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

544 lines
20 KiB

// Copyright (c) 2013, Facebook, Inc. All rights reserved.
// This source code is licensed under the BSD-style license found in the
// LICENSE file in the root directory of this source tree. An additional grant
// of patent rights can be found in the PATENTS file in the same directory.
//
// Copyright (c) 2011 The LevelDB Authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file. See the AUTHORS file for names of contributors.
#pragma once
#include <atomic>
#include <deque>
#include <set>
#include <utility>
#include <vector>
#include "db/dbformat.h"
#include "db/log_writer.h"
#include "db/snapshot.h"
#include "db/version_edit.h"
#include "memtable_list.h"
#include "port/port.h"
#include "rocksdb/db.h"
#include "rocksdb/env.h"
#include "rocksdb/memtablerep.h"
#include "rocksdb/transaction_log.h"
#include "util/autovector.h"
#include "util/stats_logger.h"
#include "db/internal_stats.h"
namespace rocksdb {
class MemTable;
class TableCache;
class Version;
class VersionEdit;
class VersionSet;
class DBImpl : public DB {
public:
DBImpl(const Options& options, const std::string& dbname);
virtual ~DBImpl();
// Implementations of the DB interface
virtual Status Put(const WriteOptions&, const Slice& key, const Slice& value);
virtual Status Merge(const WriteOptions&, const Slice& key,
const Slice& value);
virtual Status Delete(const WriteOptions&, const Slice& key);
virtual Status Write(const WriteOptions& options, WriteBatch* updates);
virtual Status Get(const ReadOptions& options,
const Slice& key,
std::string* value);
virtual std::vector<Status> MultiGet(const ReadOptions& options,
const std::vector<Slice>& keys,
std::vector<std::string>* values);
// Returns false if key doesn't exist in the database and true if it may.
// If value_found is not passed in as null, then return the value if found in
// memory. On return, if value was found, then value_found will be set to true
// , otherwise false.
virtual bool KeyMayExist(const ReadOptions& options,
const Slice& key,
std::string* value,
bool* value_found = nullptr);
virtual Iterator* NewIterator(const ReadOptions&);
virtual const Snapshot* GetSnapshot();
virtual void ReleaseSnapshot(const Snapshot* snapshot);
virtual bool GetProperty(const Slice& property, std::string* value);
virtual void GetApproximateSizes(const Range* range, int n, uint64_t* sizes);
virtual Status CompactRange(const Slice* begin, const Slice* end,
bool reduce_level = false, int target_level = -1);
virtual int NumberLevels();
virtual int MaxMemCompactionLevel();
virtual int Level0StopWriteTrigger();
[RocksDB] BackupableDB Summary: In this diff I present you BackupableDB v1. You can easily use it to backup your DB and it will do incremental snapshots for you. Let's first describe how you would use BackupableDB. It's inheriting StackableDB interface so you can easily construct it with your DB object -- it will add a method RollTheSnapshot() to the DB object. When you call RollTheSnapshot(), current snapshot of the DB will be stored in the backup dir. To restore, you can just call RestoreDBFromBackup() on a BackupableDB (which is a static method) and it will restore all files from the backup dir. In the next version, it will even support automatic backuping every X minutes. There are multiple things you can configure: 1. backup_env and db_env can be different, which is awesome because then you can easily backup to HDFS or wherever you feel like. 2. sync - if true, it *guarantees* backup consistency on machine reboot 3. number of snapshots to keep - this will keep last N snapshots around if you want, for some reason, be able to restore from an earlier snapshot. All the backuping is done in incremental fashion - if we already have 00010.sst, we will not copy it again. *IMPORTANT* -- This is based on assumption that 00010.sst never changes - two files named 00010.sst from the same DB will always be exactly the same. Is this true? I always copy manifest, current and log files. 4. You can decide if you want to flush the memtables before you backup, or you're fine with backing up the log files -- either way, you get a complete and consistent view of the database at a time of backup. 