
|
if (manual_compaction_ == &manual) {
|
|
// Cancel my manual compaction since we aborted early for some reason.
|
|
manual_compaction_ = nullptr;
|
|
}
|
|
}
|
|
|
|
Status DBImpl::TEST_CompactMemTable() {
|
|
// nullptr batch means just wait for earlier writes to be done
|
|
Status s = Write(WriteOptions(), nullptr);
|
|
if (s.ok()) {
|
|
// Wait until the compaction completes
|
|
MutexLock l(&mutex_);
|
|
while (imm_ != nullptr && bg_error_.ok()) {
|
|
background_work_finished_signal_.Wait();
|
|
}
|
|
if (imm_ != nullptr) {
|
|
s = bg_error_;
|
|
}
|
|
}
|
|
return s;
|
|
}
|
|
|
|
void DBImpl::RecordBackgroundError(const Status& s) {
|
|
mutex_.AssertHeld();
|
|
if (bg_error_.ok()) {
|
|
bg_error_ = s;
|
|
background_work_finished_signal_.SignalAll();
|
|
// 注释：唤醒后台 GC 线程
|
|
garbage_collection_work_signal_.SignalAll();
|
|
}
|
|
}
|
|
|
|
void DBImpl::MaybeScheduleCompaction() {
|
|
mutex_.AssertHeld();
|
|
if (background_compaction_scheduled_) {
|
|
// Already scheduled
|
|
} else if (shutting_down_.load(std::memory_order_acquire)) {
|
|
// DB is being deleted; no more background compactions
|
|
} else if (!bg_error_.ok()) {
|
|
// Already got an error; no more changes
|
|
} else if (imm_ == nullptr && manual_compaction_ == nullptr &&
|
|
!versions_->NeedsCompaction()) {
|
|
// No work to be done
|
|
} else {
|
|
background_compaction_scheduled_ = true;
|
|
env_->Schedule(&DBImpl::BGWork, this);
|
|
}
|
|
}
|
|
|
|
// 注释：获取所有 VLogs
|
|
Status DBImpl::GetAllValueLog(std::string dir, std::vector<uint64_t>& logs) {
|
|
logs.clear();
|
|
std::vector<std::string> filenames;
|
|
// 获取 VLogs 列表
|
|
Status s = env_->GetChildren(dir, &filenames);
|
|
if (!s.ok()) {
|
|
return s;
|
|
}
|
|
uint64_t number;
|
|
FileType type;
|
|
for (size_t i = 0; i < filenames.size(); i++) {
|
|
if (ParseFileName(filenames[i], &number, &type)) {
|
|
// 获取所有 .log 文件
|
|
if (type == kLogFile)
|
|
logs.push_back(number);
|
|
}
|
|
}
|
|
return s;
|
|
}
|
|
|
|
// 注释：手动进行离线回收
|
|
Status DBImpl::OutLineGarbageCollection(){
|
|
MutexLock l(&mutex_);
|
|
Status s;
|
|
// map_file_info_ 非空，则根据它进行回收，否则进行所有 VLog 的回收
|
|
if (!garbage_collection_management_->EmptyMap()) {
|
|
garbage_collection_management_->CollectionMap();
|
|
uint64_t last_sequence = versions_->LastSequence();
|
|
garbage_collection_management_->ConvertQueue(last_sequence);
|
|
versions_->SetLastSequence(last_sequence);
|
|
MaybeScheduleGarbageCollection();
|
|
return Status();
|
|
}
|
|
return s;
|
|
}
|
|
|
|
// 注释：在线 GC，读取并回收一个 vlog 文件，
|
|
// next_sequence 指的是第一个没有用到的 sequence（由于是在线 GC ，所以需要提前指定）
|
|
Status DBImpl::CollectionValueLog(uint64_t fid, uint64_t& next_sequence) {
|
|
|
|
struct LogReporter : public log::VlogReader::Reporter {
|
|
Status* status;
|
|
void Corruption(size_t bytes, const Status& s) override {
|
|
if (this->status->ok()) *this->status = s;
|
|
}
|
|
};
|
|
LogReporter report;
|
|
std::string logName = LogFileName(dbname_, fid);
|
|
SequentialFile* lfile;
|
|
Status status = env_->NewSequentialFile(logName, &lfile);
|
|
if (!status.ok()) {
|
|
Log(options_.info_log, "Garbage Collection Open file error: %s", status.ToString().c_str());
|
|
return status;
|
|
}
|
|
log::VlogReader reader(lfile, &report);
|
|
|
|
Slice record;
|
|
std::string scratch;
|
|
// record_offset 是每条 record 相对 VLog head 的偏移
|
|
uint64_t record_offset = 0;
|
|
uint64_t size_offset = 0;
|
|
WriteOptions opt(options_.background_garbage_collection_separate_);
|
|
WriteBatch batch(opt.separate_threshold);
|
|
batch.setGarbageCollection(true);
|
|
WriteBatchInternal::SetSequence(&batch, next_sequence);
|
|
|
|
while (reader.ReadRecord(&record, &scratch)) {
|
|
const char* head_record_ptr = record.data();
|
|
record.remove_prefix(log::vHeaderSize + log::wHeaderSize);
|
|
|
|
while (record.size() > 0) {
|
|
const char* head_kv_ptr = record.data();
|
|
uint64_t kv_offset = record_offset + head_kv_ptr - head_record_ptr;
|
|
ValueType type = static_cast<ValueType>(record[0]);
|
|
record.remove_prefix(1);
|
|
Slice key;
|
|
Slice value;
|
|
std::string get_value;
|
|
GetLengthPrefixedSlice(&record,&key);
|
|
if (type != kTypeDeletion) {
|
|
GetLengthPrefixedSlice(&record,&value);
|
|
}
|
|
if (type != kTypeSeparation) {
|
|
continue;
|
|
}
|
|
|
|
status = this->GetLsm(key,&get_value);
|
|
// 1. 从 LSM-tree 中找不到 key，说明这个 key 被删除了，vlog中要丢弃
|
|
// 2. 找到了 key，但是最新的 kv 对不是 KV 分离的情况，也丢弃
|
|
if (status.IsNotFound() || !status.IsSeparated()) {
|
|
continue;
|
|
}
|
|
if (!status.ok()) {
|
|
std::cout<<"read the file error "<<std::endl;
|
|
return status;
|
|
}
|
|
|
|
// 判断是否要丢弃旧值
|
|
Slice get_slice(get_value);
|
|
uint64_t lsm_fid;
|
|
uint64_t lsm_offset;
|
|
|
|
GetVarint64(&get_slice,&lsm_fid);
|
|
GetVarint64(&get_slice,&lsm_offset);
|
|
|
|
if (fid == lsm_fid && lsm_offset == kv_offset) {
|
|
batch.Put(key, value);
|
|
++next_sequence;
|
|
if (kv_offset - size_offset > config::gcWriteBatchSize) {
|
|
Write(opt, &batch);
|
|
batch.Clear();
|
|
batch.setGarbageCollection(true);
|
|
WriteBatchInternal::SetSequence(&batch, next_sequence);
|
|
uint64_t kv_size;
|
|
GetVarint64(&get_slice,&kv_size);
|
|
size_offset = kv_offset + kv_size;
|
|
}
|
|
}
|
|
}
|
|
record_offset += record.data() - head_record_ptr;
|
|
}
|
|
|
|
Write(opt, &batch);
|
|
status = env_->RemoveFile(logName);
