rocksdb/env/env.cc

1264 lines
44 KiB
C++
Raw Permalink 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.
#include "rocksdb/env.h"
#include <thread>
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
#include "env/composite_env_wrapper.h"
#include "env/emulated_clock.h"
#include "env/mock_env.h"
Experimental support for SST unique IDs (#8990) Summary: * New public header unique_id.h and function GetUniqueIdFromTableProperties which computes a universally unique identifier based on table properties of table files from recent RocksDB versions. * Generation of DB session IDs is refactored so that they are guaranteed unique in the lifetime of a process running RocksDB. (SemiStructuredUniqueIdGen, new test included.) Along with file numbers, this enables SST unique IDs to be guaranteed unique among SSTs generated in a single process, and "better than random" between processes. See https://github.com/pdillinger/unique_id * In addition to public API producing 'external' unique IDs, there is a function for producing 'internal' unique IDs, with functions for converting between the two. In short, the external ID is "safe" for things people might do with it, and the internal ID enables more "power user" features for the future. Specifically, the external ID goes through a hashing layer so that any subset of bits in the external ID can be used as a hash of the full ID, while also preserving uniqueness guarantees in the first 128 bits (bijective both on first 128 bits and on full 192 bits). Intended follow-up: * Use the internal unique IDs in cache keys. (Avoid conflicts with https://github.com/facebook/rocksdb/issues/8912) (The file offset can be XORed into the third 64-bit value of the unique ID.) * Publish the external unique IDs in FileStorageInfo (https://github.com/facebook/rocksdb/issues/8968) Pull Request resolved: https://github.com/facebook/rocksdb/pull/8990 Test Plan: Unit tests added, and checking of unique ids in stress test. NOTE in stress test we do not generate nearly enough files to thoroughly stress uniqueness, but the test trims off pieces of the ID to check for uniqueness so that we can infer (with some assumptions) stronger properties in the aggregate. Reviewed By: zhichao-cao, mrambacher Differential Revision: D31582865 Pulled By: pdillinger fbshipit-source-id: 1f620c4c86af9abe2a8d177b9ccf2ad2b9f48243
2021-10-19 08:28:28 +02:00
#include "env/unique_id_gen.h"
#include "logging/env_logger.h"
#include "memory/arena.h"
#include "options/db_options.h"
#include "port/port.h"
#include "rocksdb/convenience.h"
#include "rocksdb/options.h"
#include "rocksdb/system_clock.h"
#include "rocksdb/utilities/customizable_util.h"
#include "rocksdb/utilities/object_registry.h"
#include "rocksdb/utilities/options_type.h"
#include "util/autovector.h"
namespace ROCKSDB_NAMESPACE {
Create a CustomEnv class; Add WinFileSystem; Make LegacyFileSystemWrapper private (#7703) Summary: This PR does the following: -> Creates a WinFileSystem class. This class is the Windows equivalent of the PosixFileSystem and will be used on Windows systems. -> Introduces a CustomEnv class. A CustomEnv is an Env that takes a FileSystem as constructor argument. I believe there will only ever be two implementations of this class (PosixEnv and WinEnv). There is still a CustomEnvWrapper class that takes an Env and a FileSystem and wraps the Env calls with the input Env but uses the FileSystem for the FileSystem calls -> Eliminates the public uses of the LegacyFileSystemWrapper. With this change in place, there are effectively the following patterns of Env: - "Base Env classes" (PosixEnv, WinEnv). These classes implement the core Env functions (e.g. Threads) and have a hard-coded input FileSystem. These classes inherit from CompositeEnv, implement the core Env functions (threads) and delegate the FileSystem-like calls to the input file system. - Wrapped Composite Env classes (MemEnv). These classes take in an Env and a FileSystem. The core env functions are re-directed to the wrapped env. The file system calls are redirected to the input file system - Legacy Wrapped Env classes. These classes take in an Env input (but no FileSystem). The core env functions are re-directed to the wrapped env. A "Legacy File System" is created using this env and the file system calls directed to the env itself. With these changes in place, the PosixEnv becomes a singleton -- there is only ever one created. Any other use of the PosixEnv is via another wrapped env. This cleans up some of the issues with the env construction and destruction. Additionally, there were places in the code that required had an Env when they required a FileSystem. Many of these places would wrap the Env with a LegacyFileSystemWrapper instead of using the env->GetFileSystem(). These places were changed, thereby removing layers of additional redirection (LegacyFileSystem --> Env --> Env::FileSystem). Pull Request resolved: https://github.com/facebook/rocksdb/pull/7703 Reviewed By: zhichao-cao Differential Revision: D25762190 Pulled By: anand1976 fbshipit-source-id: 1a088e97fc916f28ac69c149cd1dcad0ab31704b
2021-01-06 19:48:24 +01:00
namespace {
#ifndef ROCKSDB_LITE
static int RegisterBuiltinEnvs(ObjectLibrary& library,
const std::string& /*arg*/) {
library.AddFactory<Env>(MockEnv::kClassName(), [](const std::string& /*uri*/,
std::unique_ptr<Env>* guard,
std::string* /* errmsg */) {
guard->reset(MockEnv::Create(Env::Default()));
return guard->get();
});
library.AddFactory<Env>(
CompositeEnvWrapper::kClassName(),
[](const std::string& /*uri*/, std::unique_ptr<Env>* guard,
std::string* /* errmsg */) {
guard->reset(new CompositeEnvWrapper(Env::Default()));
return guard->get();
});
size_t num_types;
return static_cast<int>(library.GetFactoryCount(&num_types));
}
#endif // ROCKSDB_LITE
static void RegisterSystemEnvs() {
#ifndef ROCKSDB_LITE
static std::once_flag loaded;
std::call_once(loaded, [&]() {
RegisterBuiltinEnvs(*(ObjectLibrary::Default().get()), "");
});
#endif // ROCKSDB_LITE
}
class LegacySystemClock : public SystemClock {
private:
Env* env_;
public:
explicit LegacySystemClock(Env* env) : env_(env) {}
const char* Name() const override { return "LegacySystemClock"; }
// Returns the number of micro-seconds since some fixed point in time.
// It is often used as system time such as in GenericRateLimiter
// and other places so a port needs to return system time in order to work.
uint64_t NowMicros() override { return env_->NowMicros(); }
// Returns the number of nano-seconds since some fixed point in time. Only
// useful for computing deltas of time in one run.
// Default implementation simply relies on NowMicros.
// In platform-specific implementations, NowNanos() should return time points
// that are MONOTONIC.
uint64_t NowNanos() override { return env_->NowNanos(); }
uint64_t CPUMicros() override { return CPUNanos() / 1000; }
uint64_t CPUNanos() override { return env_->NowCPUNanos(); }
// Sleep/delay the thread for the prescribed number of micro-seconds.
void SleepForMicroseconds(int micros) override {
env_->SleepForMicroseconds(micros);
}
// Get the number of seconds since the Epoch, 1970-01-01 00:00:00 (UTC).
// Only overwrites *unix_time on success.
Status GetCurrentTime(int64_t* unix_time) override {
return env_->GetCurrentTime(unix_time);
}
// Converts seconds-since-Jan-01-1970 to a printable string
std::string TimeToString(uint64_t time) override {
return env_->TimeToString(time);
}
#ifndef ROCKSDB_LITE
std::string SerializeOptions(const ConfigOptions& /*config_options*/,
const std::string& /*prefix*/) const override {
// We do not want the LegacySystemClock to appear in the serialized output.