5. More things you can find in BackupableDBOptions Here is the directory structure I use: backup_dir/CURRENT_SNAPSHOT - just 4 bytes holding the latest snapshot 0, 1, 2, ... - files containing serialized version of each snapshot - containing a list of files files/*.sst - sst files shared between snapshots - if one snapshot references 00010.sst and another one needs to backup it from the DB, it will just reference the same file files/ 0/, 1/, 2/, ... - snapshot directories containing private snapshot files - current, manifest and log files All the files are ref counted and deleted immediatelly when they get out of scope. Some other stuff in this diff: 1. Added GetEnv() method to the DB. Discussed with @haobo and we agreed that it seems right thing to do. 2. Fixed StackableDB interface. The way it was set up before, I was not able to implement BackupableDB. Test Plan: I have a unittest, but please don't look at this yet. I just hacked it up to help me with debugging. I will write a lot of good tests and update the diff. Also, `make asan_check` Reviewers: dhruba, haobo, emayanke Reviewed By: dhruba CC: leveldb, haobo Differential Revision: https://reviews.facebook.net/D14295
11 years ago
virtual const std::string& GetName() const;
virtual Env* GetEnv() const;
virtual const Options& GetOptions() const;
virtual Status Flush(const FlushOptions& options);
virtual Status DisableFileDeletions();
virtual Status EnableFileDeletions(bool force);
// All the returned filenames start with "/"
virtual Status GetLiveFiles(std::vector<std::string>&,
uint64_t* manifest_file_size,
bool flush_memtable = true);
virtual Status GetSortedWalFiles(VectorLogPtr& files);
virtual SequenceNumber GetLatestSequenceNumber() const;
virtual Status GetUpdatesSince(SequenceNumber seq_number,
unique_ptr<TransactionLogIterator>* iter);
virtual Status DeleteFile(std::string name);
virtual void GetLiveFilesMetaData(
std::vector<LiveFileMetaData> *metadata);
virtual Status GetDbIdentity(std::string& identity);
Status RunManualCompaction(int input_level,
int output_level,
const Slice* begin,
const Slice* end);
// Extra methods (for testing) that are not in the public DB interface
// Compact any files in the named level that overlap [*begin, *end]
Status TEST_CompactRange(int level,
const Slice* begin,
const Slice* end);
// Force current memtable contents to be flushed.
Status TEST_FlushMemTable();
// Wait for memtable compaction
Status TEST_WaitForFlushMemTable();
// Wait for any compaction
Status TEST_WaitForCompact();
// Return an internal iterator over the current state of the database.
// The keys of this iterator are internal keys (see format.h).
// The returned iterator should be deleted when no longer needed.
Iterator* TEST_NewInternalIterator();
// Return the maximum overlapping data (in bytes) at next level for any
// file at a level >= 1.
int64_t TEST_MaxNextLevelOverlappingBytes();
// Simulate a db crash, no elegant closing of database.
void TEST_Destroy_DBImpl();
// Return the current manifest file no.
uint64_t TEST_Current_Manifest_FileNo();
// Trigger's a background call for testing.
void TEST_PurgeObsoleteteWAL();
// get total level0 file size. Only for testing.
uint64_t TEST_GetLevel0TotalSize();
void TEST_SetDefaultTimeToCheck(uint64_t default_interval_to_delete_obsolete_WAL)
{
default_interval_to_delete_obsolete_WAL_ = default_interval_to_delete_obsolete_WAL;
}
// holds references to memtable, all immutable memtables and version
struct SuperVersion {
MemTable* mem;
MemTableListVersion Summary: MemTableListVersion is to MemTableList what Version is to VersionSet. I took almost the same ideas to develop MemTableListVersion. The reason is to have copying std::list done in background, while flushing, rather than in foreground (MultiGet() and NewIterator()) under a mutex! Also, whenever we copied MemTableList, we copied also some MemTableList metadata (flush_requested_, commit_in_progress_, etc.), which was wasteful. This diff avoids std::list copy under a mutex in both MultiGet() and NewIterator(). I created a small database with some number of immutable memtables, and creating 100.000 iterators in a single-thread (!) decreased from {188739, 215703, 198028} to {154352, 164035, 159817}. A lot of the savings come from code under a mutex, so we should see much higher savings with multiple threads. Creating new iterator is very important to LogDevice team. I also think this diff will make SuperVersion obsolete for performance reasons. I will try it in the next diff. SuperVersion gave us huge savings on Get() code path, but I think that most of the savings came from copying MemTableList under a mutex. If we had MemTableListVersion, we would never need to copy the entire object (like we still do in NewIterator() and MultiGet()) Test Plan: `make check` works. I will also do `make valgrind_check` before commit Reviewers: dhruba, haobo, kailiu, sdong, emayanke, tnovak Reviewed By: kailiu CC: leveldb Differential Revision: https://reviews.facebook.net/D15255
11 years ago
MemTableListVersion* imm;
Version* current;
std::atomic<uint32_t> refs;
