rocksdb/db/repair.cc

749 lines
27 KiB
C++
Raw Normal View History

// Copyright (c) 2011-present, Facebook, Inc. All rights reserved.
// This source code is licensed under both the GPLv2 (found in the
// COPYING file in the root directory) and Apache 2.0 License
// (found in the LICENSE.Apache file in the root directory).
//
// 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.
//
// Repairer does best effort recovery to recover as much data as possible after
// a disaster without compromising consistency. It does not guarantee bringing
// the database to a time consistent state.
//
// Repair process is broken into 4 phases:
// (a) Find files
// (b) Convert logs to tables
// (c) Extract metadata
// (d) Write Descriptor
//
// (a) Find files
//
// The repairer goes through all the files in the directory, and classifies them
// based on their file name. Any file that cannot be identified by name will be
// ignored.
//
// (b) Convert logs to table
//
// Every log file that is active is replayed. All sections of the file where the
// checksum does not match is skipped over. We intentionally give preference to
// data consistency.
//
// (c) Extract metadata
//
// We scan every table to compute
// (1) smallest/largest for the table
// (2) largest sequence number in the table
// (3) oldest blob file referred to by the table (if applicable)
//
// If we are unable to scan the file, then we ignore the table.
//
// (d) Write Descriptor
//
// We generate descriptor contents:
// - log number is set to zero
// - next-file-number is set to 1 + largest file number we found
// - last-sequence-number is set to largest sequence# found across
// all tables (see 2c)
// - compaction pointers are cleared
// - every table file is added at level 0
//
// Possible optimization 1:
// (a) Compute total size and use to pick appropriate max-level M
// (b) Sort tables by largest sequence# in the table
// (c) For each table: if it overlaps earlier table, place in level-0,
// else place in level-M.
// (d) We can provide options for time consistent recovery and unsafe recovery
// (ignore checksum failure when applicable)
// Possible optimization 2:
// Store per-table metadata (smallest, largest, largest-seq#, ...)
// in the table's meta section to speed up ScanTable.
#ifndef ROCKSDB_LITE
#include <cinttypes>
#include "db/builder.h"
#include "db/db_impl/db_impl.h"
#include "db/dbformat.h"
#include "db/log_reader.h"
#include "db/log_writer.h"
#include "db/memtable.h"
#include "db/table_cache.h"
#include "db/version_edit.h"
#include "db/write_batch_internal.h"
#include "file/filename.h"
#include "file/writable_file_writer.h"
#include "logging/logging.h"
#include "options/cf_options.h"
#include "rocksdb/comparator.h"
#include "rocksdb/db.h"
#include "rocksdb/env.h"
#include "rocksdb/options.h"
#include "rocksdb/write_buffer_manager.h"
#include "table/scoped_arena_iterator.h"
#include "util/string_util.h"
namespace ROCKSDB_NAMESPACE {
namespace {
class Repairer {
public:
Repairer(const std::string& dbname, const DBOptions& db_options,
const std::vector<ColumnFamilyDescriptor>& column_families,
const ColumnFamilyOptions& default_cf_opts,
const ColumnFamilyOptions& unknown_cf_opts, bool create_unknown_cfs)
: dbname_(dbname),
db_session_id_(DBImpl::GenerateDbSessionId(db_options.env)),
env_(db_options.env),
file_options_(),
db_options_(SanitizeOptions(dbname_, db_options)),
immutable_db_options_(ImmutableDBOptions(db_options_)),
icmp_(default_cf_opts.comparator),
default_cf_opts_(
SanitizeOptions(immutable_db_options_, default_cf_opts)),
default_iopts_(
ImmutableOptions(immutable_db_options_, default_cf_opts_)),
unknown_cf_opts_(
SanitizeOptions(immutable_db_options_, unknown_cf_opts)),
create_unknown_cfs_(create_unknown_cfs),
raw_table_cache_(
// TableCache can be small since we expect each table to be opened
// once.
NewLRUCache(10, db_options_.table_cache_numshardbits)),
table_cache_(new TableCache(default_iopts_, &file_options_,
raw_table_cache_.get(),
/*block_cache_tracer=*/nullptr,
/*io_tracer=*/nullptr, db_session_id_)),
wb_(db_options_.db_write_buffer_size),
wc_(db_options_.delayed_write_rate),
vset_(dbname_, &immutable_db_options_, file_options_,
raw_table_cache_.get(), &wb_, &wc_,
/*block_cache_tracer=*/nullptr, /*io_tracer=*/nullptr,
db_session_id_),
next_file_number_(1),
db_lock_(nullptr),
closed_(false) {
for (const auto& cfd : column_families) {
cf_name_to_opts_[cfd.name] = cfd.options;
}
}
const ColumnFamilyOptions* GetColumnFamilyOptions(
const std::string& cf_name) {
if (cf_name_to_opts_.find(cf_name) == cf_name_to_opts_.end()) {
if (create_unknown_cfs_) {
return &unknown_cf_opts_;
}
return nullptr;
}
return &cf_name_to_opts_[cf_name];
}
// Adds a column family to the VersionSet with cf_options_ and updates
// manifest.