// This clock is an internal class for those who do not implement one and
// would be part of the Env. As such, do not serialize it here.
return "";
}
#endif // ROCKSDB_LITE
};
class LegacySequentialFileWrapper : public FSSequentialFile {
public:
explicit LegacySequentialFileWrapper(
std::unique_ptr<SequentialFile>&& _target)
: target_(std::move(_target)) {}
IOStatus Read(size_t n, const IOOptions& /*options*/, Slice* result,
char* scratch, IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->Read(n, result, scratch));
}
IOStatus Skip(uint64_t n) override {
return status_to_io_status(target_->Skip(n));
}
bool use_direct_io() const override { return target_->use_direct_io(); }
size_t GetRequiredBufferAlignment() const override {
return target_->GetRequiredBufferAlignment();
}
IOStatus InvalidateCache(size_t offset, size_t length) override {
return status_to_io_status(target_->InvalidateCache(offset, length));
}
IOStatus PositionedRead(uint64_t offset, size_t n,
const IOOptions& /*options*/, Slice* result,
char* scratch, IODebugContext* /*dbg*/) override {
return status_to_io_status(
target_->PositionedRead(offset, n, result, scratch));
}
private:
std::unique_ptr<SequentialFile> target_;
};
class LegacyRandomAccessFileWrapper : public FSRandomAccessFile {
public:
explicit LegacyRandomAccessFileWrapper(
std::unique_ptr<RandomAccessFile>&& target)
: target_(std::move(target)) {}
IOStatus Read(uint64_t offset, size_t n, const IOOptions& /*options*/,
Slice* result, char* scratch,
IODebugContext* /*dbg*/) const override {
return status_to_io_status(target_->Read(offset, n, result, scratch));
}
IOStatus MultiRead(FSReadRequest* fs_reqs, size_t num_reqs,
const IOOptions& /*options*/,
IODebugContext* /*dbg*/) override {
std::vector<ReadRequest> reqs;
Status status;
reqs.reserve(num_reqs);
for (size_t i = 0; i < num_reqs; ++i) {
ReadRequest req;
req.offset = fs_reqs[i].offset;
req.len = fs_reqs[i].len;
req.scratch = fs_reqs[i].scratch;
req.status = Status::OK();
reqs.emplace_back(req);
}
status = target_->MultiRead(reqs.data(), num_reqs);
for (size_t i = 0; i < num_reqs; ++i) {
fs_reqs[i].result = reqs[i].result;
fs_reqs[i].status = status_to_io_status(std::move(reqs[i].status));
}
return status_to_io_status(std::move(status));
}
IOStatus Prefetch(uint64_t offset, size_t n, const IOOptions& /*options*/,
IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->Prefetch(offset, n));
}
size_t GetUniqueId(char* id, size_t max_size) const override {
return target_->GetUniqueId(id, max_size);
}
void Hint(AccessPattern pattern) override {
target_->Hint((RandomAccessFile::AccessPattern)pattern);
}
bool use_direct_io() const override { return target_->use_direct_io(); }
size_t GetRequiredBufferAlignment() const override {
return target_->GetRequiredBufferAlignment();
}
IOStatus InvalidateCache(size_t offset, size_t length) override {
return status_to_io_status(target_->InvalidateCache(offset, length));
}
private:
std::unique_ptr<RandomAccessFile> target_;
};
class LegacyRandomRWFileWrapper : public FSRandomRWFile {
public:
explicit LegacyRandomRWFileWrapper(std::unique_ptr<RandomRWFile>&& target)
: target_(std::move(target)) {}
bool use_direct_io() const override { return target_->use_direct_io(); }
size_t GetRequiredBufferAlignment() const override {
return target_->GetRequiredBufferAlignment();
}
IOStatus Write(uint64_t offset, const Slice& data,
const IOOptions& /*options*/,
IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->Write(offset, data));
}
IOStatus Read(uint64_t offset, size_t n, const IOOptions& /*options*/,
Slice* result, char* scratch,
IODebugContext* /*dbg*/) const override {
return status_to_io_status(target_->Read(offset, n, result, scratch));
}
IOStatus Flush(const IOOptions& /*options*/,
IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->Flush());
}
IOStatus Sync(const IOOptions& /*options*/,
IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->Sync());
}
IOStatus Fsync(const IOOptions& /*options*/,
IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->Fsync());
}
IOStatus Close(const IOOptions& /*options*/,
IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->Close());
}
private:
std::unique_ptr<RandomRWFile> target_;
};
class LegacyWritableFileWrapper : public FSWritableFile {
public:
explicit LegacyWritableFileWrapper(std::unique_ptr<WritableFile>&& _target)
: target_(std::move(_target)) {}
IOStatus Append(const Slice& data, const IOOptions& /*options*/,
IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->Append(data));
}
IOStatus Append(const Slice& data, const IOOptions& /*options*/,
const DataVerificationInfo& /*verification_info*/,
IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->Append(data));
}
IOStatus PositionedAppend(const Slice& data, uint64_t offset,
const IOOptions& /*options*/,
IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->PositionedAppend(data, offset));
}
IOStatus PositionedAppend(const Slice& data, uint64_t offset,
const IOOptions& /*options*/,
const DataVerificationInfo& /*verification_info*/,
IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->PositionedAppend(data, offset));
}
IOStatus Truncate(uint64_t size, const IOOptions& /*options*/,
IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->Truncate(size));
}
IOStatus Close(const IOOptions& /*options*/,
IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->Close());
}
IOStatus Flush(const IOOptions& /*options*/,
IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->Flush());
}
IOStatus Sync(const IOOptions& /*options*/,
IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->Sync());
}
IOStatus Fsync(const IOOptions& /*options*/,
IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->Fsync());
}
bool IsSyncThreadSafe() const override { return target_->IsSyncThreadSafe(); }
bool use_direct_io() const override { return target_->use_direct_io(); }
size_t GetRequiredBufferAlignment() const override {
return target_->GetRequiredBufferAlignment();
}
void SetWriteLifeTimeHint(Env::WriteLifeTimeHint hint) override {
target_->SetWriteLifeTimeHint(hint);
}
Env::WriteLifeTimeHint GetWriteLifeTimeHint() override {
return target_->GetWriteLifeTimeHint();
}
uint64_t GetFileSize(const IOOptions& /*options*/,
IODebugContext* /*dbg*/) override {
return target_->GetFileSize();
}
void SetPreallocationBlockSize(size_t size) override {
target_->SetPreallocationBlockSize(size);
}
void GetPreallocationStatus(size_t* block_size,
size_t* last_allocated_block) override {
target_->GetPreallocationStatus(block_size, last_allocated_block);
}
size_t GetUniqueId(char* id, size_t max_size) const override {
return target_->GetUniqueId(id, max_size);
}
IOStatus InvalidateCache(size_t offset, size_t length) override {
return status_to_io_status(target_->InvalidateCache(offset, length));
}
IOStatus RangeSync(uint64_t offset, uint64_t nbytes,
const IOOptions& /*options*/,
IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->RangeSync(offset, nbytes));
}
void PrepareWrite(size_t offset, size_t len, const IOOptions& /*options*/,
IODebugContext* /*dbg*/) override {
target_->PrepareWrite(offset, len);
}
IOStatus Allocate(uint64_t offset, uint64_t len, const IOOptions& /*options*/,
IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->Allocate(offset, len));
}
private:
std::unique_ptr<WritableFile> target_;
};
class LegacyDirectoryWrapper : public FSDirectory {
public:
explicit LegacyDirectoryWrapper(std::unique_ptr<Directory>&& target)
: target_(std::move(target)) {}
IOStatus Fsync(const IOOptions& /*options*/,
IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->Fsync());
}
size_t GetUniqueId(char* id, size_t max_size) const override {
return target_->GetUniqueId(id, max_size);
}
private:
std::unique_ptr<Directory> target_;
};