|
|
if (status.ok()) {
|
|
garbage_collection_management_->RemoveFileFromMap(fid);
|
|
}
|
|
return status;
|
|
}
|
|
|
|
// 注释：后台 GC 任务
|
|
void DBImpl::BackGroundGarbageCollection(){
|
|
uint64_t fid;
|
|
uint64_t last_sequence;
|
|
while( true){
|
|
Log(options_.info_log, "garbage collection file number: %lu", fid);
|
|
if( !garbage_collection_management_->GetGarbageCollectionQueue(fid,last_sequence) ){
|
|
return;
|
|
}
|
|
// 在线的 gc 回收的 sequence 是要提前就分配好的。
|
|
CollectionValueLog(fid,last_sequence);
|
|
}
|
|
}
|
|
|
|
// 注释：可能调度后台线程进行 GC
|
|
void DBImpl::MaybeScheduleGarbageCollection() {
|
|
mutex_.AssertHeld();
|
|
if (background_GarbageCollection_scheduled_) {
|
|
// Already scheduled
|
|
} else if (shutting_down_.load(std::memory_order_acquire)) {
|
|
// DB is being deleted; no more background compactions
|
|
} else if (!bg_error_.ok()) {
|
|
// Already got an error; no more changes
|
|
} else {
|
|
// 设置调度变量,通过 detach 线程调度; detach 线程即使主线程退出,依然可以正常执行完成
|
|
background_GarbageCollection_scheduled_ = true;
|
|
env_->ScheduleForGarbageCollection(&DBImpl::GarbageCollectionBGWork, this);
|
|
}
|
|
}
|
|
|
|
// 注释：后台 gc 线程中执行的任务
|
|
void DBImpl::GarbageCollectionBGWork(void* db) {
|
|
reinterpret_cast<DBImpl*>(db)->GarbageCollectionBackgroundCall();
|
|
}
|
|
|
|
// 注释：后台 gc 线程执行
|
|
void DBImpl::GarbageCollectionBackgroundCall() {
|
|
assert(background_GarbageCollection_scheduled_);
|
|
if (shutting_down_.load(std::memory_order_acquire)) {
|
|
// No more background work when shutting down.
|
|
} else if (!bg_error_.ok()) {
|
|
// No more background work after a background error.
|
|
} else {
|
|
BackGroundGarbageCollection();
|
|
}
|
|
|
|
background_GarbageCollection_scheduled_ = false;
|
|
// 再调用 MaybeScheduleGarbageCollection 检查是否需要再次调度
|
|
MaybeScheduleGarbageCollection();
|
|
garbage_collection_work_signal_.SignalAll();
|
|
}
|
|
// end
|
|
|
|
void DBImpl::BGWork(void* db) {
|
|
reinterpret_cast<DBImpl*>(db)->BackgroundCall();
|
|
}
|
|
|
|
void DBImpl::BackgroundCall() {
|
|
MutexLock l(&mutex_);
|
|
assert(background_compaction_scheduled_);
|
|
if (shutting_down_.load(std::memory_order_acquire)) {
|
|
// No more background work when shutting down.
|
|
} else if (!bg_error_.ok()) {
|
|
// No more background work after a background error.
|
|
} else {
|
|
BackgroundCompaction();
|
|
}
|
|
|
|
background_compaction_scheduled_ = false;
|
|
|
|
// Previous compaction may have produced too many files in a level,
|
|
// so reschedule another compaction if needed.
|
|
MaybeScheduleCompaction();
|
|
background_work_finished_signal_.SignalAll();
|
|
}
|
|
|
|
void DBImpl::BackgroundCompaction() {
|
|
mutex_.AssertHeld();
|
|
|
|
if (imm_ != nullptr) {
|
|
CompactMemTable();
|
|
return;
|
|
}
|
|
|
|
Compaction* c;
|
|
bool is_manual = (manual_compaction_ != nullptr);
|
|
InternalKey manual_end;
|
|
if (is_manual) {
|
|
ManualCompaction* m = manual_compaction_;
|
|
c = versions_->CompactRange(m->level, m->begin, m->end);
|
|
m->done = (c == nullptr);
|
|
if (c != nullptr) {
|
|
manual_end = c->input(0, c->num_input_files(0) - 1)->largest;
|
|
}
|
|
Log(options_.info_log,
|
|
"Manual compaction at level-%d from %s .. %s; will stop at %s\n",
|
|
m->level, (m->begin ? m->begin->DebugString().c_str() : "(begin)"),
|
|
(m->end ? m->end->DebugString().c_str() : "(end)"),
|
|
(m->done ? "(end)" : manual_end.DebugString().c_str()));
|
|
} else {
|
|
c = versions_->PickCompaction();
|
|
}
|
|
|
|
Status status;
|
|
if (c == nullptr) {
|
|
// Nothing to do
|
|
} else if (!is_manual && c->IsTrivialMove()) {
|
|
// Move file to next level
|
|
assert(c->num_input_files(0) == 1);
|
|
FileMetaData* f = c->input(0, 0);
|
|
c->edit()->RemoveFile(c->level(), f->number);
|
|
c->edit()->AddFile(c->level() + 1, f->number, f->file_size, f->smallest,
|
|
f->largest);
|
|
status = versions_->LogAndApply(c->edit(), &mutex_);
|
|
if (!status.ok()) {
|
|
RecordBackgroundError(status);
|
|
}
|
|
VersionSet::LevelSummaryStorage tmp;
|
|
Log(options_.info_log, "Moved #%lld to level-%d %lld bytes %s: %s\n",
|
|
static_cast<unsigned long long>(f->number), c->level() + 1,
|
|
static_cast<unsigned long long>(f->file_size),
|
|
status.ToString().c_str(), versions_->LevelSummary(&tmp));
|
|
} else {
|
|
CompactionState* compact = new CompactionState(c);
|
|
status = DoCompactionWork(compact);
|
|
if (!status.ok()) {
|
|
RecordBackgroundError(status);
|
|
}
|
|
// begin 注释： compact 后需要考虑是否将 vlog 文件进行 gc 回收，
|
|
// 如果需要则将其加入到 GC 任务队列中
|
|
// 不进行后台的 gc 回收，那么也不用更新待分配 sequence 的 vlog
|
|
if(!finish_back_garbage_collection_){
|
|
garbage_collection_management_->UpdateQueue(versions_->ImmLogFileNumber() );
|
|
}
|
|
|
|
CleanupCompaction(compact);
|
|
c->ReleaseInputs();
|
|
RemoveObsoleteFiles();
|
|
}
|
|
delete c;
|
|
|
|
if (status.ok()) {
|
|
// Done
|
|
} else if (shutting_down_.load(std::memory_order_acquire)) {
|
|
// Ignore compaction errors found during shutting down
|
|
} else {
|
|
Log(options_.info_log, "Compaction error: %s", status.ToString().c_str());
|
|
}
|
|
|
|
if (is_manual) {
|
|
ManualCompaction* m = manual_compaction_;
|
|
if (!status.ok()) {
|
|
m->done = true;
|
|
}
|
|
if (!m->done) {
|
|
// We only compacted part of the requested range. Update *m
|
|
// to the range that is left to be compacted.