Status AddColumnFamily(const std::string& cf_name, uint32_t cf_id) {
const auto* cf_opts = GetColumnFamilyOptions(cf_name);
if (cf_opts == nullptr) {
return Status::Corruption("Encountered unknown column family with name=" +
cf_name + ", id=" + ToString(cf_id));
}
Options opts(db_options_, *cf_opts);
MutableCFOptions mut_cf_opts(opts);
VersionEdit edit;
edit.SetComparatorName(opts.comparator->Name());
edit.SetLogNumber(0);
edit.SetColumnFamily(cf_id);
ColumnFamilyData* cfd;
cfd = nullptr;
edit.AddColumnFamily(cf_name);
mutex_.Lock();
Status status = vset_.LogAndApply(cfd, mut_cf_opts, &edit, &mutex_,
nullptr /* db_directory */,
false /* new_descriptor_log */, cf_opts);
mutex_.Unlock();
return status;
}
Status Close() {
Status s = Status::OK();
if (!closed_) {
if (db_lock_ != nullptr) {
s = env_->UnlockFile(db_lock_);
db_lock_ = nullptr;
}
closed_ = true;
}
return s;
}
~Repairer() { Close().PermitUncheckedError(); }
Status Run() {
Status status = env_->LockFile(LockFileName(dbname_), &db_lock_);
if (!status.ok()) {
return status;
}
status = FindFiles();
DBImpl* db_impl = nullptr;
if (status.ok()) {
// Discard older manifests and start a fresh one
for (size_t i = 0; i < manifests_.size(); i++) {
ArchiveFile(dbname_ + "/" + manifests_[i]);
}
// Just create a DBImpl temporarily so we can reuse NewDB()
db_impl = new DBImpl(db_options_, dbname_);
status = db_impl->NewDB(/*new_filenames=*/nullptr);
}
delete db_impl;
if (status.ok()) {
// Recover using the fresh manifest created by NewDB()
status =
vset_.Recover({{kDefaultColumnFamilyName, default_cf_opts_}}, false);
}
if (status.ok()) {
// Need to scan existing SST files first so the column families are
// created before we process WAL files
ExtractMetaData();
// ExtractMetaData() uses table_fds_ to know which SST files' metadata to
// extract -- we need to clear it here since metadata for existing SST
// files has been extracted already
table_fds_.clear();
ConvertLogFilesToTables();
ExtractMetaData();
status = AddTables();
}
if (status.ok()) {
uint64_t bytes = 0;
for (size_t i = 0; i < tables_.size(); i++) {
bytes += tables_[i].meta.fd.GetFileSize();
}
ROCKS_LOG_WARN(db_options_.info_log,
"**** Repaired rocksdb %s; "
"recovered %" ROCKSDB_PRIszt " files; %" PRIu64
" bytes. "
"Some data may have been lost. "
"****",
dbname_.c_str(), tables_.size(), bytes);
}
return status;
}
private:
struct TableInfo {
FileMetaData meta;
uint32_t column_family_id;
std::string column_family_name;
};
std::string const dbname_;
std::string db_session_id_;
Env* const env_;
const FileOptions file_options_;
const DBOptions db_options_;
const ImmutableDBOptions immutable_db_options_;
const InternalKeyComparator icmp_;
const ColumnFamilyOptions default_cf_opts_;
const ImmutableOptions default_iopts_; // table_cache_ holds reference
const ColumnFamilyOptions unknown_cf_opts_;
const bool create_unknown_cfs_;
std::shared_ptr<Cache> raw_table_cache_;
std::unique_ptr<TableCache> table_cache_;
WriteBufferManager wb_;
WriteController wc_;
VersionSet vset_;
std::unordered_map<std::string, ColumnFamilyOptions> cf_name_to_opts_;
InstrumentedMutex mutex_;
std::vector<std::string> manifests_;
std::vector<FileDescriptor> table_fds_;
std::vector<uint64_t> logs_;
std::vector<TableInfo> tables_;
uint64_t next_file_number_;
// Lock over the persistent DB state. Non-nullptr iff successfully
// acquired.
FileLock* db_lock_;
bool closed_;
Status FindFiles() {
std::vector<std::string> filenames;
bool found_file = false;
std::vector<std::string> to_search_paths;
for (size_t path_id = 0; path_id < db_options_.db_paths.size(); path_id++) {
to_search_paths.push_back(db_options_.db_paths[path_id].path);
}
// search wal_dir if user uses a customize wal_dir
bool same = immutable_db_options_.IsWalDirSameAsDBPath(dbname_);
if (!same) {
to_search_paths.push_back(immutable_db_options_.wal_dir);
}
for (size_t path_id = 0; path_id < to_search_paths.size(); path_id++) {
ROCKS_LOG_INFO(db_options_.info_log, "Searching path %s\n",
to_search_paths[path_id].c_str());
Status status = env_->GetChildren(to_search_paths[path_id], &filenames);
if (!status.ok()) {
return status;
}
if (!filenames.empty()) {
found_file = true;
}
uint64_t number;
FileType type;
for (size_t i = 0; i < filenames.size(); i++) {
if (ParseFileName(filenames[i], &number, &type)) {
if (type == kDescriptorFile) {
manifests_.push_back(filenames[i]);
} else {
if (number + 1 > next_file_number_) {
next_file_number_ = number + 1;
}
if (type == kWalFile) {
logs_.push_back(number);
} else if (type == kTableFile) {
table_fds_.emplace_back(number, static_cast<uint32_t>(path_id),
0);
} else {
// Ignore other files
}
}
}
}
}
if (!found_file) {
return Status::Corruption(dbname_, "repair found no files");
}
return Status::OK();
}
void ConvertLogFilesToTables() {
const auto& wal_dir = immutable_db_options_.GetWalDir();
for (size_t i = 0; i < logs_.size(); i++) {
// we should use LogFileName(wal_dir, logs_[i]) here. user might uses wal_dir option.