Create a CustomEnv class; Add WinFileSystem; Make LegacyFileSystemWrapper private (#7703) Summary: This PR does the following: -> Creates a WinFileSystem class. This class is the Windows equivalent of the PosixFileSystem and will be used on Windows systems. -> Introduces a CustomEnv class. A CustomEnv is an Env that takes a FileSystem as constructor argument. I believe there will only ever be two implementations of this class (PosixEnv and WinEnv). There is still a CustomEnvWrapper class that takes an Env and a FileSystem and wraps the Env calls with the input Env but uses the FileSystem for the FileSystem calls -> Eliminates the public uses of the LegacyFileSystemWrapper. With this change in place, there are effectively the following patterns of Env: - "Base Env classes" (PosixEnv, WinEnv). These classes implement the core Env functions (e.g. Threads) and have a hard-coded input FileSystem. These classes inherit from CompositeEnv, implement the core Env functions (threads) and delegate the FileSystem-like calls to the input file system. - Wrapped Composite Env classes (MemEnv). These classes take in an Env and a FileSystem. The core env functions are re-directed to the wrapped env. The file system calls are redirected to the input file system - Legacy Wrapped Env classes. These classes take in an Env input (but no FileSystem). The core env functions are re-directed to the wrapped env. A "Legacy File System" is created using this env and the file system calls directed to the env itself. With these changes in place, the PosixEnv becomes a singleton -- there is only ever one created. Any other use of the PosixEnv is via another wrapped env. This cleans up some of the issues with the env construction and destruction. Additionally, there were places in the code that required had an Env when they required a FileSystem. Many of these places would wrap the Env with a LegacyFileSystemWrapper instead of using the env->GetFileSystem(). These places were changed, thereby removing layers of additional redirection (LegacyFileSystem --> Env --> Env::FileSystem). Pull Request resolved: https://github.com/facebook/rocksdb/pull/7703 Reviewed By: zhichao-cao Differential Revision: D25762190 Pulled By: anand1976 fbshipit-source-id: 1a088e97fc916f28ac69c149cd1dcad0ab31704b
2021-01-06 19:48:24 +01:00
class LegacyFileSystemWrapper : public FileSystem {
public:
// Initialize an EnvWrapper that delegates all calls to *t
explicit LegacyFileSystemWrapper(Env* t) : target_(t) {}
~LegacyFileSystemWrapper() override {}
static const char* kClassName() { return "LegacyFileSystem"; }
const char* Name() const override { return kClassName(); }
Create a CustomEnv class; Add WinFileSystem; Make LegacyFileSystemWrapper private (#7703) Summary: This PR does the following: -> Creates a WinFileSystem class. This class is the Windows equivalent of the PosixFileSystem and will be used on Windows systems. -> Introduces a CustomEnv class. A CustomEnv is an Env that takes a FileSystem as constructor argument. I believe there will only ever be two implementations of this class (PosixEnv and WinEnv). There is still a CustomEnvWrapper class that takes an Env and a FileSystem and wraps the Env calls with the input Env but uses the FileSystem for the FileSystem calls -> Eliminates the public uses of the LegacyFileSystemWrapper. With this change in place, there are effectively the following patterns of Env: - "Base Env classes" (PosixEnv, WinEnv). These classes implement the core Env functions (e.g. Threads) and have a hard-coded input FileSystem. These classes inherit from CompositeEnv, implement the core Env functions (threads) and delegate the FileSystem-like calls to the input file system. - Wrapped Composite Env classes (MemEnv). These classes take in an Env and a FileSystem. The core env functions are re-directed to the wrapped env. The file system calls are redirected to the input file system - Legacy Wrapped Env classes. These classes take in an Env input (but no FileSystem). The core env functions are re-directed to the wrapped env. A "Legacy File System" is created using this env and the file system calls directed to the env itself. With these changes in place, the PosixEnv becomes a singleton -- there is only ever one created. Any other use of the PosixEnv is via another wrapped env. This cleans up some of the issues with the env construction and destruction. Additionally, there were places in the code that required had an Env when they required a FileSystem. Many of these places would wrap the Env with a LegacyFileSystemWrapper instead of using the env->GetFileSystem(). These places were changed, thereby removing layers of additional redirection (LegacyFileSystem --> Env --> Env::FileSystem). Pull Request resolved: https://github.com/facebook/rocksdb/pull/7703 Reviewed By: zhichao-cao Differential Revision: D25762190 Pulled By: anand1976 fbshipit-source-id: 1a088e97fc916f28ac69c149cd1dcad0ab31704b
2021-01-06 19:48:24 +01:00
// Return the target to which this Env forwards all calls
Env* target() const { return target_; }
// The following text is boilerplate that forwards all methods to target()
IOStatus NewSequentialFile(const std::string& f, const FileOptions& file_opts,
std::unique_ptr<FSSequentialFile>* r,
IODebugContext* /*dbg*/) override {
std::unique_ptr<SequentialFile> file;
Status s = target_->NewSequentialFile(f, &file, file_opts);
if (s.ok()) {
r->reset(new LegacySequentialFileWrapper(std::move(file)));
}
return status_to_io_status(std::move(s));
}
IOStatus NewRandomAccessFile(const std::string& f,
const FileOptions& file_opts,
std::unique_ptr<FSRandomAccessFile>* r,
IODebugContext* /*dbg*/) override {
std::unique_ptr<RandomAccessFile> file;
Status s = target_->NewRandomAccessFile(f, &file, file_opts);
if (s.ok()) {
r->reset(new LegacyRandomAccessFileWrapper(std::move(file)));
}
return status_to_io_status(std::move(s));
}
IOStatus NewWritableFile(const std::string& f, const FileOptions& file_opts,
std::unique_ptr<FSWritableFile>* r,
IODebugContext* /*dbg*/) override {
std::unique_ptr<WritableFile> file;
Status s = target_->NewWritableFile(f, &file, file_opts);
if (s.ok()) {
r->reset(new LegacyWritableFileWrapper(std::move(file)));
}
return status_to_io_status(std::move(s));
}
IOStatus ReopenWritableFile(const std::string& fname,
const FileOptions& file_opts,
std::unique_ptr<FSWritableFile>* result,
IODebugContext* /*dbg*/) override {
std::unique_ptr<WritableFile> file;
Status s = target_->ReopenWritableFile(fname, &file, file_opts);
if (s.ok()) {
result->reset(new LegacyWritableFileWrapper(std::move(file)));
}
return status_to_io_status(std::move(s));
}
IOStatus ReuseWritableFile(const std::string& fname,
const std::string& old_fname,
const FileOptions& file_opts,
std::unique_ptr<FSWritableFile>* r,
IODebugContext* /*dbg*/) override {
std::unique_ptr<WritableFile> file;
Status s = target_->ReuseWritableFile(fname, old_fname, &file, file_opts);
if (s.ok()) {
r->reset(new LegacyWritableFileWrapper(std::move(file)));
}
return status_to_io_status(std::move(s));
}
IOStatus NewRandomRWFile(const std::string& fname,
const FileOptions& file_opts,
std::unique_ptr<FSRandomRWFile>* result,
IODebugContext* /*dbg*/) override {
std::unique_ptr<RandomRWFile> file;
Status s = target_->NewRandomRWFile(fname, &file, file_opts);
if (s.ok()) {
result->reset(new LegacyRandomRWFileWrapper(std::move(file)));
}
return status_to_io_status(std::move(s));
}
IOStatus NewMemoryMappedFileBuffer(
const std::string& fname,
std::unique_ptr<MemoryMappedFileBuffer>* result) override {
return status_to_io_status(
target_->NewMemoryMappedFileBuffer(fname, result));
}
IOStatus NewDirectory(const std::string& name, const IOOptions& /*io_opts*/,
std::unique_ptr<FSDirectory>* result,
IODebugContext* /*dbg*/) override {