MemTableListVersion Summary: MemTableListVersion is to MemTableList what Version is to VersionSet. I took almost the same ideas to develop MemTableListVersion. The reason is to have copying std::list done in background, while flushing, rather than in foreground (MultiGet() and NewIterator()) under a mutex! Also, whenever we copied MemTableList, we copied also some MemTableList metadata (flush_requested_, commit_in_progress_, etc.), which was wasteful. This diff avoids std::list copy under a mutex in both MultiGet() and NewIterator(). I created a small database with some number of immutable memtables, and creating 100.000 iterators in a single-thread (!) decreased from {188739, 215703, 198028} to {154352, 164035, 159817}. A lot of the savings come from code under a mutex, so we should see much higher savings with multiple threads. Creating new iterator is very important to LogDevice team. I also think this diff will make SuperVersion obsolete for performance reasons. I will try it in the next diff. SuperVersion gave us huge savings on Get() code path, but I think that most of the savings came from copying MemTableList under a mutex. If we had MemTableListVersion, we would never need to copy the entire object (like we still do in NewIterator() and MultiGet()) Test Plan: `make check` works. I will also do `make valgrind_check` before commit Reviewers: dhruba, haobo, kailiu, sdong, emayanke, tnovak Reviewed By: kailiu CC: leveldb Differential Revision: https://reviews.facebook.net/D15255
11 years ago
// We need to_delete because during Cleanup(), imm->Unref() returns
// all memtables that we need to free through this vector. We then
// delete all those memtables outside of mutex, during destruction
autovector<MemTable*> to_delete;
// should be called outside the mutex
SuperVersion() = default;
~SuperVersion();
SuperVersion* Ref();
// Returns true if this was the last reference and caller should
// call Clenaup() and delete the object
bool Unref();
// call these two methods with db mutex held
// Cleanup unrefs mem, imm and current. Also, it stores all memtables
// that needs to be deleted in to_delete vector. Unrefing those
// objects needs to be done in the mutex
void Cleanup();
MemTableListVersion Summary: MemTableListVersion is to MemTableList what Version is to VersionSet. I took almost the same ideas to develop MemTableListVersion. The reason is to have copying std::list done in background, while flushing, rather than in foreground (MultiGet() and NewIterator()) under a mutex! Also, whenever we copied MemTableList, we copied also some MemTableList metadata (flush_requested_, commit_in_progress_, etc.), which was wasteful. This diff avoids std::list copy under a mutex in both MultiGet() and NewIterator(). I created a small database with some number of immutable memtables, and creating 100.000 iterators in a single-thread (!) decreased from {188739, 215703, 198028} to {154352, 164035, 159817}. A lot of the savings come from code under a mutex, so we should see much higher savings with multiple threads. Creating new iterator is very important to LogDevice team. I also think this diff will make SuperVersion obsolete for performance reasons. I will try it in the next diff. SuperVersion gave us huge savings on Get() code path, but I think that most of the savings came from copying MemTableList under a mutex. If we had MemTableListVersion, we would never need to copy the entire object (like we still do in NewIterator() and MultiGet()) Test Plan: `make check` works. I will also do `make valgrind_check` before commit Reviewers: dhruba, haobo, kailiu, sdong, emayanke, tnovak Reviewed By: kailiu CC: leveldb Differential Revision: https://reviews.facebook.net/D15255
11 years ago
void Init(MemTable* new_mem, MemTableListVersion* new_imm,
Version* new_current);
};
// needed for CleanupIteratorState
struct DeletionState {
inline bool HaveSomethingToDelete() const {
return candidate_files.size() ||
sst_delete_files.size() ||
log_delete_files.size();
}
// a list of all files that we'll consider deleting
// (every once in a while this is filled up with all files
// in the DB directory)
std::vector<std::string> candidate_files;
// the list of all live sst files that cannot be deleted
std::vector<uint64_t> sst_live;
// a list of sst files that we need to delete
std::vector<FileMetaData*> sst_delete_files;
// a list of log files that we need to delete
std::vector<uint64_t> log_delete_files;
// a list of memtables to be free
autovector<MemTable*> memtables_to_free;
SuperVersion* superversion_to_free; // if nullptr nothing to free
SuperVersion* new_superversion; // if nullptr no new superversion
// the current manifest_file_number, log_number and prev_log_number
// that corresponds to the set of files in 'live'.
