2016-02-10 00:12:00 +01:00
|
|
|
// Copyright (c) 2011-present, Facebook, Inc. All rights reserved.
|
2017-07-16 01:03:42 +02:00
|
|
|
// 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).
|
2014-10-30 01:43:37 +01:00
|
|
|
//
|
|
|
|
// 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/wal_manager.h"
|
|
|
|
|
|
|
|
#include <algorithm>
|
2019-09-20 21:00:55 +02:00
|
|
|
#include <cinttypes>
|
2014-10-30 01:43:37 +01:00
|
|
|
#include <memory>
|
2019-09-20 21:00:55 +02:00
|
|
|
#include <vector>
|
2014-10-30 01:43:37 +01:00
|
|
|
|
|
|
|
#include "db/log_reader.h"
|
|
|
|
#include "db/log_writer.h"
|
2017-04-04 03:27:24 +02:00
|
|
|
#include "db/transaction_log_impl.h"
|
2014-10-30 01:43:37 +01:00
|
|
|
#include "db/write_batch_internal.h"
|
2019-05-30 05:44:08 +02:00
|
|
|
#include "file/file_util.h"
|
|
|
|
#include "file/filename.h"
|
2019-09-16 19:31:27 +02:00
|
|
|
#include "file/sequence_file_reader.h"
|
2019-06-01 02:19:43 +02:00
|
|
|
#include "logging/logging.h"
|
2014-10-30 01:43:37 +01:00
|
|
|
#include "port/port.h"
|
|
|
|
#include "rocksdb/env.h"
|
|
|
|
#include "rocksdb/options.h"
|
|
|
|
#include "rocksdb/write_batch.h"
|
2019-05-31 02:39:43 +02:00
|
|
|
#include "test_util/sync_point.h"
|
2017-07-29 01:23:50 +02:00
|
|
|
#include "util/cast_util.h"
|
2014-10-30 01:43:37 +01:00
|
|
|
#include "util/coding.h"
|
|
|
|
#include "util/mutexlock.h"
|
2014-11-25 05:44:49 +01:00
|
|
|
#include "util/string_util.h"
|
2014-10-30 01:43:37 +01:00
|
|
|
|
2020-02-20 21:07:53 +01:00
|
|
|
namespace ROCKSDB_NAMESPACE {
|
2014-10-30 01:43:37 +01:00
|
|
|
|
|
|
|
#ifndef ROCKSDB_LITE
|
|
|
|
|
2018-01-12 03:48:19 +01:00
|
|
|
Status WalManager::DeleteFile(const std::string& fname, uint64_t number) {
|
|
|
|
auto s = env_->DeleteFile(db_options_.wal_dir + "/" + fname);
|
|
|
|
if (s.ok()) {
|
|
|
|
MutexLock l(&read_first_record_cache_mutex_);
|
|
|
|
read_first_record_cache_.erase(number);
|
|
|
|
}
|
|
|
|
return s;
|
|
|
|
}
|
|
|
|
|
2014-10-30 01:43:37 +01:00
|
|
|
Status WalManager::GetSortedWalFiles(VectorLogPtr& files) {
|
|
|
|
// First get sorted files in db dir, then get sorted files from archived
|
|
|
|
// dir, to avoid a race condition where a log file is moved to archived
|
|
|
|
// dir in between.
|
|
|
|
Status s;
|
|
|
|
// list wal files in main db dir.
|
|
|
|
VectorLogPtr logs;
|
|
|
|
s = GetSortedWalsOfType(db_options_.wal_dir, logs, kAliveLogFile);
|
|
|
|
if (!s.ok()) {
|
|
|
|
return s;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Reproduce the race condition where a log file is moved
|
|
|
|
// to archived dir, between these two sync points, used in
|
|
|
|
// (DBTest,TransactionLogIteratorRace)
|
|
|
|
TEST_SYNC_POINT("WalManager::GetSortedWalFiles:1");
|
|
|
|
TEST_SYNC_POINT("WalManager::GetSortedWalFiles:2");
|
|
|
|
|
|
|
|
files.clear();
|
|
|
|
// list wal files in archive dir.