std::string logname = LogFileName(wal_dir, logs_[i]);
Status status = ConvertLogToTable(wal_dir, logs_[i]);
if (!status.ok()) {
ROCKS_LOG_WARN(db_options_.info_log,
"Log #%" PRIu64 ": ignoring conversion error: %s",
logs_[i], status.ToString().c_str());
}
ArchiveFile(logname);
}
}
Status ConvertLogToTable(const std::string& wal_dir, uint64_t log) {
struct LogReporter : public log::Reader::Reporter {
Env* env;
std::shared_ptr<Logger> info_log;
uint64_t lognum;
void Corruption(size_t bytes, const Status& s) override {
// We print error messages for corruption, but continue repairing.
ROCKS_LOG_ERROR(info_log, "Log #%" PRIu64 ": dropping %d bytes; %s",
lognum, static_cast<int>(bytes), s.ToString().c_str());
}
};
// Open the log file
std::string logname = LogFileName(wal_dir, log);
const auto& fs = env_->GetFileSystem();
std::unique_ptr<SequentialFileReader> lfile_reader;
Status status = SequentialFileReader::Create(
fs, logname, fs->OptimizeForLogRead(file_options_), &lfile_reader,
nullptr);
if (!status.ok()) {
return status;
}
// Create the log reader.
LogReporter reporter;
reporter.env = env_;
reporter.info_log = db_options_.info_log;
reporter.lognum = log;
// We intentionally make log::Reader do checksumming so that
// corruptions cause entire commits to be skipped instead of
// propagating bad information (like overly large sequence
// numbers).
log::Reader reader(db_options_.info_log, std::move(lfile_reader), &reporter,
Support for single-primary, multi-secondary instances (#4899) Summary: This PR allows RocksDB to run in single-primary, multi-secondary process mode. The writer is a regular RocksDB (e.g. an `DBImpl`) instance playing the role of a primary. Multiple `DBImplSecondary` processes (secondaries) share the same set of SST files, MANIFEST, WAL files with the primary. Secondaries tail the MANIFEST of the primary and apply updates to their own in-memory state of the file system, e.g. `VersionStorageInfo`. This PR has several components: 1. (Originally in #4745). Add a `PathNotFound` subcode to `IOError` to denote the failure when a secondary tries to open a file which has been deleted by the primary. 2. (Similar to #4602). Add `FragmentBufferedReader` to handle partially-read, trailing record at the end of a log from where future read can continue. 3. (Originally in #4710 and #4820). Add implementation of the secondary, i.e. `DBImplSecondary`. 3.1 Tail the primary's MANIFEST during recovery. 3.2 Tail the primary's MANIFEST during normal processing by calling `ReadAndApply`. 3.3 Tailing WAL will be in a future PR. 4. Add an example in 'examples/multi_processes_example.cc' to demonstrate the usage of secondary RocksDB instance in a multi-process setting. Instructions to run the example can be found at the beginning of the source code. Pull Request resolved: https://github.com/facebook/rocksdb/pull/4899 Differential Revision: D14510945 Pulled By: riversand963 fbshipit-source-id: 4ac1c5693e6012ad23f7b4b42d3c374fecbe8886
2019-03-27 00:41:31 +01:00
true /*enable checksum*/, log);
// Initialize per-column family memtables
for (auto* cfd : *vset_.GetColumnFamilySet()) {
cfd->CreateNewMemtable(*cfd->GetLatestMutableCFOptions(),
kMaxSequenceNumber);
}
auto cf_mems = new ColumnFamilyMemTablesImpl(vset_.GetColumnFamilySet());
// Read all the records and add to a memtable
std::string scratch;
Slice record;
WriteBatch batch;
int counter = 0;
while (reader.ReadRecord(&record, &scratch)) {
if (record.size() < WriteBatchInternal::kHeader) {
reporter.Corruption(
record.size(), Status::Corruption("log record too small"));
continue;
}
Status record_status = WriteBatchInternal::SetContents(&batch, record);
if (record_status.ok()) {
record_status =
WriteBatchInternal::InsertInto(&batch, cf_mems, nullptr, nullptr);
}
if (record_status.ok()) {
counter += WriteBatchInternal::Count(&batch);
} else {
ROCKS_LOG_WARN(db_options_.info_log, "Log #%" PRIu64 ": ignoring %s",
log, record_status.ToString().c_str());
}
}
// Dump a table for each column family with entries in this log file.
for (auto* cfd : *vset_.GetColumnFamilySet()) {
// Do not record a version edit for this conversion to a Table
// since ExtractMetaData() will also generate edits.