std::unique_ptr<Directory> dir;
Status s = target_->NewDirectory(name, &dir);
if (s.ok()) {
result->reset(new LegacyDirectoryWrapper(std::move(dir)));
}
return status_to_io_status(std::move(s));
}
IOStatus FileExists(const std::string& f, const IOOptions& /*io_opts*/,
IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->FileExists(f));
}
IOStatus GetChildren(const std::string& dir, const IOOptions& /*io_opts*/,
std::vector<std::string>* r,
IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->GetChildren(dir, r));
}
IOStatus GetChildrenFileAttributes(const std::string& dir,
const IOOptions& /*options*/,
std::vector<FileAttributes>* result,
IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->GetChildrenFileAttributes(dir, result));
}
IOStatus DeleteFile(const std::string& f, const IOOptions& /*options*/,
IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->DeleteFile(f));
}
IOStatus Truncate(const std::string& fname, size_t size,
const IOOptions& /*options*/,
IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->Truncate(fname, size));
}
IOStatus CreateDir(const std::string& d, const IOOptions& /*options*/,
IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->CreateDir(d));
}
IOStatus CreateDirIfMissing(const std::string& d,
const IOOptions& /*options*/,
IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->CreateDirIfMissing(d));
}
IOStatus DeleteDir(const std::string& d, const IOOptions& /*options*/,
IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->DeleteDir(d));
}
IOStatus GetFileSize(const std::string& f, const IOOptions& /*options*/,
uint64_t* s, IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->GetFileSize(f, s));
}
IOStatus GetFileModificationTime(const std::string& fname,
const IOOptions& /*options*/,
uint64_t* file_mtime,
IODebugContext* /*dbg*/) override {
return status_to_io_status(
target_->GetFileModificationTime(fname, file_mtime));
}
IOStatus GetAbsolutePath(const std::string& db_path,
const IOOptions& /*options*/,
std::string* output_path,
IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->GetAbsolutePath(db_path, output_path));
}
IOStatus RenameFile(const std::string& s, const std::string& t,
const IOOptions& /*options*/,
IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->RenameFile(s, t));
}
IOStatus LinkFile(const std::string& s, const std::string& t,
const IOOptions& /*options*/,
IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->LinkFile(s, t));
}
IOStatus NumFileLinks(const std::string& fname, const IOOptions& /*options*/,
uint64_t* count, IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->NumFileLinks(fname, count));
}
IOStatus AreFilesSame(const std::string& first, const std::string& second,
const IOOptions& /*options*/, bool* res,
IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->AreFilesSame(first, second, res));
}
IOStatus LockFile(const std::string& f, const IOOptions& /*options*/,
FileLock** l, IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->LockFile(f, l));
}
IOStatus UnlockFile(FileLock* l, const IOOptions& /*options*/,
IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->UnlockFile(l));
}
IOStatus GetTestDirectory(const IOOptions& /*options*/, std::string* path,
IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->GetTestDirectory(path));
}
IOStatus NewLogger(const std::string& fname, const IOOptions& /*options*/,
std::shared_ptr<Logger>* result,
IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->NewLogger(fname, result));
}
void SanitizeFileOptions(FileOptions* opts) const override {
target_->SanitizeEnvOptions(opts);
}
FileOptions OptimizeForLogRead(
const FileOptions& file_options) const override {
return target_->OptimizeForLogRead(file_options);
}
FileOptions OptimizeForManifestRead(
const FileOptions& file_options) const override {
return target_->OptimizeForManifestRead(file_options);
}
FileOptions OptimizeForLogWrite(const FileOptions& file_options,
const DBOptions& db_options) const override {
return target_->OptimizeForLogWrite(file_options, db_options);
}
FileOptions OptimizeForManifestWrite(
const FileOptions& file_options) const override {
return target_->OptimizeForManifestWrite(file_options);
}
FileOptions OptimizeForCompactionTableWrite(
const FileOptions& file_options,
const ImmutableDBOptions& immutable_ops) const override {
return target_->OptimizeForCompactionTableWrite(file_options,
immutable_ops);
}
FileOptions OptimizeForCompactionTableRead(
const FileOptions& file_options,
const ImmutableDBOptions& db_options) const override {
return target_->OptimizeForCompactionTableRead(file_options, db_options);
}
FileOptions OptimizeForBlobFileRead(
const FileOptions& file_options,
const ImmutableDBOptions& db_options) const override {
return target_->OptimizeForBlobFileRead(file_options, db_options);
}
Create a CustomEnv class; Add WinFileSystem; Make LegacyFileSystemWrapper private (#7703) Summary: This PR does the following: -> Creates a WinFileSystem class. This class is the Windows equivalent of the PosixFileSystem and will be used on Windows systems. -> Introduces a CustomEnv class. A CustomEnv is an Env that takes a FileSystem as constructor argument. I believe there will only ever be two implementations of this class (PosixEnv and WinEnv). There is still a CustomEnvWrapper class that takes an Env and a FileSystem and wraps the Env calls with the input Env but uses the FileSystem for the FileSystem calls -> Eliminates the public uses of the LegacyFileSystemWrapper. With this change in place, there are effectively the following patterns of Env: - "Base Env classes" (PosixEnv, WinEnv). These classes implement the core Env functions (e.g. Threads) and have a hard-coded input FileSystem. These classes inherit from CompositeEnv, implement the core Env functions (threads) and delegate the FileSystem-like calls to the input file system. - Wrapped Composite Env classes (MemEnv). These classes take in an Env and a FileSystem. The core env functions are re-directed to the wrapped env. The file system calls are redirected to the input file system - Legacy Wrapped Env classes. These classes take in an Env input (but no FileSystem). The core env functions are re-directed to the wrapped env. A "Legacy File System" is created using this env and the file system calls directed to the env itself. With these changes in place, the PosixEnv becomes a singleton -- there is only ever one created. Any other use of the PosixEnv is via another wrapped env. This cleans up some of the issues with the env construction and destruction. Additionally, there were places in the code that required had an Env when they required a FileSystem. Many of these places would wrap the Env with a LegacyFileSystemWrapper instead of using the env->GetFileSystem(). These places were changed, thereby removing layers of additional redirection (LegacyFileSystem --> Env --> Env::FileSystem). Pull Request resolved: https://github.com/facebook/rocksdb/pull/7703 Reviewed By: zhichao-cao Differential Revision: D25762190 Pulled By: anand1976 fbshipit-source-id: 1a088e97fc916f28ac69c149cd1dcad0ab31704b
2021-01-06 19:48:24 +01:00
#ifdef GetFreeSpace
#undef GetFreeSpace
#endif
IOStatus GetFreeSpace(const std::string& path, const IOOptions& /*options*/,
uint64_t* diskfree, IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->GetFreeSpace(path, diskfree));
}
IOStatus IsDirectory(const std::string& path, const IOOptions& /*options*/,
bool* is_dir, IODebugContext* /*dbg*/) override {
return status_to_io_status(target_->IsDirectory(path, is_dir));
}
#ifndef ROCKSDB_LITE
std::string SerializeOptions(const ConfigOptions& /*config_options*/,
const std::string& /*prefix*/) const override {
// We do not want the LegacyFileSystem to appear in the serialized output.