uint64_t manifest_file_number, log_number, prev_log_number;
explicit DeletionState(bool create_superversion = false) {
manifest_file_number = 0;
log_number = 0;
prev_log_number = 0;
superversion_to_free = nullptr;
new_superversion =
create_superversion ? new SuperVersion() : nullptr;
}
~DeletionState() {
// free pending memtables
for (auto m : memtables_to_free) {
delete m;
}
// free superversion. if nullptr, this will be noop
delete superversion_to_free;
// if new_superversion was not used, it will be non-nullptr and needs
// to be freed here
delete new_superversion;
}
};
// Returns the list of live files in 'live' and the list
// of all files in the filesystem in 'candidate_files'.
// If force == false and the last call was less than
// options_.delete_obsolete_files_period_micros microseconds ago,
// it will not fill up the deletion_state
void FindObsoleteFiles(DeletionState& deletion_state,
bool force,
bool no_full_scan = false);
// Diffs the files listed in filenames and those that do not
// belong to live files are posibly removed. Also, removes all the
// files in sst_delete_files and log_delete_files.
// It is not necessary to hold the mutex when invoking this method.
void PurgeObsoleteFiles(DeletionState& deletion_state);
protected:
Env* const env_;
const std::string dbname_;
unique_ptr<VersionSet> versions_;
const InternalKeyComparator internal_comparator_;
const Options options_; // options_.comparator == &internal_comparator_
const Comparator* user_comparator() const {
return internal_comparator_.user_comparator();
}
SuperVersion* GetSuperVersion() {
return super_version_;
}
Iterator* NewInternalIterator(const ReadOptions&,
SequenceNumber* latest_snapshot);
private:
friend class DB;
friend class TailingIterator;
struct CompactionState;
struct Writer;
Status NewDB();
// Recover the descriptor from persistent storage. May do a significant
Refactor Recover() code Summary: This diff does two things: * Rethinks how we call Recover() with read_only option. Before, we call it with pointer to memtable where we'd like to apply those changes to. This memtable is set in db_impl_readonly.cc and it's actually DBImpl::mem_. Why don't we just apply updates to mem_ right away? It seems more intuitive. * Changes when we apply updates to manifest. Before, the process is to recover all the logs, flush it to sst files and then do one giant commit that atomically adds all recovered sst files and sets the next log number. This works good enough, but causes some small troubles for my column family approach, since I can't have one VersionEdit apply to more than single column family[1]. The change here is to commit the files recovered from logs right away. Here is the state of the world before the change: 1. Recover log 5, add new sst files to edit 2. Recover log 7, add new sst files to edit 3. Recover log 8, add new sst files to edit 4. Commit all added sst files to manifest and mark log files 5, 7 and 8 as recoverd (via SetLogNumber(9) function) After the change, we'll do: 1. Recover log 5, commit the new sst files and set log 5 as recovered 2. Recover log 7, commit the new sst files and set log 7 as recovered 3. Recover log 8, commit the new sst files and set log 8 as recovered The added (small) benefit is that if we fail after (2), the new recovery will only have to recover log 8. In previous case, we'll have to restart the recovery from the beginning. The bigger benefit will be to enable easier integration of multiple column families in Recovery code path. [1] I'm happy to dicuss this decison, but I believe this is the cleanest way to go. It also makes backward compatibility much easier. We don't have a requirement of adding multiple column families atomically. Test Plan: make check Reviewers: dhruba, haobo, kailiu, sdong Reviewed By: kailiu CC: leveldb Differential Revision: https://reviews.facebook.net/D15237
11 years ago
// amount of work to recover recently logged updates.