|
|
|
|
std::string archivedir = ArchivalDirectory(db_options_.wal_dir);
|
2015-07-21 02:20:40 +02:00
|
|
|
Status exists = env_->FileExists(archivedir);
|
|
|
|
if (exists.ok()) {
|
2014-10-30 01:43:37 +01:00
|
|
|
s = GetSortedWalsOfType(archivedir, files, kArchivedLogFile);
|
|
|
|
if (!s.ok()) {
|
|
|
|
return s;
|
|
|
|
}
|
2015-07-21 02:20:40 +02:00
|
|
|
} else if (!exists.IsNotFound()) {
|
|
|
|
assert(s.IsIOError());
|
|
|
|
return s;
|
2014-10-30 01:43:37 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
uint64_t latest_archived_log_number = 0;
|
|
|
|
if (!files.empty()) {
|
|
|
|
latest_archived_log_number = files.back()->LogNumber();
|
2017-03-16 03:22:52 +01:00
|
|
|
ROCKS_LOG_INFO(db_options_.info_log, "Latest Archived log: %" PRIu64,
|
|
|
|
latest_archived_log_number);
|
2014-10-30 01:43:37 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
files.reserve(files.size() + logs.size());
|
|
|
|
for (auto& log : logs) {
|
|
|
|
if (log->LogNumber() > latest_archived_log_number) {
|
|
|
|
files.push_back(std::move(log));
|
|
|
|
} else {
|
|
|
|
// When the race condition happens, we could see the
|
|
|
|
// same log in both db dir and archived dir. Simply
|
|
|
|
// ignore the one in db dir. Note that, if we read
|
|
|
|
// archived dir first, we would have missed the log file.
|
2017-03-16 03:22:52 +01:00
|
|
|
ROCKS_LOG_WARN(db_options_.info_log, "%s already moved to archive",
|
|
|
|
log->PathName().c_str());
|
2014-10-30 01:43:37 +01:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return s;
|
|
|
|
}
|
|
|
|
|
|
|
|
Status WalManager::GetUpdatesSince(
|
|
|
|
SequenceNumber seq, std::unique_ptr<TransactionLogIterator>* iter,
|
|
|
|
const TransactionLogIterator::ReadOptions& read_options,
|
|
|
|
VersionSet* version_set) {
|
|
|
|
|
|
|
|
// Get all sorted Wal Files.
|
|
|
|
// Do binary search and open files and find the seq number.
|
|
|
|
|
|
|
|
std::unique_ptr<VectorLogPtr> wal_files(new VectorLogPtr);
|
|
|
|
Status s = GetSortedWalFiles(*wal_files);
|
|
|
|
if (!s.ok()) {
|
|
|
|
return s;
|
|
|
|
}
|
|
|
|
|
|
|
|
s = RetainProbableWalFiles(*wal_files, seq);
|
|
|
|
if (!s.ok()) {
|
|
|
|
return s;
|
|
|
|
}
|
|
|
|
iter->reset(new TransactionLogIteratorImpl(
|
Introduce a new storage specific Env API (#5761)
Summary:
The current Env API encompasses both storage/file operations, as well as OS related operations. Most of the APIs return a Status, which does not have enough metadata about an error, such as whether its retry-able or not, scope (i.e fault domain) of the error etc., that may be required in order to properly handle a storage error. The file APIs also do not provide enough control over the IO SLA, such as timeout, prioritization, hinting about placement and redundancy etc.
This PR separates out the file/storage APIs from Env into a new FileSystem class. The APIs are updated to return an IOStatus with metadata about the error, as well as to take an IOOptions structure as input in order to allow more control over the IO.
The user can set both ```options.env``` and ```options.file_system``` to specify that RocksDB should use the former for OS related operations and the latter for storage operations. Internally, a ```CompositeEnvWrapper``` has been introduced that inherits from ```Env``` and redirects individual methods to either an ```Env``` implementation or the ```FileSystem``` as appropriate. When options are sanitized during ```DB::Open```, ```options.env``` is replaced with a newly allocated ```CompositeEnvWrapper``` instance if both env and file_system have been specified. This way, the rest of the RocksDB code can continue to function as before.
This PR also ports PosixEnv to the new API by splitting it into two - PosixEnv and PosixFileSystem. PosixEnv is defined as a sub-class of CompositeEnvWrapper, and threading/time functions are overridden with Posix specific implementations in order to avoid an extra level of indirection.