MemTable* mem = cfd->mem();
if (mem->IsEmpty()) {
continue;
}
FileMetaData meta;
meta.fd = FileDescriptor(next_file_number_++, 0, 0);
ReadOptions ro;
ro.total_order_seek = true;
Arena arena;
ScopedArenaIterator iter(mem->NewIterator(ro, &arena));
int64_t _current_time = 0;
immutable_db_options_.clock->GetCurrentTime(&_current_time)
.PermitUncheckedError(); // ignore error
FIFO Compaction with TTL Summary: Introducing FIFO compactions with TTL. FIFO compaction is based on size only which makes it tricky to enable in production as use cases can have organic growth. A user requested an option to drop files based on the time of their creation instead of the total size. To address that request: - Added a new TTL option to FIFO compaction options. - Updated FIFO compaction score to take TTL into consideration. - Added a new table property, creation_time, to keep track of when the SST file is created. - Creation_time is set as below: - On Flush: Set to the time of flush. - On Compaction: Set to the max creation_time of all the files involved in the compaction. - On Repair and Recovery: Set to the time of repair/recovery. - Old files created prior to this code change will have a creation_time of 0. - FIFO compaction with TTL is enabled when ttl > 0. All files older than ttl will be deleted during compaction. i.e. `if (file.creation_time < (current_time - ttl)) then delete(file)`. This will enable cases where you might want to delete all files older than, say, 1 day. - FIFO compaction will fall back to the prior way of deleting files based on size if: - the creation_time of all files involved in compaction is 0. - the total size (of all SST files combined) does not drop below `compaction_options_fifo.max_table_files_size` even if the files older than ttl are deleted. This feature is not supported if max_open_files != -1 or with table formats other than Block-based. **Test Plan:** Added tests. **Benchmark results:** Base: FIFO with max size: 100MB :: ``` svemuri@dev15905 ~/rocksdb (fifo-compaction) $ TEST_TMPDIR=/dev/shm ./db_bench --benchmarks=readwhilewriting --num=5000000 --threads=16 --compaction_style=2 --fifo_compaction_max_table_files_size_mb=100 readwhilewriting : 1.924 micros/op 519858 ops/sec; 13.6 MB/s (1176277 of 5000000 found) ``` With TTL (a low one for testing) :: ``` svemuri@dev15905 ~/rocksdb (fifo-compaction) $ TEST_TMPDIR=/dev/shm ./db_bench --benchmarks=readwhilewriting --num=5000000 --threads=16 --compaction_style=2 --fifo_compaction_max_table_files_size_mb=100 --fifo_compaction_ttl=20 readwhilewriting : 1.902 micros/op 525817 ops/sec; 13.7 MB/s (1185057 of 5000000 found) ``` Example Log lines: ``` 2017/06/26-15:17:24.609249 7fd5a45ff700 (Original Log Time 2017/06/26-15:17:24.609177) [db/compaction_picker.cc:1471] [default] FIFO compaction: picking file 40 with creation time 1498515423 for deletion 2017/06/26-15:17:24.609255 7fd5a45ff700 (Original Log Time 2017/06/26-15:17:24.609234) [db/db_impl_compaction_flush.cc:1541] [default] Deleted 1 files ... 2017/06/26-15:17:25.553185 7fd5a61a5800 [DEBUG] [db/db_impl_files.cc:309] [JOB 0] Delete /dev/shm/dbbench/000040.sst type=2 #40 -- OK 2017/06/26-15:17:25.553205 7fd5a61a5800 EVENT_LOG_v1 {"time_micros": 1498515445553199, "job": 0, "event": "table_file_deletion", "file_number": 40} ``` SST Files remaining in the dbbench dir, after db_bench execution completed: ``` svemuri@dev15905 ~/rocksdb (fifo-compaction) $ ls -l /dev/shm//dbbench/*.sst -rw-r--r--. 1 svemuri users 30749887 Jun 26 15:17 /dev/shm//dbbench/000042.sst -rw-r--r--. 1 svemuri users 30768779 Jun 26 15:17 /dev/shm//dbbench/000044.sst -rw-r--r--. 1 svemuri users 30757481 Jun 26 15:17 /dev/shm//dbbench/000046.sst ``` Closes https://github.com/facebook/rocksdb/pull/2480 Differential Revision: D5305116 Pulled By: sagar0 fbshipit-source-id: 3e5cfcf5dd07ed2211b5b37492eb235b45139174
2017-06-28 02:02:20 +02:00
const uint64_t current_time = static_cast<uint64_t>(_current_time);
SnapshotChecker* snapshot_checker = DisableGCSnapshotChecker::Instance();
auto write_hint = cfd->CalculateSSTWriteHint(0);
std::vector<std::unique_ptr<FragmentedRangeTombstoneIterator>>
range_del_iters;
auto range_del_iter =
mem->NewRangeTombstoneIterator(ro, kMaxSequenceNumber);
if (range_del_iter != nullptr) {
range_del_iters.emplace_back(range_del_iter);
}
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
IOStatus io_s;
CompressionOptions default_compression;
TableBuilderOptions tboptions(
*cfd->ioptions(), *cfd->GetLatestMutableCFOptions(),
cfd->internal_comparator(), cfd->int_tbl_prop_collector_factories(),