// This clock is an internal class for those who do not implement one and
// would be part of the Env. As such, do not serialize it here.
return "";
}
#endif // ROCKSDB_LITE
Create a CustomEnv class; Add WinFileSystem; Make LegacyFileSystemWrapper private (#7703) Summary: This PR does the following: -> Creates a WinFileSystem class. This class is the Windows equivalent of the PosixFileSystem and will be used on Windows systems. -> Introduces a CustomEnv class. A CustomEnv is an Env that takes a FileSystem as constructor argument. I believe there will only ever be two implementations of this class (PosixEnv and WinEnv). There is still a CustomEnvWrapper class that takes an Env and a FileSystem and wraps the Env calls with the input Env but uses the FileSystem for the FileSystem calls -> Eliminates the public uses of the LegacyFileSystemWrapper. With this change in place, there are effectively the following patterns of Env: - "Base Env classes" (PosixEnv, WinEnv). These classes implement the core Env functions (e.g. Threads) and have a hard-coded input FileSystem. These classes inherit from CompositeEnv, implement the core Env functions (threads) and delegate the FileSystem-like calls to the input file system. - Wrapped Composite Env classes (MemEnv). These classes take in an Env and a FileSystem. The core env functions are re-directed to the wrapped env. The file system calls are redirected to the input file system - Legacy Wrapped Env classes. These classes take in an Env input (but no FileSystem). The core env functions are re-directed to the wrapped env. A "Legacy File System" is created using this env and the file system calls directed to the env itself. With these changes in place, the PosixEnv becomes a singleton -- there is only ever one created. Any other use of the PosixEnv is via another wrapped env. This cleans up some of the issues with the env construction and destruction. Additionally, there were places in the code that required had an Env when they required a FileSystem. Many of these places would wrap the Env with a LegacyFileSystemWrapper instead of using the env->GetFileSystem(). These places were changed, thereby removing layers of additional redirection (LegacyFileSystem --> Env --> Env::FileSystem). Pull Request resolved: https://github.com/facebook/rocksdb/pull/7703 Reviewed By: zhichao-cao Differential Revision: D25762190 Pulled By: anand1976 fbshipit-source-id: 1a088e97fc916f28ac69c149cd1dcad0ab31704b
2021-01-06 19:48:24 +01:00
private:
Env* target_;
};
} // end anonymous namespace
Simplify migration to FileSystem API (#6552) Summary: The current Env/FileSystem API separation has a couple of issues - 1. It requires the user to specify 2 options - ```Options::env``` and ```Options::file_system``` - which means they have to make code changes to benefit from the new APIs. Furthermore, there is a risk of accessing the same APIs in two different ways, through Env in the old way and through FileSystem in the new way. The two may not always match, for example, if env is ```PosixEnv``` and FileSystem is a custom implementation. Any stray RocksDB calls to env will use the ```PosixEnv``` implementation rather than the file_system implementation. 2. There needs to be a simple way for the FileSystem developer to instantiate an Env for backward compatibility purposes. This PR solves the above issues and simplifies the migration in the following ways - 1. Embed a shared_ptr to the ```FileSystem``` in the ```Env```, and remove ```Options::file_system``` as a configurable option. This way, no code changes will be required in application code to benefit from the new API. The default Env constructor uses a ```LegacyFileSystemWrapper``` as the embedded ```FileSystem```. 1a. - This also makes it more robust by ensuring that even if RocksDB has some stray calls to Env APIs rather than FileSystem, they will go through the same object and thus there is no risk of getting out of sync. 2. Provide a ```NewCompositeEnv()``` API that can be used to construct a PosixEnv with a custom FileSystem implementation. This eliminates an indirection to call Env APIs, and relieves the FileSystem developer of the burden of having to implement wrappers for the Env APIs. 3. Add a couple of missing FileSystem APIs - ```SanitizeEnvOptions()``` and ```NewLogger()``` Tests: 1. New unit tests 2. make check and make asan_check Pull Request resolved: https://github.com/facebook/rocksdb/pull/6552 Reviewed By: riversand963 Differential Revision: D20592038 Pulled By: anand1976 fbshipit-source-id: c3801ad4153f96d21d5a3ae26c92ba454d1bf1f7
2020-03-24 05:50:42 +01:00
Env::Env() : thread_status_updater_(nullptr) {
file_system_ = std::make_shared<LegacyFileSystemWrapper>(this);
system_clock_ = std::make_shared<LegacySystemClock>(this);
Simplify migration to FileSystem API (#6552) Summary: The current Env/FileSystem API separation has a couple of issues - 1. It requires the user to specify 2 options - ```Options::env``` and ```Options::file_system``` - which means they have to make code changes to benefit from the new APIs. Furthermore, there is a risk of accessing the same APIs in two different ways, through Env in the old way and through FileSystem in the new way. The two may not always match, for example, if env is ```PosixEnv``` and FileSystem is a custom implementation. Any stray RocksDB calls to env will use the ```PosixEnv``` implementation rather than the file_system implementation. 2. There needs to be a simple way for the FileSystem developer to instantiate an Env for backward compatibility purposes. This PR solves the above issues and simplifies the migration in the following ways - 1. Embed a shared_ptr to the ```FileSystem``` in the ```Env```, and remove ```Options::file_system``` as a configurable option. This way, no code changes will be required in application code to benefit from the new API. The default Env constructor uses a ```LegacyFileSystemWrapper``` as the embedded ```FileSystem```. 1a. - This also makes it more robust by ensuring that even if RocksDB has some stray calls to Env APIs rather than FileSystem, they will go through the same object and thus there is no risk of getting out of sync. 2. Provide a ```NewCompositeEnv()``` API that can be used to construct a PosixEnv with a custom FileSystem implementation. This eliminates an indirection to call Env APIs, and relieves the FileSystem developer of the burden of having to implement wrappers for the Env APIs. 3. Add a couple of missing FileSystem APIs - ```SanitizeEnvOptions()``` and ```NewLogger()``` Tests: 1. New unit tests 2. make check and make asan_check Pull Request resolved: https://github.com/facebook/rocksdb/pull/6552 Reviewed By: riversand963 Differential Revision: D20592038 Pulled By: anand1976 fbshipit-source-id: c3801ad4153f96d21d5a3ae26c92ba454d1bf1f7
2020-03-24 05:50:42 +01:00
}
Env::Env(const std::shared_ptr<FileSystem>& fs)
: thread_status_updater_(nullptr), file_system_(fs) {
system_clock_ = std::make_shared<LegacySystemClock>(this);
}
Env::Env(const std::shared_ptr<FileSystem>& fs,
const std::shared_ptr<SystemClock>& clock)
: thread_status_updater_(nullptr), file_system_(fs), system_clock_(clock) {}
Simplify migration to FileSystem API (#6552) Summary: The current Env/FileSystem API separation has a couple of issues - 1. It requires the user to specify 2 options - ```Options::env``` and ```Options::file_system``` - which means they have to make code changes to benefit from the new APIs. Furthermore, there is a risk of accessing the same APIs in two different ways, through Env in the old way and through FileSystem in the new way. The two may not always match, for example, if env is ```PosixEnv``` and FileSystem is a custom implementation. Any stray RocksDB calls to env will use the ```PosixEnv``` implementation rather than the file_system implementation. 2. There needs to be a simple way for the FileSystem developer to instantiate an Env for backward compatibility purposes. This PR solves the above issues and simplifies the migration in the following ways - 1. Embed a shared_ptr to the ```FileSystem``` in the ```Env```, and remove ```Options::file_system``` as a configurable option. This way, no code changes will be required in application code to benefit from the new API. The default Env constructor uses a ```LegacyFileSystemWrapper``` as the embedded ```FileSystem```. 1a. - This also makes it more robust by ensuring that even if RocksDB has some stray calls to Env APIs rather than FileSystem, they will go through the same object and thus there is no risk of getting out of sync. 2. Provide a ```NewCompositeEnv()``` API that can be used to construct a PosixEnv with a custom FileSystem implementation. This eliminates an indirection to call Env APIs, and relieves the FileSystem developer of the burden of having to implement wrappers for the Env APIs. 3. Add a couple of missing FileSystem APIs - ```SanitizeEnvOptions()``` and ```NewLogger()``` Tests: 1. New unit tests 2. make check and make asan_check Pull Request resolved: https://github.com/facebook/rocksdb/pull/6552 Reviewed By: riversand963 Differential Revision: D20592038 Pulled By: anand1976 fbshipit-source-id: c3801ad4153f96d21d5a3ae26c92ba454d1bf1f7