The ```CompositeEnvWrapper``` translates ```IOStatus``` return code to ```Status```, and sets the severity to ```kSoftError``` if the io_status is retryable. The error handling code in RocksDB can then recover the DB automatically.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/5761
Differential Revision: D18868376
Pulled By: anand1976
fbshipit-source-id: 39efe18a162ea746fabac6360ff529baba48486f
2019-12-13 23:47:08 +01:00
|
|
|
db_options_.wal_dir, &db_options_, read_options, file_options_, seq,
|
2020-08-13 02:28:10 +02:00
|
|
|
std::move(wal_files), version_set, seq_per_batch_, io_tracer_));
|
2014-10-30 01:43:37 +01:00
|
|
|
return (*iter)->status();
|
|
|
|
}
|
|
|
|
|
|
|
|
// 1. Go through all archived files and
|
|
|
|
// a. if ttl is enabled, delete outdated files
|
|
|
|
// b. if archive size limit is enabled, delete empty files,
|
|
|
|
// compute file number and size.
|
|
|
|
// 2. If size limit is enabled:
|
|
|
|
// a. compute how many files should be deleted
|
|
|
|
// b. get sorted non-empty archived logs
|
|
|
|
// c. delete what should be deleted
|
|
|
|
void WalManager::PurgeObsoleteWALFiles() {
|
2016-09-24 01:34:04 +02:00
|
|
|
bool const ttl_enabled = db_options_.wal_ttl_seconds > 0;
|
|
|
|
bool const size_limit_enabled = db_options_.wal_size_limit_mb > 0;
|
2014-10-30 01:43:37 +01:00
|
|
|
if (!ttl_enabled && !size_limit_enabled) {
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
int64_t current_time;
|
|
|
|
Status s = env_->GetCurrentTime(¤t_time);
|
|
|
|
if (!s.ok()) {
|
2017-03-16 03:22:52 +01:00
|
|
|
ROCKS_LOG_ERROR(db_options_.info_log, "Can't get current time: %s",
|
|
|
|
s.ToString().c_str());
|
2014-10-30 01:43:37 +01:00
|
|
|
assert(false);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
uint64_t const now_seconds = static_cast<uint64_t>(current_time);
|
|
|
|
uint64_t const time_to_check = (ttl_enabled && !size_limit_enabled)
|
2016-09-24 01:34:04 +02:00
|
|
|
? db_options_.wal_ttl_seconds / 2
|
2014-10-30 01:43:37 +01:00
|
|
|
: kDefaultIntervalToDeleteObsoleteWAL;
|
|
|
|
|
|
|
|
if (purge_wal_files_last_run_ + time_to_check > now_seconds) {
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
purge_wal_files_last_run_ = now_seconds;
|
|
|
|
|
|
|
|
std::string archival_dir = ArchivalDirectory(db_options_.wal_dir);
|
|
|
|
std::vector<std::string> files;
|
|
|
|
s = env_->GetChildren(archival_dir, &files);
|
|
|
|
if (!s.ok()) {
|
2017-03-16 03:22:52 +01:00
|
|
|
ROCKS_LOG_ERROR(db_options_.info_log, "Can't get archive files: %s",
|
|
|
|
s.ToString().c_str());
|
2014-10-30 01:43:37 +01:00
|
|
|
assert(false);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
size_t log_files_num = 0;
|
|
|
|
uint64_t log_file_size = 0;
|
|
|
|
|
|
|
|
for (auto& f : files) {
|
|
|
|
uint64_t number;
|
|
|
|
FileType type;
|
2020-10-23 02:04:39 +02:00
|
|
|
if (ParseFileName(f, &number, &type) && type == kWalFile) {
|
2014-10-30 01:43:37 +01:00
|
|
|
std::string const file_path = archival_dir + "/" + f;
|
|
|
|
if (ttl_enabled) {
|
|
|
|
uint64_t file_m_time;
|
2014-10-31 19:59:54 +01:00
|
|
|
s = env_->GetFileModificationTime(file_path, &file_m_time);
|
2014-10-30 01:43:37 +01:00
|
|
|
if (!s.ok()) {
|
2017-03-16 03:22:52 +01:00
|
|
|
ROCKS_LOG_WARN(db_options_.info_log,
|
|
|
|
"Can't get file mod time: %s: %s", file_path.c_str(),
|
|
|
|
s.ToString().c_str());
|
2014-10-30 01:43:37 +01:00
|
|
|
continue;
|
|
|
|
}
|
2016-09-24 01:34:04 +02:00