Add more LSM info to FilterBuildingContext (#8246) Summary: Add `num_levels`, `is_bottommost`, and table file creation `reason` to `FilterBuildingContext`, in anticipation of more powerful Bloom-like filter support. To support this, added `is_bottommost` and `reason` to `TableBuilderOptions`, which allowed removing `reason` parameter from `rocksdb::BuildTable`. I attempted to remove `skip_filters` from `TableBuilderOptions`, because filter construction decisions should arise from options, not one-off parameters. I could not completely remove it because the public API for SstFileWriter takes a `skip_filters` parameter, and translating this into an option change would mean awkwardly replacing the table_factory if it is BlockBasedTableFactory with new filter_policy=nullptr option. I marked this public skip_filters option as deprecated because of this oddity. (skip_filters on the read side probably makes sense.) At least `skip_filters` is now largely hidden for users of `TableBuilderOptions` and is no longer used for implementing the optimize_filters_for_hits option. Bringing the logic for that option closer to handling of FilterBuildingContext makes it more obvious that hese two are using the same notion of "bottommost." (Planned: configuration options for Bloom-like filters that generalize `optimize_filters_for_hits`) Recommended follow-up: Try to get away from "bottommost level" naming of things, which is inaccurate (see VersionStorageInfo::RangeMightExistAfterSortedRun), and move to "bottommost run" or just "bottommost." Pull Request resolved: https://github.com/facebook/rocksdb/pull/8246 Test Plan: extended an existing unit test to exercise and check various filter building contexts. Also, existing tests for optimize_filters_for_hits validate some of the "bottommost" handling, which is now closely connected to FilterBuildingContext::is_bottommost through TableBuilderOptions::is_bottommost Reviewed By: mrambacher Differential Revision: D28099346 Pulled By: pdillinger fbshipit-source-id: 2c1072e29c24d4ac404c761a7b7663292372600a
2021-04-30 22:49:24 +02:00
kNoCompression, default_compression, cfd->GetID(), cfd->GetName(),
-1 /* level */, false /* is_bottommost */,
TableFileCreationReason::kRecovery, current_time,
0 /* oldest_key_time */, 0 /* file_creation_time */,
"DB Repairer" /* db_id */, db_session_id_, 0 /*target_file_size*/,
meta.fd.GetNumber());
status = BuildTable(
dbname_, /* versions */ nullptr, immutable_db_options_, tboptions,
file_options_, table_cache_.get(), iter.get(),
std::move(range_del_iters), &meta, nullptr /* blob_file_additions */,
{}, kMaxSequenceNumber, snapshot_checker,
Add more LSM info to FilterBuildingContext (#8246) Summary: Add `num_levels`, `is_bottommost`, and table file creation `reason` to `FilterBuildingContext`, in anticipation of more powerful Bloom-like filter support. To support this, added `is_bottommost` and `reason` to `TableBuilderOptions`, which allowed removing `reason` parameter from `rocksdb::BuildTable`. I attempted to remove `skip_filters` from `TableBuilderOptions`, because filter construction decisions should arise from options, not one-off parameters. I could not completely remove it because the public API for SstFileWriter takes a `skip_filters` parameter, and translating this into an option change would mean awkwardly replacing the table_factory if it is BlockBasedTableFactory with new filter_policy=nullptr option. I marked this public skip_filters option as deprecated because of this oddity. (skip_filters on the read side probably makes sense.) At least `skip_filters` is now largely hidden for users of `TableBuilderOptions` and is no longer used for implementing the optimize_filters_for_hits option. Bringing the logic for that option closer to handling of FilterBuildingContext makes it more obvious that hese two are using the same notion of "bottommost." (Planned: configuration options for Bloom-like filters that generalize `optimize_filters_for_hits`) Recommended follow-up: Try to get away from "bottommost level" naming of things, which is inaccurate (see VersionStorageInfo::RangeMightExistAfterSortedRun), and move to "bottommost run" or just "bottommost." Pull Request resolved: https://github.com/facebook/rocksdb/pull/8246 Test Plan: extended an existing unit test to exercise and check various filter building contexts. Also, existing tests for optimize_filters_for_hits validate some of the "bottommost" handling, which is now closely connected to FilterBuildingContext::is_bottommost through TableBuilderOptions::is_bottommost Reviewed By: mrambacher Differential Revision: D28099346 Pulled By: pdillinger fbshipit-source-id: 2c1072e29c24d4ac404c761a7b7663292372600a
2021-04-30 22:49:24 +02:00
false /* paranoid_file_checks*/, nullptr /* internal_stats */, &io_s,
nullptr /*IOTracer*/, BlobFileCreationReason::kRecovery,
nullptr /* event_logger */, 0 /* job_id */, Env::IO_HIGH,