2020-03-24 05:50:42 +01:00
Env::~Env() {
}
Status Env::NewLogger(const std::string& fname,
std::shared_ptr<Logger>* result) {
return NewEnvLogger(fname, this, result);
}
Status Env::LoadEnv(const std::string& value, Env** result) {
return CreateFromString(ConfigOptions(), value, result);
}
Status Env::CreateFromString(const ConfigOptions& config_options,
const std::string& value, Env** result) {
Env* base = Env::Default();
if (value.empty() || base->IsInstanceOf(value)) {
*result = base;
return Status::OK();
} else {
RegisterSystemEnvs();
Env* env = *result;
Status s = LoadStaticObject<Env>(config_options, value, nullptr, &env);
if (s.ok()) {
*result = env;
}
return s;
}
}
Status Env::LoadEnv(const std::string& value, Env** result,
std::shared_ptr<Env>* guard) {
return CreateFromString(ConfigOptions(), value, result, guard);
}
Status Env::CreateFromString(const ConfigOptions& config_options,
const std::string& value, Env** result,
std::shared_ptr<Env>* guard) {
assert(result);
assert(guard != nullptr);
std::unique_ptr<Env> uniq;
Env* env = *result;
std::string id;
std::unordered_map<std::string, std::string> opt_map;
Status status =
Customizable::GetOptionsMap(config_options, env, value, &id, &opt_map);
if (!status.ok()) { // GetOptionsMap failed
return status;
}
Env* base = Env::Default();
if (id.empty() || base->IsInstanceOf(id)) {
env = base;
status = Status::OK();
} else {
RegisterSystemEnvs();
#ifndef ROCKSDB_LITE
Return different Status based on ObjectRegistry::NewObject calls (#9333) Summary: This fix addresses https://github.com/facebook/rocksdb/issues/9299. If attempting to create a new object via the ObjectRegistry and a factory is not found, the ObjectRegistry will return a "NotSupported" status. This is the same behavior as previously. If the factory is found but could not successfully create the object, an "InvalidArgument" status is returned. If the factory returned a reason why (in the errmsg), this message will be in the returned status. In practice, there are two options in the ConfigOptions that control how these errors are propagated: - If "ignore_unknown_options=true", then both InvalidArgument and NotSupported status codes will be swallowed internally. Both cases will return success - If "ignore_unsupported_options=true", then having no factory will return success but a failing factory will return an error - If both options are false, both cases (no and failing factory) will return errors. In practice this likely only changes Customizable that may be partially available. For example, the JEMallocMemoryAllocator is a built-in allocator that is registered with the system but may not be compiled in. In this case, the status code for this allocator changed from NotSupported("JEMalloc not available") to InvalidArgumen("JEMalloc not available"). Other Customizable builtins/plugins would have the same semantics. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9333 Reviewed By: pdillinger Differential Revision: D33517681 Pulled By: mrambacher fbshipit-source-id: 8033052d4a4a7b88c2d9f90147b1b4467e51f6fd
2022-02-11 14:10:10 +01:00
// First, try to load the Env as a unique object.
status = config_options.registry->NewObject<Env>(id, &env, &uniq);
#else
status =
Status::NotSupported("Cannot load environment in LITE mode", value);
#endif
}
if (config_options.ignore_unsupported_options && status.IsNotSupported()) {
status = Status::OK();
} else if (status.ok()) {
status = Customizable::ConfigureNewObject(config_options, env, opt_map);
}
if (status.ok()) {
guard->reset(uniq.release());
*result = env;
}
return status;
}
Status Env::CreateFromUri(const ConfigOptions& config_options,
const std::string& env_uri, const std::string& fs_uri,
Env** result, std::shared_ptr<Env>* guard) {
*result = config_options.env;
if (env_uri.empty() && fs_uri.empty()) {
// Neither specified. Use the default
guard->reset();
return Status::OK();
} else if (!env_uri.empty() && !fs_uri.empty()) {
// Both specified. Cannot choose. Return Invalid
return Status::InvalidArgument("cannot specify both fs_uri and env_uri");
} else if (fs_uri.empty()) { // Only have an ENV URI. Create an Env from it
return CreateFromString(config_options, env_uri, result, guard);
} else {
std::shared_ptr<FileSystem> fs;
Status s = FileSystem::CreateFromString(config_options, fs_uri, &fs);
if (s.ok()) {
guard->reset(new CompositeEnvWrapper(*result, fs));
*result = guard->get();
}
return s;
}
}
std::string Env::PriorityToString(Env::Priority priority) {
switch (priority) {
case Env::Priority::BOTTOM:
return "Bottom";
case Env::Priority::LOW:
return "Low";
case Env::Priority::HIGH:
return "High";
case Env::Priority::USER:
return "User";
case Env::Priority::TOTAL:
assert(false);
}
return "Invalid";
}
uint64_t Env::GetThreadID() const {
std::hash<std::thread::id> hasher;
return hasher(std::this_thread::get_id());
}
Status Env::ReuseWritableFile(const std::string& fname,
const std::string& old_fname,
std::unique_ptr<WritableFile>* result,
const EnvOptions& options) {
Status s = RenameFile(old_fname, fname);
if (!s.ok()) {
return s;
}
return NewWritableFile(fname, result, options);
}
Status Env::GetChildrenFileAttributes(const std::string& dir,
std::vector<FileAttributes>* result) {
assert(result != nullptr);
std::vector<std::string> child_fnames;
Status s = GetChildren(dir, &child_fnames);
if (!s.ok()) {
return s;
}
result->resize(child_fnames.size());
size_t result_size = 0;
for (size_t i = 0; i < child_fnames.size(); ++i) {
const std::string path = dir + "/" + child_fnames[i];
if (!(s = GetFileSize(path, &(*result)[result_size].size_bytes)).ok()) {
if (FileExists(path).IsNotFound()) {
// The file may have been deleted since we listed the directory
continue;
}
return s;
}
(*result)[result_size].name = std::move(child_fnames[i]);
result_size++;
}
result->resize(result_size);
return Status::OK();
}
Status Env::GetHostNameString(std::string* result) {
std::array<char, kMaxHostNameLen> hostname_buf{};
Status s = GetHostName(hostname_buf.data(), hostname_buf.size());
if (s.ok()) {
hostname_buf[hostname_buf.size() - 1] = '\0';
result->assign(hostname_buf.data());
}
return s;
}
Built-in support for generating unique IDs, bug fix (#8708) Summary: Env::GenerateUniqueId() works fine on Windows and on POSIX where /proc/sys/kernel/random/uuid exists. Our other implementation is flawed and easily produces collision in a new multi-threaded test. As we rely more heavily on DB session ID uniqueness, this becomes a serious issue. This change combines several individually suitable entropy sources for reliable generation of random unique IDs, with goal of uniqueness and portability, not cryptographic strength nor maximum speed. Specifically: * Moves code for getting UUIDs from the OS to port::GenerateRfcUuid rather than in Env implementation details. Callers are now told whether the operation fails or succeeds. * Adds an internal API GenerateRawUniqueId for generating high-quality 128-bit unique identifiers, by combining entropy from three "tracks": * Lots of info from default Env like time, process id, and hostname. * std::random_device * port::GenerateRfcUuid (when working) * Built-in implementations of Env::GenerateUniqueId() will now always produce an RFC 4122 UUID string, either from platform-specific API or by converting the output of GenerateRawUniqueId. DB session IDs now use GenerateRawUniqueId while DB IDs (not as critical) try to use port::GenerateRfcUuid but fall back on GenerateRawUniqueId with conversion to an RFC 4122 UUID. GenerateRawUniqueId is declared and defined under env/ rather than util/ or even port/ because of the Env dependency. Likely follow-up: enhance GenerateRawUniqueId to be faster after the first call and to guarantee uniqueness within the lifetime of a single process (imparting the same property onto DB session IDs). Pull Request resolved: https://github.com/facebook/rocksdb/pull/8708 Test Plan: A new mini-stress test in env_test checks the various public and internal APIs for uniqueness, including each track of GenerateRawUniqueId individually. We can't hope to verify anywhere close to 128 bits of entropy, but it can at least detect flaws as bad as the old code. Serial execution of the new tests takes about 350 ms on my machine. Reviewed By: zhichao-cao, mrambacher Differential Revision: D30563780 Pulled By: pdillinger fbshipit-source-id: de4c9ff4b2f581cf784fcedb5f39f16e5185c364
2021-08-31 00:19:39 +02:00
std::string Env::GenerateUniqueId() {
std::string result;
bool success = port::GenerateRfcUuid(&result);
if (!success) {
// Fall back on our own way of generating a unique ID and adapt it to
// RFC 4122 variant 1 version 4 (a random ID).