|
|
|
if (now_seconds - file_m_time > db_options_.wal_ttl_seconds) {
|
2019-07-07 06:04:22 +02:00
|
|
|
s = DeleteDBFile(&db_options_, file_path, archival_dir, false,
|
|
|
|
/*force_fg=*/!wal_in_db_path_);
|
2014-10-30 01:43:37 +01:00
|
|
|
if (!s.ok()) {
|
2017-03-16 03:22:52 +01:00
|
|
|
ROCKS_LOG_WARN(db_options_.info_log, "Can't delete file: %s: %s",
|
|
|
|
file_path.c_str(), s.ToString().c_str());
|
2014-10-30 01:43:37 +01:00
|
|
|
continue;
|
|
|
|
} else {
|
|
|
|
MutexLock l(&read_first_record_cache_mutex_);
|
|
|
|
read_first_record_cache_.erase(number);
|
|
|
|
}
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if (size_limit_enabled) {
|
|
|
|
uint64_t file_size;
|
2014-10-31 19:59:54 +01:00
|
|
|
s = env_->GetFileSize(file_path, &file_size);
|
2014-10-30 01:43:37 +01:00
|
|
|
if (!s.ok()) {
|
2017-03-16 03:22:52 +01:00
|
|
|
ROCKS_LOG_ERROR(db_options_.info_log,
|
|
|
|
"Unable to get file size: %s: %s", file_path.c_str(),
|
|
|
|
s.ToString().c_str());
|
2014-10-30 01:43:37 +01:00
|
|
|
return;
|
|
|
|
} else {
|
|
|
|
if (file_size > 0) {
|
|
|
|
log_file_size = std::max(log_file_size, file_size);
|
|
|
|
++log_files_num;
|
|
|
|
} else {
|
2019-07-07 06:04:22 +02:00
|
|
|
s = DeleteDBFile(&db_options_, file_path, archival_dir, false,
|
|
|
|
/*force_fg=*/!wal_in_db_path_);
|
2014-10-30 01:43:37 +01:00
|
|
|
if (!s.ok()) {
|
2017-03-16 03:22:52 +01:00
|
|
|
ROCKS_LOG_WARN(db_options_.info_log,
|
|
|
|
"Unable to delete file: %s: %s", file_path.c_str(),
|
|
|
|
s.ToString().c_str());
|
2014-10-30 01:43:37 +01:00
|
|
|
continue;
|
|
|
|
} else {
|
|
|
|
MutexLock l(&read_first_record_cache_mutex_);
|
|
|
|
read_first_record_cache_.erase(number);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if (0 == log_files_num || !size_limit_enabled) {
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
size_t const files_keep_num =
|
2018-09-06 03:07:53 +02:00
|
|
|
static_cast<size_t>(db_options_.wal_size_limit_mb * 1024 * 1024 / log_file_size);
|
2014-10-30 01:43:37 +01:00
|
|
|
if (log_files_num <= files_keep_num) {
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
size_t files_del_num = log_files_num - files_keep_num;
|
|
|
|
VectorLogPtr archived_logs;
|
|
|
|
GetSortedWalsOfType(archival_dir, archived_logs, kArchivedLogFile);
|
|
|
|
|
|
|
|
if (files_del_num > archived_logs.size()) {
|
2017-03-16 03:22:52 +01:00
|
|
|
ROCKS_LOG_WARN(db_options_.info_log,
|
|
|
|
"Trying to delete more archived log files than "
|
|
|
|
"exist. Deleting all");
|
2014-10-30 01:43:37 +01:00
|
|
|
files_del_num = archived_logs.size();
|
|
|
|
}
|
|
|
|
|
|
|
|
for (size_t i = 0; i < files_del_num; ++i) {
|
|
|
|
std::string const file_path = archived_logs[i]->PathName();
|
2019-03-28 23:13:02 +01:00
|
|
|
s = DeleteDBFile(&db_options_, db_options_.wal_dir + "/" + file_path,
|
2019-07-07 06:04:22 +02:00
|
|
|
db_options_.wal_dir, false,
|
|
|
|
/*force_fg=*/!wal_in_db_path_);
|
2014-10-30 01:43:37 +01:00
|
|
|
if (!s.ok()) {
|
2017-03-16 03:22:52 +01:00
|
|
|
ROCKS_LOG_WARN(db_options_.info_log, "Unable to delete file: %s: %s",
|
|
|
|
file_path.c_str(), s.ToString().c_str());
|
2014-10-30 01:43:37 +01:00
|
|
|
continue;
|
|
|
|
} else {
|
|
|
|
MutexLock l(&read_first_record_cache_mutex_);
|
|
|
|
read_first_record_cache_.erase(archived_logs[i]->LogNumber());
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
void WalManager::ArchiveWALFile(const std::string& fname, uint64_t number) {