nullptr /* table_properties */, write_hint);
ROCKS_LOG_INFO(db_options_.info_log,
"Log #%" PRIu64 ": %d ops saved to Table #%" PRIu64 " %s",
log, counter, meta.fd.GetNumber(),
status.ToString().c_str());
if (status.ok()) {
if (meta.fd.GetFileSize() > 0) {
table_fds_.push_back(meta.fd);
}
} else {
break;
}
}
delete cf_mems;
return status;
}
void ExtractMetaData() {
for (size_t i = 0; i < table_fds_.size(); i++) {
TableInfo t;
t.meta.fd = table_fds_[i];
Status status = ScanTable(&t);
if (!status.ok()) {
std::string fname = TableFileName(
db_options_.db_paths, t.meta.fd.GetNumber(), t.meta.fd.GetPathId());
char file_num_buf[kFormatFileNumberBufSize];
FormatFileNumber(t.meta.fd.GetNumber(), t.meta.fd.GetPathId(),
file_num_buf, sizeof(file_num_buf));
ROCKS_LOG_WARN(db_options_.info_log, "Table #%s: ignoring %s",
file_num_buf, status.ToString().c_str());
ArchiveFile(fname);
} else {
tables_.push_back(t);
}
}
}
Status ScanTable(TableInfo* t) {
std::string fname = TableFileName(
db_options_.db_paths, t->meta.fd.GetNumber(), t->meta.fd.GetPathId());
int counter = 0;
uint64_t file_size;
Status status = env_->GetFileSize(fname, &file_size);
t->meta.fd = FileDescriptor(t->meta.fd.GetNumber(), t->meta.fd.GetPathId(),
file_size);
std::shared_ptr<const TableProperties> props;
if (status.ok()) {
status = table_cache_->GetTableProperties(file_options_, icmp_,
t->meta.fd, &props);
}
if (status.ok()) {
t->column_family_id = static_cast<uint32_t>(props->column_family_id);
if (t->column_family_id ==
TablePropertiesCollectorFactory::Context::kUnknownColumnFamily) {
ROCKS_LOG_WARN(
db_options_.info_log,
"Table #%" PRIu64
": column family unknown (probably due to legacy format); "
"adding to default column family id 0.",
t->meta.fd.GetNumber());
t->column_family_id = 0;
}
if (vset_.GetColumnFamilySet()->GetColumnFamily(t->column_family_id) ==
nullptr) {
status =
AddColumnFamily(props->column_family_name, t->column_family_id);
}
t->meta.oldest_ancester_time = props->creation_time;
}
ColumnFamilyData* cfd = nullptr;
if (status.ok()) {
cfd = vset_.GetColumnFamilySet()->GetColumnFamily(t->column_family_id);
if (cfd->GetName() != props->column_family_name) {
ROCKS_LOG_ERROR(
db_options_.info_log,
"Table #%" PRIu64
": inconsistent column family name '%s'; expected '%s' for column "
"family id %" PRIu32 ".",
t->meta.fd.GetNumber(), props->column_family_name.c_str(),
cfd->GetName().c_str(), t->column_family_id);
status = Status::Corruption(dbname_, "inconsistent column family name");
}
}
if (status.ok()) {
ReadOptions ropts;
ropts.total_order_seek = true;
InternalIterator* iter = table_cache_->NewIterator(
ropts, file_options_, cfd->internal_comparator(), t->meta,
nullptr /* range_del_agg */,
Fast path for detecting unchanged prefix_extractor (#9407) Summary: Fixes a major performance regression in 6.26, where extra CPU is spent in SliceTransform::AsString when reads involve a prefix_extractor (Get, MultiGet, Seek). Common case performance is now better than 6.25. This change creates a "fast path" for verifying that the current prefix extractor is unchanged and compatible with what was used to generate a table file. This fast path detects the common case by pointer comparison on the current prefix_extractor and a "known good" prefix extractor (if applicable) that is saved at the time the table reader is opened. The "known good" prefix extractor is saved as another shared_ptr copy (in an existing field, however) to ensure the pointer is not recycled. When the prefix_extractor has changed to a different instance but same compatible configuration (rare, odd), performance is still a regression compared to 6.25, but this is likely acceptable because of the oddity of such a case. The performance of incompatible prefix_extractor is essentially unchanged. Also fixed a minor case (ForwardIterator) where a prefix_extractor could be used via a raw pointer after being freed as a shared_ptr, if replaced via SetOptions. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9407 Test Plan: ## Performance Populate DB with `TEST_TMPDIR=/dev/shm/rocksdb ./db_bench -benchmarks=fillrandom -num=10000000 -disable_wal=1 -write_buffer_size=10000000 -bloom_bits=16 -compaction_style=2 -fifo_compaction_max_table_files_size_mb=10000 -fifo_compaction_allow_compaction=0 -prefix_size=12` Running head-to-head comparisons simultaneously with `TEST_TMPDIR=/dev/shm/rocksdb ./db_bench -use_existing_db -readonly -benchmarks=seekrandom -num=10000000 -duration=20 -disable_wal=1 -bloom_bits=16 -compaction_style=2 -fifo_compaction_max_table_files_size_mb=10000 -fifo_compaction_allow_compaction=0 -prefix_size=12` Below each is compared by ops/sec vs. baseline which is version 6.25 (multiple baseline runs because of variable machine load) v6.26: 4833 vs. 6698 (<- major regression!) v6.27: 4737 vs. 6397 (still) New: 6704 vs. 6461 (better than baseline in common case) Disabled fastpath: 4843 vs. 6389 (e.g. if prefix extractor instance changes but is still compatible) Changed prefix size (no usable filter) in new: 787 vs. 5927 Changed prefix size (no usable filter) in new & baseline: 773 vs. 784 Reviewed By: mrambacher Differential Revision: D33677812 Pulled By: pdillinger fbshipit-source-id: 571d9711c461fb97f957378a061b7e7dbc4d6a76