Status Recover(bool read_only = false, bool error_if_log_file_exist = false);
void MaybeIgnoreError(Status* s) const;
const Status CreateArchivalDirectory();
// Delete any unneeded files and stale in-memory entries.
void DeleteObsoleteFiles();
// Flush the in-memory write buffer to storage. Switches to a new
// log-file/memtable and writes a new descriptor iff successful.
Status FlushMemTableToOutputFile(bool* madeProgress,
DeletionState& deletion_state);
Refactor Recover() code Summary: This diff does two things: * Rethinks how we call Recover() with read_only option. Before, we call it with pointer to memtable where we'd like to apply those changes to. This memtable is set in db_impl_readonly.cc and it's actually DBImpl::mem_. Why don't we just apply updates to mem_ right away? It seems more intuitive. * Changes when we apply updates to manifest. Before, the process is to recover all the logs, flush it to sst files and then do one giant commit that atomically adds all recovered sst files and sets the next log number. This works good enough, but causes some small troubles for my column family approach, since I can't have one VersionEdit apply to more than single column family[1]. The change here is to commit the files recovered from logs right away. Here is the state of the world before the change: 1. Recover log 5, add new sst files to edit 2. Recover log 7, add new sst files to edit 3. Recover log 8, add new sst files to edit 4. Commit all added sst files to manifest and mark log files 5, 7 and 8 as recoverd (via SetLogNumber(9) function) After the change, we'll do: 1. Recover log 5, commit the new sst files and set log 5 as recovered 2. Recover log 7, commit the new sst files and set log 7 as recovered 3. Recover log 8, commit the new sst files and set log 8 as recovered The added (small) benefit is that if we fail after (2), the new recovery will only have to recover log 8. In previous case, we'll have to restart the recovery from the beginning. The bigger benefit will be to enable easier integration of multiple column families in Recovery code path. [1] I'm happy to dicuss this decison, but I believe this is the cleanest way to go. It also makes backward compatibility much easier. We don't have a requirement of adding multiple column families atomically. Test Plan: make check Reviewers: dhruba, haobo, kailiu, sdong Reviewed By: kailiu CC: leveldb Differential Revision: https://reviews.facebook.net/D15237
11 years ago
Status RecoverLogFile(uint64_t log_number, SequenceNumber* max_sequence,
bool read_only);
// The following two methods are used to flush a memtable to
// storage. The first one is used atdatabase RecoveryTime (when the
// database is opened) and is heavyweight because it holds the mutex
// for the entire period. The second method WriteLevel0Table supports
// concurrent flush memtables to storage.
Status WriteLevel0TableForRecovery(MemTable* mem, VersionEdit* edit);
Status WriteLevel0Table(autovector<MemTable*>& mems, VersionEdit* edit,
uint64_t* filenumber);
uint64_t SlowdownAmount(int n, double bottom, double top);
// MakeRoomForWrite will return superversion_to_free through an arugment,
// which the caller needs to delete. We do it because caller can delete
// the superversion outside of mutex
Status MakeRoomForWrite(bool force /* compact even if there is room? */,
SuperVersion** superversion_to_free);
void BuildBatchGroup(Writer** last_writer,
autovector<WriteBatch*>* write_batch_group);
// Force current memtable contents to be flushed.
Status FlushMemTable(const FlushOptions& options);
// Wait for memtable flushed
Status WaitForFlushMemTable();
void MaybeScheduleLogDBDeployStats();
static void BGLogDBDeployStats(void* db);
void LogDBDeployStats();
void MaybeScheduleFlushOrCompaction();
static void BGWorkCompaction(void* db);
static void BGWorkFlush(void* db);
void BackgroundCallCompaction();
void BackgroundCallFlush();
Status BackgroundCompaction(bool* madeProgress,DeletionState& deletion_state);
Status BackgroundFlush(bool* madeProgress, DeletionState& deletion_state);
void CleanupCompaction(CompactionState* compact, Status status);
Status DoCompactionWork(CompactionState* compact,
DeletionState& deletion_state);
Status OpenCompactionOutputFile(CompactionState* compact);
Status FinishCompactionOutputFile(CompactionState* compact, Iterator* input);
Status InstallCompactionResults(CompactionState* compact);
void AllocateCompactionOutputFileNumbers(CompactionState* compact);
void ReleaseCompactionUnusedFileNumbers(CompactionState* compact);
void PurgeObsoleteWALFiles();
Status AppendSortedWalsOfType(const std::string& path,
VectorLogPtr& log_files,
WalFileType type);
// Requires: all_logs should be sorted with earliest log file first
// Retains all log files in all_logs which contain updates with seq no.