|
|
|
|
auto archived_log_name = ArchivedLogFileName(db_options_.wal_dir, number);
|
|
|
|
// The sync point below is used in (DBTest,TransactionLogIteratorRace)
|
|
|
|
TEST_SYNC_POINT("WalManager::PurgeObsoleteFiles:1");
|
|
|
|
Status s = env_->RenameFile(fname, archived_log_name);
|
|
|
|
// The sync point below is used in (DBTest,TransactionLogIteratorRace)
|
|
|
|
TEST_SYNC_POINT("WalManager::PurgeObsoleteFiles:2");
|
2017-03-16 03:22:52 +01:00
|
|
|
ROCKS_LOG_INFO(db_options_.info_log, "Move log file %s to %s -- %s\n",
|
|
|
|
fname.c_str(), archived_log_name.c_str(),
|
|
|
|
s.ToString().c_str());
|
2014-10-30 01:43:37 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
Status WalManager::GetSortedWalsOfType(const std::string& path,
|
|
|
|
VectorLogPtr& log_files,
|
|
|
|
WalFileType log_type) {
|
|
|
|
std::vector<std::string> all_files;
|
|
|
|
const Status status = env_->GetChildren(path, &all_files);
|
|
|
|
if (!status.ok()) {
|
|
|
|
return status;
|
|
|
|
}
|
|
|
|
log_files.reserve(all_files.size());
|
|
|
|
for (const auto& f : all_files) {
|
|
|
|
uint64_t number;
|
|
|
|
FileType type;
|
2020-10-23 02:04:39 +02:00
|
|
|
if (ParseFileName(f, &number, &type) && type == kWalFile) {
|
2014-10-30 01:43:37 +01:00
|
|
|
SequenceNumber sequence;
|
|
|
|
Status s = ReadFirstRecord(log_type, number, &sequence);
|
|
|
|
if (!s.ok()) {
|
|
|
|
return s;
|
|
|
|
}
|
|
|
|
if (sequence == 0) {
|
|
|
|
// empty file
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Reproduce the race condition where a log file is moved
|
|
|
|
// to archived dir, between these two sync points, used in
|
|
|
|
// (DBTest,TransactionLogIteratorRace)
|
|
|
|
TEST_SYNC_POINT("WalManager::GetSortedWalsOfType:1");
|
|
|
|
TEST_SYNC_POINT("WalManager::GetSortedWalsOfType:2");
|
|
|
|
|
|
|
|
uint64_t size_bytes;
|
|
|
|
s = env_->GetFileSize(LogFileName(path, number), &size_bytes);
|
|
|
|
// re-try in case the alive log file has been moved to archive.
|
2019-09-10 18:40:21 +02:00
|
|
|
if (!s.ok() && log_type == kAliveLogFile) {
|
|
|
|
std::string archived_file = ArchivedLogFileName(path, number);
|
|
|
|
if (env_->FileExists(archived_file).ok()) {
|
|
|
|
s = env_->GetFileSize(archived_file, &size_bytes);
|
|
|
|
if (!s.ok() && env_->FileExists(archived_file).IsNotFound()) {
|
|
|
|
// oops, the file just got deleted from archived dir! move on
|
|
|
|
s = Status::OK();
|
|
|
|
continue;
|
|
|
|
}
|
2015-01-24 02:35:12 +01:00
|
|
|
}
|
2014-10-30 01:43:37 +01:00
|
|
|
}
|
|
|
|
if (!s.ok()) {
|
|
|
|
return s;
|
|
|
|
}
|
|
|
|
|
2015-12-23 21:10:00 +01:00
|
|
|
log_files.push_back(std::unique_ptr<LogFile>(
|
|
|
|
new LogFileImpl(number, log_type, sequence, size_bytes)));
|
2014-10-30 01:43:37 +01:00
|
|
|
}
|
|
|
|
}
|
2019-09-04 20:36:47 +02:00
|
|
|
std::sort(
|
|
|
|
log_files.begin(), log_files.end(),
|
|
|
|
[](const std::unique_ptr<LogFile>& a, const std::unique_ptr<LogFile>& b) {
|
2020-04-29 22:06:27 +02:00
|
|
|
LogFileImpl* a_impl = static_cast_with_check<LogFileImpl>(a.get());
|
|
|
|
LogFileImpl* b_impl = static_cast_with_check<LogFileImpl>(b.get());
|
2019-09-04 20:36:47 +02:00
|
|
|
return *a_impl < *b_impl;
|
|
|
|
});
|
2014-10-30 01:43:37 +01:00
|
|
|
return status;
|
|
|
|
}
|
|
|
|
|
|
|
|
Status WalManager::RetainProbableWalFiles(VectorLogPtr& all_logs,
|
|
|
|
const SequenceNumber target) {
|
|
|
|
int64_t start = 0; // signed to avoid overflow when target is < first file.