2022-01-21 20:36:36 +01:00
cfd->GetLatestMutableCFOptions()->prefix_extractor,
/*table_reader_ptr=*/nullptr, /*file_read_hist=*/nullptr,
TableReaderCaller::kRepair, /*arena=*/nullptr, /*skip_filters=*/false,
/*level=*/-1, /*max_file_size_for_l0_meta_pin=*/0,
/*smallest_compaction_key=*/nullptr,
Properly report IO errors when IndexType::kBinarySearchWithFirstKey is used (#6621) Summary: Context: Index type `kBinarySearchWithFirstKey` added the ability for sst file iterator to sometimes report a key from index without reading the corresponding data block. This is useful when sst blocks are cut at some meaningful boundaries (e.g. one block per key prefix), and many seeks land between blocks (e.g. for each prefix, the ranges of keys in different sst files are nearly disjoint, so a typical seek needs to read a data block from only one file even if all files have the prefix). But this added a new error condition, which rocksdb code was really not equipped to deal with: `InternalIterator::value()` may fail with an IO error or Status::Incomplete, but it's just a method returning a Slice, with no way to report error instead. Before this PR, this type of error wasn't handled at all (an empty slice was returned), and kBinarySearchWithFirstKey implementation was considered a prototype. Now that we (LogDevice) have experimented with kBinarySearchWithFirstKey for a while and confirmed that it's really useful, this PR is adding the missing error handling. It's a pretty inconvenient situation implementation-wise. The error needs to be reported from InternalIterator when trying to access value. But there are ~700 call sites of `InternalIterator::value()`, most of which either can't hit the error condition (because the iterator is reading from memtable or from index or something) or wouldn't benefit from the deferred loading of the value (e.g. compaction iterator that reads all values anyway). Adding error handling to all these call sites would needlessly bloat the code. So instead I made the deferred value loading optional: only the call sites that may use deferred loading have to call the new method `PrepareValue()` before calling `value()`. The feature is enabled with a new bool argument `allow_unprepared_value` to a bunch of methods that create iterators (it wouldn't make sense to put it in ReadOptions because it's completely internal to iterators, with virtually no user-visible effect). Lmk if you have better ideas. Note that the deferred value loading only happens for *internal* iterators. The user-visible iterator (DBIter) always prepares the value before returning from Seek/Next/etc. We could go further and add an API to defer that value loading too, but that's most likely not useful for LogDevice, so it doesn't seem worth the complexity for now. Pull Request resolved: https://github.com/facebook/rocksdb/pull/6621 Test Plan: make -j5 check . Will also deploy to some logdevice test clusters and look at stats. Reviewed By: siying Differential Revision: D20786930 Pulled By: al13n321 fbshipit-source-id: 6da77d918bad3780522e918f17f4d5513d3e99ee
2020-04-16 02:37:23 +02:00
/*largest_compaction_key=*/nullptr,
/*allow_unprepared_value=*/false);
ParsedInternalKey parsed;
for (iter->SeekToFirst(); iter->Valid(); iter->Next()) {
Slice key = iter->key();
Status pik_status =
ParseInternalKey(key, &parsed, db_options_.allow_data_in_errors);
if (!pik_status.ok()) {
ROCKS_LOG_ERROR(db_options_.info_log,
"Table #%" PRIu64 ": unparsable key - %s",
t->meta.fd.GetNumber(), pik_status.getState());
continue;
}
counter++;
t->meta.UpdateBoundaries(key, iter->value(), parsed.sequence,
parsed.type);
}
if (!iter->status().ok()) {
status = iter->status();
}
delete iter;
ROCKS_LOG_INFO(db_options_.info_log, "Table #%" PRIu64 ": %d entries %s",
t->meta.fd.GetNumber(), counter,
status.ToString().c_str());
}
if (status.ok()) {
// XXX/FIXME: This is just basic, naive handling of range tombstones,
// like call to UpdateBoundariesForRange in builder.cc where we assume
// an SST file is a full sorted run. This probably needs the extra logic
// from compaction_job.cc around call to UpdateBoundariesForRange (to
// handle range tombstones extendingg beyond range of other entries).