|
|
m->tmp_storage = manual_end;
|
|
m->begin = &m->tmp_storage;
|
|
}
|
|
manual_compaction_ = nullptr;
|
|
}
|
|
}
|
|
|
|
void DBImpl::CleanupCompaction(CompactionState* compact) {
|
|
mutex_.AssertHeld();
|
|
if (compact->builder != nullptr) {
|
|
// May happen if we get a shutdown call in the middle of compaction
|
|
compact->builder->Abandon();
|
|
delete compact->builder;
|
|
} else {
|
|
assert(compact->outfile == nullptr);
|
|
}
|
|
delete compact->outfile;
|
|
for (size_t i = 0; i < compact->outputs.size(); i++) {
|
|
const CompactionState::Output& out = compact->outputs[i];
|
|
pending_outputs_.erase(out.number);
|
|
}
|
|
delete compact;
|
|
}
|
|
|
|
Status DBImpl::OpenCompactionOutputFile(CompactionState* compact) {
|
|
assert(compact != nullptr);
|
|
assert(compact->builder == nullptr);
|
|
uint64_t file_number;
|
|
{
|
|
mutex_.Lock();
|
|
file_number = versions_->NewFileNumber();
|
|
pending_outputs_.insert(file_number);
|
|
CompactionState::Output out;
|
|
out.number = file_number;
|
|
out.smallest.Clear();
|
|
out.largest.Clear();
|
|
compact->outputs.push_back(out);
|
|
mutex_.Unlock();
|
|
}
|
|
|
|
// Make the output file
|
|
std::string fname = TableFileName(dbname_, file_number);
|
|
Status s = env_->NewWritableFile(fname, &compact->outfile);
|
|
if (s.ok()) {
|
|
compact->builder = new TableBuilder(options_, compact->outfile);
|
|
}
|
|
return s;
|
|
}
|
|
|
|
Status DBImpl::FinishCompactionOutputFile(CompactionState* compact,
|
|
Iterator* input) {
|
|
assert(compact != nullptr);
|
|
assert(compact->outfile != nullptr);
|
|
assert(compact->builder != nullptr);
|
|
|
|
const uint64_t output_number = compact->current_output()->number;
|
|
assert(output_number != 0);
|
|
|
|
// Check for iterator errors
|
|
Status s = input->status();
|
|
const uint64_t current_entries = compact->builder->NumEntries();
|
|
if (s.ok()) {
|
|
s = compact->builder->Finish();
|
|
} else {
|
|
compact->builder->Abandon();
|
|
}
|
|
const uint64_t current_bytes = compact->builder->FileSize();
|
|
compact->current_output()->file_size = current_bytes;
|
|
compact->total_bytes += current_bytes;
|
|
delete compact->builder;
|
|
compact->builder = nullptr;
|
|
|
|
// Finish and check for file errors
|
|
if (s.ok()) {
|
|
s = compact->outfile->Sync();
|
|
}
|
|
if (s.ok()) {
|
|
s = compact->outfile->Close();
|
|
}
|
|
delete compact->outfile;
|
|
compact->outfile = nullptr;
|
|
|
|
if (s.ok() && current_entries > 0) {
|
|
// Verify that the table is usable
|
|
Iterator* iter =
|
|
table_cache_->NewIterator(ReadOptions(), output_number, current_bytes);
|
|
s = iter->status();
|
|
delete iter;
|
|
if (s.ok()) {
|
|
Log(options_.info_log, "Generated table #%llu@%d: %lld keys, %lld bytes",
|
|
(unsigned long long)output_number, compact->compaction->level(),
|
|
(unsigned long long)current_entries,
|
|
(unsigned long long)current_bytes);
|
|
}
|
|
}
|
|
return s;
|
|
}
|
|
|
|
Status DBImpl::InstallCompactionResults(CompactionState* compact) {
|
|
mutex_.AssertHeld();
|
|
Log(options_.info_log, "Compacted %d@%d + %d@%d files => %lld bytes",
|
|
compact->compaction->num_input_files(0), compact->compaction->level(),
|
|
compact->compaction->num_input_files(1), compact->compaction->level() + 1,
|
|
static_cast<long long>(compact->total_bytes));
|
|
|
|
// Add compaction outputs
|
|
compact->compaction->AddInputDeletions(compact->compaction->edit());
|
|
const int level = compact->compaction->level();
|
|
for (size_t i = 0; i < compact->outputs.size(); i++) {
|
|
const CompactionState::Output& out = compact->outputs[i];
|
|
compact->compaction->edit()->AddFile(level + 1, out.number, out.file_size,
|
|
out.smallest, out.largest);
|
|
}
|
|
return versions_->LogAndApply(compact->compaction->edit(), &mutex_);
|
|
}
|
|
|
|
Status DBImpl::DoCompactionWork(CompactionState* compact) {
|
|
const uint64_t start_micros = env_->NowMicros();
|
|
int64_t imm_micros = 0; // Micros spent doing imm_ compactions
|
|
|
|
Log(options_.info_log, "Compacting %d@%d + %d@%d files",
|
|
compact->compaction->num_input_files(0), compact->compaction->level(),
|
|
compact->compaction->num_input_files(1),
|
|
compact->compaction->level() + 1);
|
|
|
|
assert(versions_->NumLevelFiles(compact->compaction->level()) > 0);
|
|
assert(compact->builder == nullptr);
|
|
assert(compact->outfile == nullptr);
|
|
if (snapshots_.empty()) {
|
|
compact->smallest_snapshot = versions_->LastSequence();
|
|
} else {
|
|
compact->smallest_snapshot = snapshots_.oldest()->sequence_number();
|
|
}
|
|
|
|
Iterator* input = versions_->MakeInputIterator(compact->compaction);
|
|
|
|
// Release mutex while we're actually doing the compaction work
|
|
mutex_.Unlock();
|
|
|
|
input->SeekToFirst();
|
|
Status status;
|
|
ParsedInternalKey ikey;
|
|
std::string current_user_key;
|
|
bool has_current_user_key = false;
|
|
SequenceNumber last_sequence_for_key = kMaxSequenceNumber;
|
|
while (input->Valid() && !shutting_down_.load(std::memory_order_acquire)) {
|
|
// Prioritize immutable compaction work
|
|
if (has_imm_.load(std::memory_order_relaxed)) {
|
|
const uint64_t imm_start = env_->NowMicros();
|
|
mutex_.Lock();
|
|
if (imm_ != nullptr) {
|
|
CompactMemTable();
|
|
// Wake up MakeRoomForWrite() if necessary.
|
|
background_work_finished_signal_.SignalAll();
|
|
}
|
|
mutex_.Unlock();
|
|
imm_micros += (env_->NowMicros() - imm_start);
|
|
}
|
|
|
|
Slice key = input->key();
|
|
if (compact->compaction->ShouldStopBefore(key) &&
|
|
compact->builder != nullptr) {
|
|
status = FinishCompactionOutputFile(compact, input);
|
|
if (!status.ok()) {
|
|
break;
|
|
}
|
|
}
|
|
|
|
// Handle key/value, add to state, etc.