|
|
|
|
int64_t end = static_cast<int64_t>(all_logs.size()) - 1;
|
|
|
|
// Binary Search. avoid opening all files.
|
|
|
|
while (end >= start) {
|
|
|
|
int64_t mid = start + (end - start) / 2; // Avoid overflow.
|
2018-09-06 03:07:53 +02:00
|
|
|
SequenceNumber current_seq_num = all_logs.at(static_cast<size_t>(mid))->StartSequence();
|
2014-10-30 01:43:37 +01:00
|
|
|
if (current_seq_num == target) {
|
|
|
|
end = mid;
|
|
|
|
break;
|
|
|
|
} else if (current_seq_num < target) {
|
|
|
|
start = mid + 1;
|
|
|
|
} else {
|
|
|
|
end = mid - 1;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
// end could be -ve.
|
2018-09-06 03:07:53 +02:00
|
|
|
size_t start_index = static_cast<size_t>(std::max(static_cast<int64_t>(0), end));
|
2014-10-30 01:43:37 +01:00
|
|
|
// The last wal file is always included
|
|
|
|
all_logs.erase(all_logs.begin(), all_logs.begin() + start_index);
|
|
|
|
return Status::OK();
|
|
|
|
}
|
|
|
|
|
|
|
|
Status WalManager::ReadFirstRecord(const WalFileType type,
|
|
|
|
const uint64_t number,
|
|
|
|
SequenceNumber* sequence) {
|
2015-01-24 02:35:12 +01:00
|
|
|
*sequence = 0;
|
2014-10-30 01:43:37 +01:00
|
|
|
if (type != kAliveLogFile && type != kArchivedLogFile) {
|
2017-03-16 03:22:52 +01:00
|
|
|
ROCKS_LOG_ERROR(db_options_.info_log, "[WalManger] Unknown file type %s",
|
|
|
|
ToString(type).c_str());
|
2014-11-04 19:34:18 +01:00
|
|
|
return Status::NotSupported(
|
2014-11-25 05:44:49 +01:00
|
|
|
"File Type Not Known " + ToString(type));
|
2014-10-30 01:43:37 +01:00
|
|
|
}
|
|
|
|
{
|
|
|
|
MutexLock l(&read_first_record_cache_mutex_);
|
|
|
|
auto itr = read_first_record_cache_.find(number);
|
|
|
|
if (itr != read_first_record_cache_.end()) {
|
|
|
|
*sequence = itr->second;
|
|
|
|
return Status::OK();
|
|
|
|
}
|
|
|
|
}
|
|
|
|
Status s;
|
|
|
|
if (type == kAliveLogFile) {
|
|
|
|
std::string fname = LogFileName(db_options_.wal_dir, number);
|
2016-09-15 18:55:02 +02:00
|
|
|
s = ReadFirstLine(fname, number, sequence);
|
2019-09-10 18:40:21 +02:00
|
|
|
if (!s.ok() && env_->FileExists(fname).ok()) {
|
2014-10-30 01:43:37 +01:00
|
|
|
// return any error that is not caused by non-existing file
|
|
|
|
return s;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if (type == kArchivedLogFile || !s.ok()) {
|
|
|
|
// check if the file got moved to archive.