ReadOptions ropts;
std::unique_ptr<FragmentedRangeTombstoneIterator> r_iter;
status = table_cache_->GetRangeTombstoneIterator(
ropts, cfd->internal_comparator(), t->meta, &r_iter);
if (r_iter) {
r_iter->SeekToFirst();
while (r_iter->Valid()) {
auto tombstone = r_iter->Tombstone();
auto kv = tombstone.Serialize();
t->meta.UpdateBoundariesForRange(
kv.first, tombstone.SerializeEndKey(), tombstone.seq_,
cfd->internal_comparator());
r_iter->Next();
}
}
}
return status;
}
Status AddTables() {
std::unordered_map<uint32_t, std::vector<const TableInfo*>> cf_id_to_tables;
SequenceNumber max_sequence = 0;
for (size_t i = 0; i < tables_.size(); i++) {
cf_id_to_tables[tables_[i].column_family_id].push_back(&tables_[i]);
if (max_sequence < tables_[i].meta.fd.largest_seqno) {
max_sequence = tables_[i].meta.fd.largest_seqno;
}
}
vset_.SetLastAllocatedSequence(max_sequence);
vset_.SetLastPublishedSequence(max_sequence);
vset_.SetLastSequence(max_sequence);
for (const auto& cf_id_and_tables : cf_id_to_tables) {
auto* cfd =
vset_.GetColumnFamilySet()->GetColumnFamily(cf_id_and_tables.first);
VersionEdit edit;
edit.SetComparatorName(cfd->user_comparator()->Name());
edit.SetLogNumber(0);
edit.SetNextFile(next_file_number_);
edit.SetColumnFamily(cfd->GetID());
// TODO(opt): separate out into multiple levels
for (const auto* table : cf_id_and_tables.second) {
edit.AddFile(
0, table->meta.fd.GetNumber(), table->meta.fd.GetPathId(),
table->meta.fd.GetFileSize(), table->meta.smallest,
table->meta.largest, table->meta.fd.smallest_seqno,
table->meta.fd.largest_seqno, table->meta.marked_for_compaction,
table->meta.temperature, table->meta.oldest_blob_file_number,
table->meta.oldest_ancester_time, table->meta.file_creation_time,
table->meta.file_checksum, table->meta.file_checksum_func_name,
table->meta.min_timestamp, table->meta.max_timestamp);
}
assert(next_file_number_ > 0);
vset_.MarkFileNumberUsed(next_file_number_ - 1);
mutex_.Lock();
Status status = vset_.LogAndApply(
cfd, *cfd->GetLatestMutableCFOptions(), &edit, &mutex_,
nullptr /* db_directory */, false /* new_descriptor_log */);
mutex_.Unlock();
if (!status.ok()) {
return status;
}
}
return Status::OK();
}
void ArchiveFile(const std::string& fname) {
// Move into another directory. E.g., for
// dir/foo
// rename to
// dir/lost/foo
const char* slash = strrchr(fname.c_str(), '/');
std::string new_dir;
if (slash != nullptr) {
new_dir.assign(fname.data(), slash - fname.data());
}
new_dir.append("/lost");
env_->CreateDir(new_dir).PermitUncheckedError(); // Ignore error
std::string new_file = new_dir;
new_file.append("/");
new_file.append((slash == nullptr) ? fname.c_str() : slash + 1);
Status s = env_->RenameFile(fname, new_file);
ROCKS_LOG_INFO(db_options_.info_log, "Archiving %s: %s\n", fname.c_str(),
s.ToString().c_str());
}
};
Status GetDefaultCFOptions(
const std::vector<ColumnFamilyDescriptor>& column_families,
ColumnFamilyOptions* res) {
assert(res != nullptr);
auto iter = std::find_if(column_families.begin(), column_families.end(),
[](const ColumnFamilyDescriptor& cfd) {
return cfd.name == kDefaultColumnFamilyName;
});
if (iter == column_families.end()) {
return Status::InvalidArgument(
"column_families", "Must contain entry for default column family");
}
*res = iter->options;
return Status::OK();
}
} // anonymous namespace
Status RepairDB(const std::string& dbname, const DBOptions& db_options,
const std::vector<ColumnFamilyDescriptor>& column_families
) {
ColumnFamilyOptions default_cf_opts;
Status status = GetDefaultCFOptions(column_families, &default_cf_opts);
if (!status.ok()) {
return status;
}
Repairer repairer(dbname, db_options, column_families, default_cf_opts,
ColumnFamilyOptions() /* unknown_cf_opts */,
false /* create_unknown_cfs */);
status = repairer.Run();
if (status.ok()) {
status = repairer.Close();
}
return status;
}
Status RepairDB(const std::string& dbname, const DBOptions& db_options,
const std::vector<ColumnFamilyDescriptor>& column_families,
const ColumnFamilyOptions& unknown_cf_opts) {
ColumnFamilyOptions default_cf_opts;
Status status = GetDefaultCFOptions(column_families, &default_cf_opts);
if (!status.ok()) {
return status;
}
Repairer repairer(dbname, db_options, column_families, default_cf_opts,
unknown_cf_opts, true /* create_unknown_cfs */);
status = repairer.Run();
if (status.ok()) {
status = repairer.Close();
}
return status;
}
Status RepairDB(const std::string& dbname, const Options& options) {
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
Options opts(options);
DBOptions db_options(opts);
ColumnFamilyOptions cf_options(opts);
Repairer repairer(dbname, db_options,
{}, cf_options /* default_cf_opts */,
cf_options /* unknown_cf_opts */,
true /* create_unknown_cfs */);
Status status = repairer.Run();
if (status.ok()) {
status = repairer.Close();
}
return status;
}
} // namespace ROCKSDB_NAMESPACE
#endif // ROCKSDB_LITE