|
|
bool drop = false;
|
|
if (!ParseInternalKey(key, &ikey)) {
|
|
// Do not hide error keys
|
|
current_user_key.clear();
|
|
has_current_user_key = false;
|
|
last_sequence_for_key = kMaxSequenceNumber;
|
|
} else {
|
|
if (!has_current_user_key ||
|
|
user_comparator()->Compare(ikey.user_key, Slice(current_user_key)) !=
|
|
0) {
|
|
// First occurrence of this user key
|
|
current_user_key.assign(ikey.user_key.data(), ikey.user_key.size());
|
|
has_current_user_key = true;
|
|
last_sequence_for_key = kMaxSequenceNumber;
|
|
}
|
|
|
|
if (last_sequence_for_key <= compact->smallest_snapshot) {
|
|
// Hidden by an newer entry for same user key
|
|
drop = true; // (A)
|
|
} else if (ikey.type == kTypeDeletion &&
|
|
ikey.sequence <= compact->smallest_snapshot &&
|
|
compact->compaction->IsBaseLevelForKey(ikey.user_key)) {
|
|
// For this user key:
|
|
// (1) there is no data in higher levels
|
|
// (2) data in lower levels will have larger sequence numbers
|
|
// (3) data in layers that are being compacted here and have
|
|
// smaller sequence numbers will be dropped in the next
|
|
// few iterations of this loop (by rule (A) above).
|
|
// Therefore this deletion marker is obsolete and can be dropped.
|
|
drop = true;
|
|
}
|
|
|
|
last_sequence_for_key = ikey.sequence;
|
|
}
|
|
#if 0
|
|
Log(options_.info_log,
|
|
" Compact: %s, seq %d, type: %d %d, drop: %d, is_base: %d, "
|
|
"%d smallest_snapshot: %d",
|
|
ikey.user_key.ToString().c_str(),
|
|
(int)ikey.sequence, ikey.type, kTypeValue, drop,
|
|
compact->compaction->IsBaseLevelForKey(ikey.user_key),
|
|
(int)last_sequence_for_key, (int)compact->smallest_snapshot);
|
|
#endif
|
|
|
|
if (!drop) {
|
|
// Open output file if necessary
|
|
if (compact->builder == nullptr) {
|
|
status = OpenCompactionOutputFile(compact);
|
|
if (!status.ok()) {
|
|
break;
|
|
}
|
|
}
|
|
if (compact->builder->NumEntries() == 0) {
|
|
compact->current_output()->smallest.DecodeFrom(key);
|
|
}
|
|
compact->current_output()->largest.DecodeFrom(key);
|
|
compact->builder->Add(key, input->value());
|
|
|
|
// Close output file if it is big enough
|
|
if (compact->builder->FileSize() >=
|
|
compact->compaction->MaxOutputFileSize()) {
|
|
status = FinishCompactionOutputFile(compact, input);
|
|
if (!status.ok()) {
|
|
break;
|
|
}
|
|
}
|
|
} else {
|
|
// begin 注释：drop 掉 LSM-tree 中的 kv 数值对了，
|
|
// 对属于 KV 分离的 kv 数值对进行 GC
|
|
Slice drop_value = input->value();
|
|
if( ikey.type == kTypeSeparation ){
|
|
uint64_t fid = 0;
|
|
uint64_t offset = 0;
|
|
uint64_t size = 0;
|
|
GetVarint64(&drop_value,&fid);
|
|
GetVarint64(&drop_value,&offset);
|
|
GetVarint64(&drop_value,&size);
|
|
mutex_.Lock();
|
|
garbage_collection_management_->UpdateMap(fid,size);
|
|
mutex_.Unlock();
|
|
}
|
|
} // end
|
|
|
|
input->Next();
|
|
}
|
|
|
|
if (status.ok() && shutting_down_.load(std::memory_order_acquire)) {
|
|
status = Status::IOError("Deleting DB during compaction");
|
|
}
|
|
if (status.ok() && compact->builder != nullptr) {
|
|
status = FinishCompactionOutputFile(compact, input);
|
|
}
|
|
if (status.ok()) {
|
|
status = input->status();
|
|
}
|
|
delete input;
|
|
input = nullptr;
|
|
|
|
CompactionStats stats;
|
|
stats.micros = env_->NowMicros() - start_micros - imm_micros;
|
|
for (int which = 0; which < 2; which++) {
|
|
for (int i = 0; i < compact->compaction->num_input_files(which); i++) {
|
|
stats.bytes_read += compact->compaction->input(which, i)->file_size;
|
|
}
|
|
}
|
|
for (size_t i = 0; i < compact->outputs.size(); i++) {
|
|
stats.bytes_written += compact->outputs[i].file_size;
|
|
}
|
|
|
|
mutex_.Lock();
|
|
stats_[compact->compaction->level() + 1].Add(stats);
|
|
|
|
if (status.ok()) {
|
|
status = InstallCompactionResults(compact);
|
|
}
|
|
if (!status.ok()) {
|
|
RecordBackgroundError(status);
|
|
}
|
|
VersionSet::LevelSummaryStorage tmp;
|
|
Log(options_.info_log, "compacted to: %s", versions_->LevelSummary(&tmp));
|
|
return status;
|
|
}
|
|
|
|
namespace {
|
|
|
|
struct IterState {
|
|
port::Mutex* const mu;
|
|
Version* const version GUARDED_BY(mu);
|
|
MemTable* const mem GUARDED_BY(mu);
|
|
MemTable* const imm GUARDED_BY(mu);
|
|
|
|
IterState(port::Mutex* mutex, MemTable* mem, MemTable* imm, Version* version)
|
|
: mu(mutex), version(version), mem(mem), imm(imm) {}
|
|
};
|
|
|
|
static void CleanupIteratorState(void* arg1, void* arg2) {
|
|
IterState* state = reinterpret_cast<IterState*>(arg1);
|
|
state->mu->Lock();
|
|
state->mem->Unref();
|
|
if (state->imm != nullptr) state->imm->Unref();
|
|
state->version->Unref();
|
|
state->mu->Unlock();
|
|
delete state;
|
|
}
|
|
|
|
} // anonymous namespace
|
|
|
|
Iterator* DBImpl::NewInternalIterator(const ReadOptions& options,
|
|
SequenceNumber* latest_snapshot,
|
|
uint32_t* seed) {
|
|
mutex_.Lock();
|
|
*latest_snapshot = versions_->LastSequence();
|
|
|
|
// Collect together all needed child iterators
|
|
std::vector<Iterator*> list;
|
|
list.push_back(mem_->NewIterator());
|
|
mem_->Ref();
|
|
if (imm_ != nullptr) {
|
|
list.push_back(imm_->NewIterator());
|
|
imm_->Ref();
|
|
}
|
|
versions_->current()->AddIterators(options, &list);
|
|
Iterator* internal_iter =
|
|
NewMergingIterator(&internal_comparator_, &list[0], list.size());
|
|
versions_->current()->Ref();
|
|
|
|