|
|
|
|
std::string archived_file =
|
|
|
|
ArchivedLogFileName(db_options_.wal_dir, number);
|
2016-09-15 18:55:02 +02:00
|
|
|
s = ReadFirstLine(archived_file, number, sequence);
|
2015-01-24 02:35:12 +01:00
|
|
|
// maybe the file was deleted from archive dir. If that's the case, return
|
|
|
|
// Status::OK(). The caller with identify this as empty file because
|
|
|
|
// *sequence == 0
|
2015-07-21 02:20:40 +02:00
|
|
|
if (!s.ok() && env_->FileExists(archived_file).IsNotFound()) {
|
2015-01-24 02:35:12 +01:00
|
|
|
return Status::OK();
|
|
|
|
}
|
2014-10-30 01:43:37 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
if (s.ok() && *sequence != 0) {
|
|
|
|
MutexLock l(&read_first_record_cache_mutex_);
|
|
|
|
read_first_record_cache_.insert({number, *sequence});
|
|
|
|
}
|
|
|
|
return s;
|
|
|
|
}
|
|
|
|
|
2019-09-20 21:00:55 +02:00
|
|
|
Status WalManager::GetLiveWalFile(uint64_t number,
|
|
|
|
std::unique_ptr<LogFile>* log_file) {
|
2019-09-04 21:08:56 +02:00
|
|
|
if (!log_file) {
|
|
|
|
return Status::InvalidArgument("log_file not preallocated.");
|
|
|
|
}
|
|
|
|
|
2019-09-20 21:00:55 +02:00
|
|
|
if (!number) {
|
2019-09-04 21:08:56 +02:00
|
|
|
return Status::PathNotFound("log file not available");
|
|
|
|
}
|
|
|
|
|
|
|
|
Status s;
|
|
|
|
|
|
|
|
uint64_t size_bytes;
|
|
|
|
s = env_->GetFileSize(LogFileName(db_options_.wal_dir, number), &size_bytes);
|
|
|
|
|
|
|
|
if (!s.ok()) {
|
|
|
|
return s;
|
|
|
|
}
|
|
|
|
|
2019-09-20 21:00:55 +02:00
|
|
|
log_file->reset(new LogFileImpl(number, kAliveLogFile,
|
|
|
|
0, // SequenceNumber
|
|
|
|
size_bytes));
|
2019-09-04 21:08:56 +02:00
|
|
|
|
|
|
|
return Status::OK();
|
|
|
|
}
|
|
|
|
|
2014-10-30 01:43:37 +01:00
|
|
|
// the function returns status.ok() and sequence == 0 if the file exists, but is
|
|
|
|
// empty
|
|
|
|
Status WalManager::ReadFirstLine(const std::string& fname,
|
2016-09-15 18:55:02 +02:00
|
|
|
const uint64_t number,
|
2014-10-30 01:43:37 +01:00
|
|
|
SequenceNumber* sequence) {
|
|
|
|
struct LogReporter : public log::Reader::Reporter {
|
|
|
|
Env* env;
|
|
|
|
Logger* info_log;
|
|
|
|
const char* fname;
|
|
|
|
|
|
|
|
Status* status;
|
|
|
|
bool ignore_error; // true if db_options_.paranoid_checks==false
|
2019-02-19 22:36:04 +01:00
|
|
|
void Corruption(size_t bytes, const Status& s) override {
|
2017-03-16 03:22:52 +01:00
|
|
|
ROCKS_LOG_WARN(info_log, "[WalManager] %s%s: dropping %d bytes; %s",
|
|
|
|
(this->ignore_error ? "(ignoring error) " : ""), fname,
|
|
|
|
static_cast<int>(bytes), s.ToString().c_str());
|
2014-10-30 01:43:37 +01:00
|
|
|
if (this->status->ok()) {
|
|
|
|
// only keep the first error
|
|
|
|
*this->status = s;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
};
|
|
|
|
|
Introduce a new storage specific Env API (#5761)
Summary:
The current Env API encompasses both storage/file operations, as well as OS related operations. Most of the APIs return a Status, which does not have enough metadata about an error, such as whether its retry-able or not, scope (i.e fault domain) of the error etc., that may be required in order to properly handle a storage error. The file APIs also do not provide enough control over the IO SLA, such as timeout, prioritization, hinting about placement and redundancy etc.
This PR separates out the file/storage APIs from Env into a new FileSystem class. The APIs are updated to return an IOStatus with metadata about the error, as well as to take an IOOptions structure as input in order to allow more control over the IO.