// https://en.wikipedia.org/wiki/Universally_unique_identifier
// We already tried GenerateRfcUuid so no need to try it again in
// GenerateRawUniqueId
constexpr bool exclude_port_uuid = true;
uint64_t upper, lower;
GenerateRawUniqueId(&upper, &lower, exclude_port_uuid);
// Set 4-bit version to 4
upper = (upper & (~uint64_t{0xf000})) | 0x4000;
// Set unary-encoded variant to 1 (0b10)
lower = (lower & (~(uint64_t{3} << 62))) | (uint64_t{2} << 62);
// Use 36 character format of RFC 4122
result.resize(36U);
char* buf = &result[0];
PutBaseChars<16>(&buf, 8, upper >> 32, /*!uppercase*/ false);
*(buf++) = '-';
PutBaseChars<16>(&buf, 4, upper >> 16, /*!uppercase*/ false);
*(buf++) = '-';
PutBaseChars<16>(&buf, 4, upper, /*!uppercase*/ false);
*(buf++) = '-';
PutBaseChars<16>(&buf, 4, lower >> 48, /*!uppercase*/ false);
*(buf++) = '-';
PutBaseChars<16>(&buf, 12, lower, /*!uppercase*/ false);
assert(buf == &result[36]);
// Verify variant 1 version 4
assert(result[14] == '4');
assert(result[19] == '8' || result[19] == '9' || result[19] == 'a' ||
result[19] == 'b');
}
return result;
}
SequentialFile::~SequentialFile() {
}
RandomAccessFile::~RandomAccessFile() {
}
WritableFile::~WritableFile() {
}
MemoryMappedFileBuffer::~MemoryMappedFileBuffer() {}
Logger::~Logger() {}
Status Logger::Close() {
if (!closed_) {
closed_ = true;
return CloseImpl();
} else {
return Status::OK();
}
}
Status Logger::CloseImpl() { return Status::NotSupported(); }
FileLock::~FileLock() {
}
void LogFlush(Logger *info_log) {
if (info_log) {
info_log->Flush();
}
}
static void Logv(Logger *info_log, const char* format, va_list ap) {
if (info_log && info_log->GetInfoLogLevel() <= InfoLogLevel::INFO_LEVEL) {
info_log->Logv(InfoLogLevel::INFO_LEVEL, format, ap);
}
}
void Log(Logger* info_log, const char* format, ...) {
va_list ap;
va_start(ap, format);
Logv(info_log, format, ap);
va_end(ap);
}
void Logger::Logv(const InfoLogLevel log_level, const char* format, va_list ap) {
static const char* kInfoLogLevelNames[5] = { "DEBUG", "INFO", "WARN",
"ERROR", "FATAL" };
if (log_level < log_level_) {
return;
}
if (log_level == InfoLogLevel::INFO_LEVEL) {
// Doesn't print log level if it is INFO level.
// This is to avoid unexpected performance regression after we add
// the feature of log level. All the logs before we add the feature
// are INFO level. We don't want to add extra costs to those existing
// logging.
Logv(format, ap);
} else if (log_level == InfoLogLevel::HEADER_LEVEL) {
LogHeader(format, ap);
} else {
char new_format[500];
snprintf(new_format, sizeof(new_format) - 1, "[%s] %s",
kInfoLogLevelNames[log_level], format);
Logv(new_format, ap);
}
if (log_level >= InfoLogLevel::WARN_LEVEL &&
log_level != InfoLogLevel::HEADER_LEVEL) {
// Log messages with severity of warning or higher should be rare and are
// sometimes followed by an unclean crash. We want to be sure important
// messages are not lost in an application buffer when that happens.