IterState* cleanup = new IterState(&mutex_, mem_, imm_, versions_->current());
|
|
internal_iter->RegisterCleanup(CleanupIteratorState, cleanup, nullptr);
|
|
|
|
*seed = ++seed_;
|
|
mutex_.Unlock();
|
|
return internal_iter;
|
|
}
|
|
|
|
Iterator* DBImpl::TEST_NewInternalIterator() {
|
|
SequenceNumber ignored;
|
|
uint32_t ignored_seed;
|
|
return NewInternalIterator(ReadOptions(), &ignored, &ignored_seed);
|
|
}
|
|
|
|
int64_t DBImpl::TEST_MaxNextLevelOverlappingBytes() {
|
|
MutexLock l(&mutex_);
|
|
return versions_->MaxNextLevelOverlappingBytes();
|
|
}
|
|
|
|
/* 从 VlogReader 读取的 VLog kv 数据对中解析出 value */
|
|
bool DBImpl::ParseVlogValue(Slice key_value, Slice key,
|
|
std::string& value, uint64_t val_size) {
|
|
Slice k_v = key_value;
|
|
if (k_v[0] != kTypeSeparation) return false;
|
|
k_v.remove_prefix(1);
|
|
|
|
Slice vlog_key;
|
|
Slice vlog_value;
|
|
if (GetLengthPrefixedSlice(&k_v, &vlog_key)
|
|
&& vlog_key == key
|
|
&& GetLengthPrefixedSlice(&k_v, &vlog_value)
|
|
&& vlog_value.size() == val_size) {
|
|
value = vlog_value.ToString();
|
|
return true;
|
|
} else {
|
|
return false;
|
|
}
|
|
}
|
|
|
|
Status DBImpl::GetLsm(const Slice& key, std::string* value) {
|
|
MutexLock l(&mutex_);
|
|
ReadOptions options;
|
|
MemTable* mem = mem_;
|
|
MemTable* imm = imm_;
|
|
Version* current = versions_->current();
|
|
if( !this->snapshots_.empty() ){
|
|
options.snapshot = this->snapshots_.oldest();
|
|
}
|
|
SequenceNumber snapshot;
|
|
if (options.snapshot != nullptr) {
|
|
snapshot = static_cast<const SnapshotImpl*>(options.snapshot)->sequence_number();
|
|
} else {
|
|
snapshot = versions_->LastSequence();
|
|
}
|
|
Status s;
|
|
mem->Ref();
|
|
// imm 不一定存在，但是 mem 是一定存在的。
|
|
if (imm != nullptr) imm->Ref();
|
|
current->Ref(); // Version 读引用计数增一
|
|
Version::GetStats stats;
|
|
// Unlock while reading from files and memtables
|
|
{
|
|
mutex_.Unlock();
|
|
// First look in the memtable, then in the immutable memtable (if any).
|
|
LookupKey lkey(key, snapshot);
|
|
if (mem->Get(lkey, value, &s )) {
|
|
// Done
|
|
} else if (imm != nullptr && imm->Get(lkey, value, &s)) {
|
|
// Done
|
|
} else {
|
|
//在Version中查找是否包含指定key值
|
|
s = current->Get(options, lkey, value, &stats);
|
|
}
|
|
mutex_.Lock();
|
|
}
|
|
mem->Unref();
|
|
if (imm != nullptr) imm->Unref();
|
|
current->Unref(); //Version 读引用计数减一
|
|
return s;
|
|
}
|
|
|
|
Status DBImpl::Get(const ReadOptions& options, const Slice& key,
|
|
std::string* value) {
|
|
Status s;
|
|
MutexLock l(&mutex_);
|
|
SequenceNumber snapshot;
|
|
if (options.snapshot != nullptr) {
|
|
snapshot =
|
|
static_cast<const SnapshotImpl*>(options.snapshot)->sequence_number();
|
|
} else {
|
|
snapshot = versions_->LastSequence();
|
|
}
|
|
|
|
MemTable* mem = mem_;
|
|
MemTable* imm = imm_;
|
|
Version* current = versions_->current();
|
|
mem->Ref();
|
|
if (imm != nullptr) imm->Ref();
|
|
current->Ref();
|
|
|
|
bool have_stat_update = false;
|
|
Version::GetStats stats;
|
|
|
|
// Unlock while reading from files and memtables
|
|
{
|
|
mutex_.Unlock();
|
|
// First look in the memtable, then in the immutable memtable (if any).
|
|
LookupKey lkey(key, snapshot);
|
|
if (mem->Get(lkey, value, &s)) {
|
|
// Done
|
|
} else if (imm != nullptr && imm->Get(lkey, value, &s)) {
|
|
// Done
|
|
} else {
|
|
s = current->Get(options, lkey, value, &stats);
|
|
have_stat_update = true;
|
|
}
|
|
mutex_.Lock();
|
|
}
|
|
|
|
if (have_stat_update && current->UpdateStats(stats)) {
|
|
MaybeScheduleCompaction();
|
|
}
|
|
mem->Unref();
|
|
if (imm != nullptr) imm->Unref();
|
|
current->Unref();
|
|
|
|
/* Vlog 读取 value */
|
|
if (s.ok() && s.IsSeparated()) {
|
|
|
|
struct VlogReporter : public VlogReader::Reporter {
|
|
Status* status;
|
|
void Corruption(size_t bytes, const Status& s) override {
|
|
if (this->status->ok()) *this->status = s;
|
|
}
|
|
};
|
|
|
|
VlogReporter reporter;
|
|
Slice vlog_ptr(*value);
|
|
uint64_t file_no;
|
|
uint64_t offset;
|
|
uint64_t val_size;
|
|
size_t key_size = key.size();
|
|
|
|
/* 已知该 value 保存在 VLog，解码出 vlog_ptr（fid, offset, val_size）*/
|
|
GetVarint64(&vlog_ptr, &file_no);
|
|
GetVarint64(&vlog_ptr, &offset);
|
|
GetVarint64(&vlog_ptr, &val_size);
|
|
|
|
/* VLog 内部 kv 对的编码长度，1B 为 type */
|
|
uint64_t encoded_len = 1 + VarintLength(key_size) + key.size() + VarintLength(val_size) + val_size;
|
|
|
|
std::string fname = LogFileName(dbname_, file_no);
|
|
RandomAccessFile* file;
|
|
s = env_->NewRandomAccessFile(fname,&file);
|
|
if (!s.ok()) {
|
|
return s;
|
|
}
|
|
|
|
VlogReader vlogReader(file, &reporter);
|
|
Slice key_value;
|
|
char* scratch = new char[encoded_len];
|
|
|
|
if (vlogReader.ReadValue(offset, encoded_len, &key_value, scratch)) {
|
|
value->clear();
|
|
if (!ParseVlogValue(key_value, key, *value, val_size)) {
|
|
s = Status::Corruption("value in vlog isn't match with given key");
|
|
}
|
|
} else {
|
|
s = Status::Corruption("read vlog error");
|
|
}
|
|
|
|
delete file;
|
|
file = nullptr;
|
|
}
|
|
|
|
return s;
|
|
}
|
|
|
|
Iterator* DBImpl::NewIterator(const ReadOptions& options) {
|
|
SequenceNumber latest_snapshot;
|
|
uint32_t seed;
|
|
Iterator* iter = NewInternalIterator(options, &latest_snapshot, &seed);
|
|
return NewDBIterator(this, user_comparator(), iter,
|
|
(options.snapshot != nullptr
|
|