// Greater Than or Equal to the requested SequenceNumber.
Status RetainProbableWalFiles(VectorLogPtr& all_logs,
const SequenceNumber target);
// return true if
bool CheckWalFileExistsAndEmpty(const WalFileType type,
const uint64_t number);
Status ReadFirstRecord(const WalFileType type, const uint64_t number,
WriteBatch* const result);
Status ReadFirstLine(const std::string& fname, WriteBatch* const batch);
void PrintStatistics();
// dump rocksdb.stats to LOG
void MaybeDumpStats();
// Return the minimum empty level that could hold the total data in the
// input level. Return the input level, if such level could not be found.
int FindMinimumEmptyLevelFitting(int level);
// Move the files in the input level to the target level.
// If target_level < 0, automatically calculate the minimum level that could
// hold the data set.
Status ReFitLevel(int level, int target_level = -1);
// Returns the current SuperVersion number.
uint64_t CurrentVersionNumber() const;
// Returns a pair of iterators (mutable-only and immutable-only) used
// internally by TailingIterator and stores CurrentVersionNumber() in
// *superversion_number. These iterators are always up-to-date, i.e. can
// be used to read new data.
std::pair<Iterator*, Iterator*> GetTailingIteratorPair(
const ReadOptions& options,
uint64_t* superversion_number);
// Constant after construction
const InternalFilterPolicy internal_filter_policy_;
bool owns_info_log_;
// table_cache_ provides its own synchronization
unique_ptr<TableCache> table_cache_;
// Lock over the persistent DB state. Non-nullptr iff successfully acquired.
FileLock* db_lock_;
// State below is protected by mutex_
port::Mutex mutex_;
port::AtomicPointer shutting_down_;
port::CondVar bg_cv_; // Signalled when background work finishes
MemTable* mem_;
MemTableList imm_; // Memtable that are not changing
uint64_t logfile_number_;
unique_ptr<log::Writer> log_;
SuperVersion* super_version_;
// An ordinal representing the current SuperVersion. Updated by
// InstallSuperVersion(), i.e. incremented every time super_version_
// changes.
std::atomic<uint64_t> super_version_number_;
std::string host_name_;
std::unique_ptr<Directory> db_directory_;
// Queue of writers.
std::deque<Writer*> writers_;
WriteBatch tmp_batch_;
SnapshotList snapshots_;
// Set of table files to protect from deletion because they are
// part of ongoing compactions.
std::set<uint64_t> pending_outputs_;
Fix a deadlock in CompactRange() Summary: The way DBImpl::TEST_CompactRange() throttles down the number of bg compactions can cause it to deadlock when CompactRange() is called concurrently from multiple threads. Imagine a following scenario with only two threads (max_background_compactions is 10 and bg_compaction_scheduled_ is initially 0): 1. Thread #1 increments bg_compaction_scheduled_ (to LargeNumber), sets bg_compaction_scheduled_ to 9 (newvalue), schedules the compaction (bg_compaction_scheduled_ is now 10) and waits for it to complete. 2. Thread #2 calls TEST_CompactRange(), increments bg_compaction_scheduled_ (now LargeNumber + 10) and waits on a cv for bg_compaction_scheduled_ to drop to LargeNumber. 3. BG thread completes the first manual compaction, decrements bg_compaction_scheduled_ and wakes up all threads waiting on bg_cv_. Thread #1 runs, increments bg_compaction_scheduled_ by LargeNumber again (now 2*LargeNumber + 9). Since that's more than LargeNumber + newvalue, thread #2 also goes to sleep (waiting on bg_cv_), without resetting bg_compaction_scheduled_. This diff attempts to address the problem by introducing a new counter bg_manual_only_ (when positive, MaybeScheduleFlushOrCompaction() will only schedule manual compactions). Test Plan: I could pretty much consistently reproduce the deadlock with a program that calls CompactRange(nullptr, nullptr) immediately after Write() from multiple threads. This no longer happens with this patch. Tests (make check) pass. Reviewers: dhruba, igor, sdong, haobo Reviewed By: igor CC: leveldb Differential Revision: https://reviews.facebook.net/D14799