The user can set both ```options.env``` and ```options.file_system``` to specify that RocksDB should use the former for OS related operations and the latter for storage operations. Internally, a ```CompositeEnvWrapper``` has been introduced that inherits from ```Env``` and redirects individual methods to either an ```Env``` implementation or the ```FileSystem``` as appropriate. When options are sanitized during ```DB::Open```, ```options.env``` is replaced with a newly allocated ```CompositeEnvWrapper``` instance if both env and file_system have been specified. This way, the rest of the RocksDB code can continue to function as before.
This PR also ports PosixEnv to the new API by splitting it into two - PosixEnv and PosixFileSystem. PosixEnv is defined as a sub-class of CompositeEnvWrapper, and threading/time functions are overridden with Posix specific implementations in order to avoid an extra level of indirection.
The ```CompositeEnvWrapper``` translates ```IOStatus``` return code to ```Status```, and sets the severity to ```kSoftError``` if the io_status is retryable. The error handling code in RocksDB can then recover the DB automatically.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/5761
Differential Revision: D18868376
Pulled By: anand1976
fbshipit-source-id: 39efe18a162ea746fabac6360ff529baba48486f
2019-12-13 23:47:08 +01:00
|
|
|
std::unique_ptr<FSSequentialFile> file;
|
|
|
|
Status status = fs_->NewSequentialFile(fname,
|
|
|
|
fs_->OptimizeForLogRead(file_options_),
|
|
|
|
&file, nullptr);
|
2018-11-09 20:17:34 +01:00
|
|
|
std::unique_ptr<SequentialFileReader> file_reader(
|
2020-08-19 01:19:22 +02:00
|
|
|
new SequentialFileReader(std::move(file), fname, io_tracer_));
|
2014-10-30 01:43:37 +01:00
|
|
|
|
|
|
|
if (!status.ok()) {
|
|
|
|
return status;
|
|
|
|
}
|
|
|
|
|
|
|
|
LogReporter reporter;
|
|
|
|
reporter.env = env_;
|
|
|
|
reporter.info_log = db_options_.info_log.get();
|
|
|
|
reporter.fname = fname.c_str();
|
|
|
|
reporter.status = &status;
|
|
|
|
reporter.ignore_error = !db_options_.paranoid_checks;
|
2015-10-08 19:06:16 +02:00
|
|
|
log::Reader reader(db_options_.info_log, std::move(file_reader), &reporter,
|
2019-03-27 00:41:31 +01:00
|
|
|
true /*checksum*/, number);
|
2014-10-30 01:43:37 +01:00
|
|
|
std::string scratch;
|
|
|
|
Slice record;
|
|
|
|
|
|
|
|
if (reader.ReadRecord(&record, &scratch) &&
|
|
|
|
(status.ok() || !db_options_.paranoid_checks)) {
|
2016-03-30 19:35:22 +02:00
|
|
|
if (record.size() < WriteBatchInternal::kHeader) {
|
2014-10-30 01:43:37 +01:00
|
|
|
reporter.Corruption(record.size(),
|
|
|
|
Status::Corruption("log record too small"));
|
|
|
|
// TODO read record's till the first no corrupt entry?
|
|
|
|
} else {
|
|
|
|
WriteBatch batch;
|
2020-09-30 01:28:42 +02:00
|
|
|
// We can overwrite an existing non-OK Status since it'd only reach here
|
|
|
|
// with `paranoid_checks == false`.
|
|
|
|
status = WriteBatchInternal::SetContents(&batch, record);
|
|
|
|
if (status.ok()) {
|
|
|
|
*sequence = WriteBatchInternal::Sequence(&batch);
|
|
|
|
return status;
|
|
|
|
}
|
2014-10-30 01:43:37 +01:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2020-09-30 01:28:42 +02:00
|
|
|
// ReadRecord might have returned false on EOF, which means that the log file
|
|
|
|
// is empty. Or, a failure may have occurred while processing the first entry.
|
|
|
|
// In any case, return status and set sequence number to 0.
|
2014-10-30 01:43:37 +01:00
|
|
|
*sequence = 0;
|
|
|
|
return status;
|
|
|
|
}
|
|
|
|
|
|
|
|
#endif // ROCKSDB_LITE
|
2020-02-20 21:07:53 +01:00
|
|
|
} // namespace ROCKSDB_NAMESPACE
|