? static_cast<const SnapshotImpl*>(options.snapshot)
|
|
->sequence_number()
|
|
: latest_snapshot),
|
|
seed);
|
|
}
|
|
|
|
void DBImpl::RecordReadSample(Slice key) {
|
|
MutexLock l(&mutex_);
|
|
if (versions_->current()->RecordReadSample(key)) {
|
|
MaybeScheduleCompaction();
|
|
}
|
|
}
|
|
|
|
const Snapshot* DBImpl::GetSnapshot() {
|
|
MutexLock l(&mutex_);
|
|
// begin 注释：建立快照，对快照之后的信息不用进行 GC
|
|
finish_back_garbage_collection_ = true;
|
|
// end
|
|
return snapshots_.New(versions_->LastSequence());
|
|
}
|
|
|
|
void DBImpl::ReleaseSnapshot(const Snapshot* snapshot) {
|
|
MutexLock l(&mutex_);
|
|
snapshots_.Delete(static_cast<const SnapshotImpl*>(snapshot));
|
|
// begin 注释：没有快照，重新进行后台 GC
|
|
if (snapshots_.empty()) {
|
|
finish_back_garbage_collection_ = false;
|
|
}
|
|
// end
|
|
}
|
|
|
|
/*** DBImpl 类关于 Fields 类的 Put、Get 接口 ***/
|
|
Status DBImpl::PutFields(const WriteOptions& o, const Slice& key, const Fields& fields) {
|
|
return DBImpl::Put(o, key, Slice(fields.SerializeValue()));
|
|
}
|
|
|
|
Status DBImpl::GetFields(const ReadOptions& o, const Slice& key, Fields& fields) {
|
|
std::string value_str;
|
|
|
|
Status s = DBImpl::Get(o, key, &value_str);
|
|
if (!s.ok()) return s;
|
|
|
|
fields = Fields::ParseValue(value_str);
|
|
return Status::OK();
|
|
}
|
|
/**************************************************/
|
|
|
|
// Convenience methods
|
|
Status DBImpl::Put(const WriteOptions& o, const Slice& key, const Slice& val) {
|
|
return DB::Put(o, key, val);
|
|
}
|
|
|
|
Status DBImpl::Delete(const WriteOptions& options, const Slice& key) {
|
|
return DB::Delete(options, key);
|
|
}
|
|
|
|
Status DBImpl::Write(const WriteOptions& options, WriteBatch* updates) {
|
|
Writer w(&mutex_);
|
|
w.batch = updates;
|
|
w.sync = options.sync;
|
|
w.done = false;
|
|
|
|
MutexLock l(&mutex_);
|
|
writers_.push_back(&w);
|
|
while (!w.done && &w != writers_.front()) {
|
|
w.cv.Wait();
|
|
}
|
|
if (w.done) {
|
|
return w.status;
|
|
}
|
|
|
|
// May temporarily unlock and wait.
|
|
Status status = MakeRoomForWrite(updates == nullptr);
|
|
uint64_t last_sequence = versions_->LastSequence();
|
|
Writer* last_writer = &w;
|
|
|
|
if (status.ok() && updates != nullptr) { // nullptr batch is for compactions
|
|
WriteBatch* write_batch = BuildBatchGroup(&last_writer);
|
|
|
|
// begin 注释：GC 流程中写回的 WriteBatch 在 CollectionValueLog 函数中已经设置好了
|
|
if (!write_batch->IsGarbageCollection()) {
|
|
// 判断是否需要进行 GC
|
|
// 如需要，空出一块 sequence 区域, 触发 GC 将在 MakeRoomForWrite 里
|
|
// 先判断是否要进行 gc 后台回收
|
|
// 如果建立了快照，finish_back_garbage_collection_ 就为 true
|
|
// 此时不进行 sequence 分配
|
|
if (!finish_back_garbage_collection_
|
|
&& garbage_collection_management_->ConvertQueue(last_sequence)) {
|
|
MaybeScheduleGarbageCollection();
|
|
}
|
|
// SetSequence 在 write_batch 中写入本次的 sequence
|
|
WriteBatchInternal::SetSequence(write_batch, last_sequence + 1);
|
|
last_sequence += WriteBatchInternal::Count(write_batch);
|
|
}
|
|
// 这里设置 last_sequence 是为了确保离线 GC 的时候，
|
|
// 在 map 存在的时候需要调用 ConvertQueue 给回收任务分配 sequence
|
|
versions_->SetLastSequence(last_sequence);
|
|
vlog_kv_numbers_ += WriteBatchInternal::Count(write_batch);
|
|
// end
|
|
|
|
// Add to log and apply to memtable. We can release the lock
|
|
// during this phase since &w is currently responsible for logging
|
|
// and protects against concurrent loggers and concurrent writes
|
|
// into mem_.
|
|
{
|
|
mutex_.Unlock();
|
|
// 先写入vlog再写入memtable
|
|
// 写vlog日志 offset 表示这个 write_batch 在vlog中的偏移地址。
|
|
uint64_t offset = 0;
|
|
status = vlog_->AddRecord(WriteBatchInternal::Contents(write_batch),offset);
|
|
// status = log_->AddRecord(WriteBatchInternal::Contents(write_batch));
|
|
bool sync_error = false;
|
|
if (status.ok() && options.sync) {
|
|
status = logfile_->Sync();
|
|
if (!status.ok()) {
|
|
sync_error = true;
|
|
}
|
|
}
|
|
if (status.ok()) {
|
|
status = WriteBatchInternal::InsertInto(write_batch, mem_, logfile_number_, offset);
|
|
}
|
|
mutex_.Lock();
|
|
if (sync_error) {
|
|
// The state of the log file is indeterminate: the log record we
|
|
// just added may or may not show up when the DB is re-opened.
|
|
// So we force the DB into a mode where all future writes fail.
|
|
RecordBackgroundError(status);
|
|
}
|
|
}
|
|
if (write_batch == tmp_batch_) tmp_batch_->Clear();
|
|
|
|
versions_->SetLastSequence(last_sequence);
|
|
}
|
|
|
|
while (true) {
|
|
Writer* ready = writers_.front();
|
|
writers_.pop_front();
|
|
if (ready != &w) {
|
|
ready->status = status;
|
|
ready->done = true;
|
|
ready->cv.Signal();
|
|
}
|
|
if (ready == last_writer) break;
|
|
}
|
|
|
|
// Notify new head of write queue
|
|
if (!writers_.empty()) {
|
|
writers_.front()->cv.Signal();
|
|
}
|
|
|
|
return status;
|
|
}
|
|
|
|
// REQUIRES: Writer list must be non-empty
|
|
// REQUIRES: First writer must have a non-null batch
|
|
WriteBatch* DBImpl::BuildBatchGroup(Writer** last_writer) {
|
|
mutex_.AssertHeld();
|
|
assert(!writers_.empty());
|
|
Writer* first = writers_.front();
|
|
WriteBatch* result = first->batch;
|
|
assert(result != nullptr);
|
|
|
|
size_t size = WriteBatchInternal::ByteSize(first->batch);
|
|
|
|
// Allow the group to grow up to a maximum size, but if the
|
|
// original write is small, limit the growth so we do not slow
|
|
// down the small write too much.