11 years ago
// count how many background compactions are running or have been scheduled
int bg_compaction_scheduled_;
Fix a deadlock in CompactRange() Summary: The way DBImpl::TEST_CompactRange() throttles down the number of bg compactions can cause it to deadlock when CompactRange() is called concurrently from multiple threads. Imagine a following scenario with only two threads (max_background_compactions is 10 and bg_compaction_scheduled_ is initially 0): 1. Thread #1 increments bg_compaction_scheduled_ (to LargeNumber), sets bg_compaction_scheduled_ to 9 (newvalue), schedules the compaction (bg_compaction_scheduled_ is now 10) and waits for it to complete. 2. Thread #2 calls TEST_CompactRange(), increments bg_compaction_scheduled_ (now LargeNumber + 10) and waits on a cv for bg_compaction_scheduled_ to drop to LargeNumber. 3. BG thread completes the first manual compaction, decrements bg_compaction_scheduled_ and wakes up all threads waiting on bg_cv_. Thread #1 runs, increments bg_compaction_scheduled_ by LargeNumber again (now 2*LargeNumber + 9). Since that's more than LargeNumber + newvalue, thread #2 also goes to sleep (waiting on bg_cv_), without resetting bg_compaction_scheduled_. This diff attempts to address the problem by introducing a new counter bg_manual_only_ (when positive, MaybeScheduleFlushOrCompaction() will only schedule manual compactions). Test Plan: I could pretty much consistently reproduce the deadlock with a program that calls CompactRange(nullptr, nullptr) immediately after Write() from multiple threads. This no longer happens with this patch. Tests (make check) pass. Reviewers: dhruba, igor, sdong, haobo Reviewed By: igor CC: leveldb Differential Revision: https://reviews.facebook.net/D14799
11 years ago
// If non-zero, MaybeScheduleFlushOrCompaction() will only schedule manual
// compactions (if manual_compaction_ is not null). This mechanism enables
// manual compactions to wait until all other compactions are finished.
int bg_manual_only_;
// number of background memtable flush jobs, submitted to the HIGH pool
int bg_flush_scheduled_;
// Has a background stats log thread scheduled?
bool bg_logstats_scheduled_;
// Information for a manual compaction
struct ManualCompaction {
int input_level;
int output_level;
bool done;
Status status;
bool in_progress; // compaction request being processed?
const InternalKey* begin; // nullptr means beginning of key range
const InternalKey* end; // nullptr means end of key range
InternalKey tmp_storage; // Used to keep track of compaction progress
};
ManualCompaction* manual_compaction_;
// Have we encountered a background error in paranoid mode?
Status bg_error_;
std::unique_ptr<StatsLogger> logger_;
int64_t volatile last_log_ts;
// shall we disable deletion of obsolete files
// if 0 the deletion is enabled.
// if non-zero, files will not be getting deleted
// This enables two different threads to call
// EnableFileDeletions() and DisableFileDeletions()
// without any synchronization
int disable_delete_obsolete_files_;
// last time when DeleteObsoleteFiles was invoked
uint64_t delete_obsolete_files_last_run_;
// last time when PurgeObsoleteWALFiles ran.
uint64_t purge_wal_files_last_run_;
// last time stats were dumped to LOG
std::atomic<uint64_t> last_stats_dump_time_microsec_;
// obsolete files will be deleted every this seconds if ttl deletion is
// enabled and archive size_limit is disabled.
uint64_t default_interval_to_delete_obsolete_WAL_;
bool flush_on_destroy_; // Used when disableWAL is true.