|
|
size_t max_size = 1 << 20;
|
|
if (size <= (128 << 10)) {
|
|
max_size = size + (128 << 10);
|
|
}
|
|
|
|
*last_writer = first;
|
|
std::deque<Writer*>::iterator iter = writers_.begin();
|
|
++iter; // Advance past "first"
|
|
for (; iter != writers_.end(); ++iter) {
|
|
Writer* w = *iter;
|
|
// begin 注释：写队列中如果遍历到是 gc 的 WriteBatch，停止合并
|
|
if (w->sync && !first->sync
|
|
|| first->batch->IsGarbageCollection()
|
|
|| w->batch->IsGarbageCollection()) {
|
|
// 当前 Writer要求 Sync ，而第一个 Writer 不要求 Sync，两个磁盘写入策略不一致。不做合并操作
|
|
// Do not include a sync write into a batch handled by a non-sync write.
|
|
break;
|
|
}
|
|
// end
|
|
if (w->batch != nullptr) {
|
|
size += WriteBatchInternal::ByteSize(w->batch);
|
|
if (size > max_size) {
|
|
// Do not make batch too big
|
|
break;
|
|
}
|
|
|
|
// Append to *result
|
|
if (result == first->batch) {
|
|
// Switch to temporary batch instead of disturbing caller's batch
|
|
result = tmp_batch_;
|
|
assert(WriteBatchInternal::Count(result) == 0);
|
|
WriteBatchInternal::Append(result, first->batch);
|
|
}
|
|
WriteBatchInternal::Append(result, w->batch);
|
|
}
|
|
*last_writer = w;
|
|
}
|
|
return result;
|
|
}
|
|
|
|
// REQUIRES: mutex_ is held
|
|
// REQUIRES: this thread is currently at the front of the writer queue
|
|
Status DBImpl::MakeRoomForWrite(bool force) {
|
|
mutex_.AssertHeld();
|
|
assert(!writers_.empty());
|
|
bool allow_delay = !force;
|
|
Status s;
|
|
|
|
if (logfile_->GetSize() > options_.max_value_log_size) {
|
|
uint64_t new_log_number = versions_->NewFileNumber();
|
|
WritableFile* lfile = nullptr;
|
|
s = env_->NewWritableFile(LogFileName(dbname_, new_log_number), &lfile);
|
|
if (!s.ok()) {
|
|
versions_->ReuseFileNumber(new_log_number);
|
|
}
|
|
garbage_collection_management_->WriteFileMap(logfile_number_, vlog_kv_numbers_, logfile_->GetSize());
|
|
vlog_kv_numbers_ = 0;
|
|
delete vlog_;
|
|
delete logfile_;
|
|
logfile_ = lfile;
|
|
logfile_number_ = new_log_number;
|
|
vlog_ = new log::VlogWriter(lfile);
|
|
}
|
|
|
|
while (true) {
|
|
if (!bg_error_.ok()) {
|
|
// Yield previous error
|
|
s = bg_error_;
|
|
break;
|
|
} else if (allow_delay && versions_->NumLevelFiles(0) >=
|
|
config::kL0_SlowdownWritesTrigger) {
|
|
// We are getting close to hitting a hard limit on the number of
|
|
// L0 files. Rather than delaying a single write by several
|
|
// seconds when we hit the hard limit, start delaying each
|
|
// individual write by 1ms to reduce latency variance. Also,
|
|
// this delay hands over some CPU to the compaction thread in
|
|
// case it is sharing the same core as the writer.
|
|
mutex_.Unlock();
|
|
env_->SleepForMicroseconds(1000);
|
|
allow_delay = false; // Do not delay a single write more than once
|
|
mutex_.Lock();
|
|
} else if (!force &&
|
|
(mem_->ApproximateMemoryUsage() <= options_.write_buffer_size)) {
|
|
// There is room in current memtable
|
|
break;
|
|
} else if (imm_ != nullptr) {
|
|
// We have filled up the current memtable, but the previous
|
|
// one is still being compacted, so we wait.
|
|
Log(options_.info_log, "Current memtable full; waiting...\n");
|
|
background_work_finished_signal_.Wait();
|
|
} else if (versions_->NumLevelFiles(0) >= config::kL0_StopWritesTrigger) {
|
|
// There are too many level-0 files.
|
|
Log(options_.info_log, "Too many L0 files; waiting...\n");
|
|
background_work_finished_signal_.Wait();
|
|
} else {
|
|
// Attempt to switch to a new memtable and trigger compaction of old
|
|
assert(versions_->PrevLogNumber() == 0);
|
|
|
|
mem_->SetLogFileNumber(logfile_number_);
|
|
|
|
imm_ = mem_;
|
|
has_imm_.store(true, std::memory_order_release);
|
|
mem_ = new MemTable(internal_comparator_);
|
|
mem_->Ref();
|
|
force = false; // Do not force another compaction if have room
|
|
MaybeScheduleCompaction();
|
|
}
|
|
}
|
|
return s;
|
|
}
|
|
|
|
bool DBImpl::GetProperty(const Slice& property, std::string* value) {
|
|
value->clear();
|
|
|
|
MutexLock l(&mutex_);
|
|
Slice in = property;
|
|
Slice prefix("leveldb.");
|
|
if (!in.starts_with(prefix)) return false;
|
|
in.remove_prefix(prefix.size());
|
|
|
|
if (in.starts_with("num-files-at-level")) {
|
|
in.remove_prefix(strlen("num-files-at-level"));
|
|
uint64_t level;
|
|
bool ok = ConsumeDecimalNumber(&in, &level) && in.empty();
|
|
if (!ok || level >= config::kNumLevels) {
|
|
return false;
|
|
} else {
|
|
char buf[100];
|
|
std::snprintf(buf, sizeof(buf), "%d",
|
|
versions_->NumLevelFiles(static_cast<int>(level)));
|
|
*value = buf;
|
|
return true;
|
|
}
|
|
} else if (in == "stats") {
|
|
char buf[200];
|
|
std::snprintf(buf, sizeof(buf),
|
|
" Compactions\n"
|
|
"Level Files Size(MB) Time(sec) Read(MB) Write(MB)\n"
|
|
"--------------------------------------------------\n");
|
|
value->append(buf);
|
|
for (int level = 0; level < config::kNumLevels; level++) {
|
|
int files = versions_->NumLevelFiles(level);
|
|
if (stats_[level].micros > 0 || files > 0) {
|
|
std::snprintf(buf, sizeof(buf), "%3d %8d %8.0f %9.0f %8.0f %9.0f\n",
|
|
level, files, versions_->NumLevelBytes(level) / 1048576.0,
|
|
stats_[level].micros / 1e6,
|
|
stats_[level].bytes_read / 1048576.0,
|
|
stats_[level].bytes_written / 1048576.0);
|
|
value->append(buf);
|
|
}
|
|
}
|
|
return true;
|
|
} else if (in == "sstables") {
|
|
*value = versions_->current()->DebugString();
|
|
return true;
|
|
} else if (in == "approximate-memory-usage") {
|
|
size_t total_usage = options_.block_cache->TotalCharge();
|
|
if (mem_) {
|
|
total_usage += mem_->ApproximateMemoryUsage();
|
|
}
|
|
if (imm_) {
|
|
total_usage += imm_->ApproximateMemoryUsage();
|
|
}
|
|
char buf[50];
|
|
std::snprintf(buf, sizeof(buf), "%llu",
|
|
static_cast<unsigned long long>(total_usage));
|
|
value->append(buf);
|
|
return true;
|
|
}
|
|
|
|
return false;
|
|
}
|
|
|
|
void DBImpl::GetApproximateSizes(const Range* range, int n, uint64_t* sizes) {
|
|
// TODO(opt): better implementation
|
|
MutexLock l(&mutex_);
|
|
Version* v = versions_->current();
|
|
v->Ref();
|
|
|
|
for (int i = 0; i < n; i++) {
|
|
// Convert user_key into a corresponding internal key.
|
|
InternalKey k1(range[i].start, kMaxSequenceNumber, kValueTypeForSeek);
|
|
InternalKey k2(range[i].limit, kMaxSequenceNumber, kValueTypeForSeek);
|
|
uint64_t start = versions_->ApproximateOffsetOf(v, k1);
|
|
uint64_t limit = versions_->ApproximateOffsetOf(v, k2);
|
|
sizes[i] = (limit >= start ? limit - start : 0);
|
|
}
|
|
|
|
v->Unref();
|
|
}
|
|
|
|
// Default implementations of convenience methods that subclasses of DB
|
|
// can call if they wish
|
|
Status DB::Put(const WriteOptions& opt, const Slice& key, const Slice& value) {
|
|
WriteBatch batch(opt.separate_threshold);
|
|
batch.Put(key, value);
|
|
return Write(opt, &batch);
|
|
}
|
|
|
|
Status DB::Delete(const WriteOptions& opt, const Slice& key) {
|
|
WriteBatch batch;
|
|
batch.Delete(key);
|
|
return Write(opt, &batch);
|
|
}
|
|
|
|
DB::~DB() = default;
|
|
|
|
Status DB::Open(const Options& options, const std::string& dbname, DB** dbptr) {
|
|
*dbptr = nullptr;
|
|
|
|
DBImpl* impl = new DBImpl(options, dbname);
|
|
impl->mutex_.Lock();
|
|
VersionEdit edit;
|
|
// Recover handles create_if_missing, error_if_exists
|
|
bool save_manifest = false;
|
|
Status s = impl->Recover(&edit, &save_manifest);
|
|
|
|
// begin 注释： Recover 之后，获取所有 VLogs
|
|
std::vector<uint64_t> logs;
|
|
s = impl->GetAllValueLog(dbname, logs);
|
|
sort(logs.begin(),logs.end());
|
|
// end
|
|
|
|
if (s.ok() && impl->mem_ == nullptr) {
|
|
// Create new log and a corresponding memtable.
|
|
uint64_t new_log_number = impl->versions_->NewFileNumber();
|
|
WritableFile* lfile;
|
|
s = options.env->NewWritableFile(LogFileName(dbname, new_log_number),
|
|
&lfile);
|
|
if (s.ok()) {
|
|
edit.SetLogNumber(new_log_number);
|
|
impl->logfile_ = lfile;
|
|
impl->logfile_number_ = new_log_number;
|
|
impl->vlog_ = new log::VlogWriter(lfile);
|
|
impl->mem_ = new MemTable(impl->internal_comparator_);
|
|
impl->mem_->Ref();
|
|
}
|
|
}
|
|
if (s.ok() && save_manifest) {
|
|
edit.SetPrevLogNumber(0); // No older logs needed after recovery.
|
|
edit.SetLogNumber(impl->logfile_number_);
|
|
// s = impl->versions_->LogAndApply(&edit, &impl->mutex_);
|
|
// begin 注释：把 imm_last_sequence 设置到新的 manifest 当中，
|
|
// 即 RecoverLogFile 中判断上一次断电时的数据库状态的 imm -> sst 的情况，
|
|
// 表示一次成功的全盘恢复
|
|
impl->versions_->StartImmLastSequence(true);
|
|
s = impl->versions_->LogAndApply(&edit, &impl->mutex_);
|
|
impl->versions_->StartImmLastSequence(false);
|
|
// end
|
|
}
|
|
if (s.ok()) {
|
|
impl->RemoveObsoleteFiles();
|
|
impl->MaybeScheduleCompaction();
|
|
}
|
|
|
|
// begin 开始全盘回收
|
|
|
|
if (s.ok() && impl->options_.start_garbage_collection) {
|
|
if( s.ok() ){
|
|
int size = logs.size();
|
|
for( int i = 0; i < size ; i++){
|
|
uint64_t fid = logs[i];
|
|
uint64_t next_sequence = impl->versions_->LastSequence() + 1;
|
|
std::cout<<" collection file : "<<fid<<std::endl;
|
|
impl->mutex_.Unlock();
|
|
Status stmp = impl->CollectionValueLog(fid, next_sequence);
|
|
impl->mutex_.Lock();
|
|
if (!stmp.ok()) s = stmp;
|
|
impl->versions_->SetLastSequence(next_sequence - 1);
|
|
}
|
|
}
|
|
}
|
|
// end
|
|
impl->mutex_.Unlock();
|
|
if (s.ok()) {
|
|
assert(impl->mem_ != nullptr);
|
|
*dbptr = impl;
|
|
} else {
|
|
delete impl;
|
|
}
|
|
return s;
|
|
}
|
|
|
|
Snapshot::~Snapshot() = default;
|
|
|
|
Status DestroyDB(const std::string& dbname, const Options& options) {
|
|
Env* env = options.env;
|
|
std::vector<std::string> filenames;
|
|
Status result = env->GetChildren(dbname, &filenames);
|
|
if (!result.ok()) {
|
|
// Ignore error in case directory does not exist
|
|
return Status::OK();
|
|
}
|
|
|
|
FileLock* lock;
|
|
const std::string lockname = LockFileName(dbname);
|
|
result = env->LockFile(lockname, &lock);
|
|
if (result.ok()) {
|
|
uint64_t number;
|
|
FileType type;
|
|
for (size_t i = 0; i < filenames.size(); i++) {
|
|
if (ParseFileName(filenames[i], &number, &type) &&
|
|
type != kDBLockFile) { // Lock file will be deleted at end
|
|
Status del = env->RemoveFile(dbname + "/" + filenames[i]);
|
|
if (result.ok() && !del.ok()) {
|
|
result = del;
|
|
}
|
|
}
|
|
}
|
|
env->UnlockFile(lock); // Ignore error since state is already gone
|
|
env->RemoveFile(lockname);
|
|
env->RemoveDir(dbname); // Ignore error in case dir contains other files
|
|
}
|
|
return result;
|
|
}
|
|
|
|
} // namespace leveldb
|