InternalStats internal_stats_;
Improve output for GetProperty('leveldb.stats') Summary: Display separate values for read, write & total compaction IO. Display compaction amplification and write amplification. Add similar values for the period since the last call to GetProperty. Results since the server started are reported as "cumulative" stats. Results since the last call to GetProperty are reported as "interval" stats. Level Files Size(MB) Time(sec) Read(MB) Write(MB) Rn(MB) Rnp1(MB) Wnew(MB) Amplify Read(MB/s) Write(MB/s) Rn Rnp1 Wnp1 NewW Count Ln-stall ---------------------------------------------------------------------------------------------------------------------------------------------------------------------- 0 7 13 21 0 211 0 0 211 0.0 0.0 10.1 0 0 0 0 113 0.0 1 79 157 88 993 989 198 795 194 9.0 11.3 11.2 106 405 502 97 14 0.0 2 19 36 5 63 63 37 27 36 2.4 12.3 12.2 19 14 32 18 12 0.0 >>>>>>>>>>>>>>>>>>>>>>>>> text below has been is new and/or reformatted Uptime(secs): 122.2 total, 0.9 interval Compaction IO cumulative (GB): 0.21 new, 1.03 read, 1.23 write, 2.26 read+write Compaction IO cumulative (MB/sec): 1.7 new, 8.6 read, 10.3 write, 19.0 read+write Amplification cumulative: 6.0 write, 11.0 compaction Compaction IO interval (MB): 5.59 new, 0.00 read, 5.59 write, 5.59 read+write Compaction IO interval (MB/sec): 6.5 new, 0.0 read, 6.5 write, 6.5 read+write Amplification interval: 1.0 write, 1.0 compaction >>>>>>>>>>>>>>>>>>>>>>>> text above is new and/or reformatted Stalls(secs): 90.574 level0_slowdown, 0.000 level0_numfiles, 10.165 memtable_compaction, 0.000 leveln_slowdown Task ID: # Blame Rev: Test Plan: make check, run db_bench Revert Plan: Database Impact: Memcache Impact: Other Notes: EImportant: - begin *PUBLIC* platform impact section - Bugzilla: # - end platform impact - Reviewers: haobo Reviewed By: haobo CC: leveldb Differential Revision: https://reviews.facebook.net/D11049
12 years ago
static const int KEEP_LOG_FILE_NUM = 1000;
std::string db_absolute_path_;
// count of the number of contiguous delaying writes
int delayed_writes_;
// The options to access storage files
const EnvOptions storage_options_;
// A value of true temporarily disables scheduling of background work
bool bg_work_gate_closed_;
// Guard against multiple concurrent refitting
bool refitting_level_;
// No copying allowed
DBImpl(const DBImpl&);
void operator=(const DBImpl&);
// dump the delayed_writes_ to the log file and reset counter.
void DelayLoggingAndReset();
// Return the earliest snapshot where seqno is visible.
// Store the snapshot right before that, if any, in prev_snapshot
inline SequenceNumber findEarliestVisibleSnapshot(
SequenceNumber in,
std::vector<SequenceNumber>& snapshots,
SequenceNumber* prev_snapshot);
// will return a pointer to SuperVersion* if previous SuperVersion
// if its reference count is zero and needs deletion or nullptr if not
// As argument takes a pointer to allocated SuperVersion
// Foreground threads call this function directly (they don't carry
// deletion state and have to handle their own creation and deletion
// of SuperVersion)
SuperVersion* InstallSuperVersion(SuperVersion* new_superversion);
// Background threads call this function, which is just a wrapper around
// the InstallSuperVersion() function above. Background threads carry
// deletion_state which can have new_superversion already allocated.
void InstallSuperVersion(DeletionState& deletion_state);
// Function that Get and KeyMayExist call with no_io true or false
// Note: 'value_found' from KeyMayExist propagates here
Status GetImpl(const ReadOptions& options,
const Slice& key,
std::string* value,
bool* value_found = nullptr);
};
// Sanitize db options. The caller should delete result.info_log if
// it is not equal to src.info_log.
extern Options SanitizeOptions(const std::string& db,
const InternalKeyComparator* icmp,
const InternalFilterPolicy* ipolicy,
const Options& src);
// Determine compression type, based on user options, level of the output
// file and whether compression is disabled.
// If enable_compression is false, then compression is always disabled no
// matter what the values of the other two parameters are.
// Otherwise, the compression type is determined based on options and level.
CompressionType GetCompressionType(const Options& options, int level,
const bool enable_compression);
// Determine compression type for L0 file written by memtable flush.
CompressionType GetCompressionFlush(const Options& options);
} // namespace rocksdb