Flush();
}
}
static void Logv(const InfoLogLevel log_level, Logger *info_log, const char *format, va_list ap) {
if (info_log && info_log->GetInfoLogLevel() <= log_level) {
if (log_level == InfoLogLevel::HEADER_LEVEL) {
info_log->LogHeader(format, ap);
} else {
info_log->Logv(log_level, format, ap);
}
}
}
void Log(const InfoLogLevel log_level, Logger* info_log, const char* format,
...) {
va_list ap;
va_start(ap, format);
Logv(log_level, info_log, format, ap);
va_end(ap);
}
static void Headerv(Logger *info_log, const char *format, va_list ap) {
if (info_log) {
info_log->LogHeader(format, ap);
}
}
void Header(Logger* info_log, const char* format, ...) {
va_list ap;
va_start(ap, format);
Headerv(info_log, format, ap);
va_end(ap);
}
static void Debugv(Logger* info_log, const char* format, va_list ap) {
if (info_log && info_log->GetInfoLogLevel() <= InfoLogLevel::DEBUG_LEVEL) {
info_log->Logv(InfoLogLevel::DEBUG_LEVEL, format, ap);
}
}
void Debug(Logger* info_log, const char* format, ...) {
va_list ap;
va_start(ap, format);
Debugv(info_log, format, ap);
va_end(ap);
}
static void Infov(Logger* info_log, const char* format, va_list ap) {
if (info_log && info_log->GetInfoLogLevel() <= InfoLogLevel::INFO_LEVEL) {
info_log->Logv(InfoLogLevel::INFO_LEVEL, format, ap);
}
}
void Info(Logger* info_log, const char* format, ...) {
va_list ap;
va_start(ap, format);
Infov(info_log, format, ap);
va_end(ap);
}
static void Warnv(Logger* info_log, const char* format, va_list ap) {
if (info_log && info_log->GetInfoLogLevel() <= InfoLogLevel::WARN_LEVEL) {
info_log->Logv(InfoLogLevel::WARN_LEVEL, format, ap);
}
}
void Warn(Logger* info_log, const char* format, ...) {
va_list ap;
va_start(ap, format);
Warnv(info_log, format, ap);
va_end(ap);
}
static void Errorv(Logger* info_log, const char* format, va_list ap) {
if (info_log && info_log->GetInfoLogLevel() <= InfoLogLevel::ERROR_LEVEL) {
info_log->Logv(InfoLogLevel::ERROR_LEVEL, format, ap);
}
}
void Error(Logger* info_log, const char* format, ...) {
va_list ap;
va_start(ap, format);
Errorv(info_log, format, ap);
va_end(ap);
}
static void Fatalv(Logger* info_log, const char* format, va_list ap) {
if (info_log && info_log->GetInfoLogLevel() <= InfoLogLevel::FATAL_LEVEL) {
info_log->Logv(InfoLogLevel::FATAL_LEVEL, format, ap);
}
}
void Fatal(Logger* info_log, const char* format, ...) {
va_list ap;
va_start(ap, format);
Fatalv(info_log, format, ap);
va_end(ap);
}
void LogFlush(const std::shared_ptr<Logger>& info_log) {
LogFlush(info_log.get());
}
void Log(const InfoLogLevel log_level, const std::shared_ptr<Logger>& info_log,
const char* format, ...) {
va_list ap;
va_start(ap, format);
Logv(log_level, info_log.get(), format, ap);
va_end(ap);
}
void Header(const std::shared_ptr<Logger>& info_log, const char* format, ...) {
va_list ap;
va_start(ap, format);
Headerv(info_log.get(), format, ap);
va_end(ap);
}
void Debug(const std::shared_ptr<Logger>& info_log, const char* format, ...) {
va_list ap;
va_start(ap, format);
Debugv(info_log.get(), format, ap);
va_end(ap);
}
void Info(const std::shared_ptr<Logger>& info_log, const char* format, ...) {
va_list ap;
va_start(ap, format);
Infov(info_log.get(), format, ap);
va_end(ap);
}
void Warn(const std::shared_ptr<Logger>& info_log, const char* format, ...) {
va_list ap;
va_start(ap, format);
Warnv(info_log.get(), format, ap);
va_end(ap);
}
void Error(const std::shared_ptr<Logger>& info_log, const char* format, ...) {
va_list ap;
va_start(ap, format);
Errorv(info_log.get(), format, ap);
va_end(ap);
}
void Fatal(const std::shared_ptr<Logger>& info_log, const char* format, ...) {
va_list ap;
va_start(ap, format);
Fatalv(info_log.get(), format, ap);
va_end(ap);
}
void Log(const std::shared_ptr<Logger>& info_log, const char* format, ...) {
va_list ap;
va_start(ap, format);
Logv(info_log.get(), format, ap);
va_end(ap);
}
Status WriteStringToFile(Env* env, const Slice& data, const std::string& fname,
bool should_sync) {
Create a CustomEnv class; Add WinFileSystem; Make LegacyFileSystemWrapper private (#7703) Summary: This PR does the following: -> Creates a WinFileSystem class. This class is the Windows equivalent of the PosixFileSystem and will be used on Windows systems. -> Introduces a CustomEnv class. A CustomEnv is an Env that takes a FileSystem as constructor argument. I believe there will only ever be two implementations of this class (PosixEnv and WinEnv). There is still a CustomEnvWrapper class that takes an Env and a FileSystem and wraps the Env calls with the input Env but uses the FileSystem for the FileSystem calls -> Eliminates the public uses of the LegacyFileSystemWrapper. With this change in place, there are effectively the following patterns of Env: - "Base Env classes" (PosixEnv, WinEnv). These classes implement the core Env functions (e.g. Threads) and have a hard-coded input FileSystem. These classes inherit from CompositeEnv, implement the core Env functions (threads) and delegate the FileSystem-like calls to the input file system. - Wrapped Composite Env classes (MemEnv). These classes take in an Env and a FileSystem. The core env functions are re-directed to the wrapped env. The file system calls are redirected to the input file system - Legacy Wrapped Env classes. These classes take in an Env input (but no FileSystem). The core env functions are re-directed to the wrapped env. A "Legacy File System" is created using this env and the file system calls directed to the env itself. With these changes in place, the PosixEnv becomes a singleton -- there is only ever one created. Any other use of the PosixEnv is via another wrapped env. This cleans up some of the issues with the env construction and destruction. Additionally, there were places in the code that required had an Env when they required a FileSystem. Many of these places would wrap the Env with a LegacyFileSystemWrapper instead of using the env->GetFileSystem(). These places were changed, thereby removing layers of additional redirection (LegacyFileSystem --> Env --> Env::FileSystem). Pull Request resolved: https://github.com/facebook/rocksdb/pull/7703 Reviewed By: zhichao-cao Differential Revision: D25762190 Pulled By: anand1976 fbshipit-source-id: 1a088e97fc916f28ac69c149cd1dcad0ab31704b
2021-01-06 19:48:24 +01:00
const auto& fs = env->GetFileSystem();
return WriteStringToFile(fs.get(), data, fname, should_sync);
}
Status ReadFileToString(Env* env, const std::string& fname, std::string* data) {
Create a CustomEnv class; Add WinFileSystem; Make LegacyFileSystemWrapper private (#7703) Summary: This PR does the following: -> Creates a WinFileSystem class. This class is the Windows equivalent of the PosixFileSystem and will be used on Windows systems. -> Introduces a CustomEnv class. A CustomEnv is an Env that takes a FileSystem as constructor argument. I believe there will only ever be two implementations of this class (PosixEnv and WinEnv). There is still a CustomEnvWrapper class that takes an Env and a FileSystem and wraps the Env calls with the input Env but uses the FileSystem for the FileSystem calls -> Eliminates the public uses of the LegacyFileSystemWrapper. With this change in place, there are effectively the following patterns of Env: - "Base Env classes" (PosixEnv, WinEnv). These classes implement the core Env functions (e.g. Threads) and have a hard-coded input FileSystem. These classes inherit from CompositeEnv, implement the core Env functions (threads) and delegate the FileSystem-like calls to the input file system. - Wrapped Composite Env classes (MemEnv). These classes take in an Env and a FileSystem. The core env functions are re-directed to the wrapped env. The file system calls are redirected to the input file system - Legacy Wrapped Env classes. These classes take in an Env input (but no FileSystem). The core env functions are re-directed to the wrapped env. A "Legacy File System" is created using this env and the file system calls directed to the env itself. With these changes in place, the PosixEnv becomes a singleton -- there is only ever one created. Any other use of the PosixEnv is via another wrapped env. This cleans up some of the issues with the env construction and destruction. Additionally, there were places in the code that required had an Env when they required a FileSystem. Many of these places would wrap the Env with a LegacyFileSystemWrapper instead of using the env->GetFileSystem(). These places were changed, thereby removing layers of additional redirection (LegacyFileSystem --> Env --> Env::FileSystem). Pull Request resolved: https://github.com/facebook/rocksdb/pull/7703 Reviewed By: zhichao-cao Differential Revision: D25762190 Pulled By: anand1976 fbshipit-source-id: 1a088e97fc916f28ac69c149cd1dcad0ab31704b
2021-01-06 19:48:24 +01:00
const auto& fs = env->GetFileSystem();
return ReadFileToString(fs.get(), fname, data);
}
namespace { // anonymous namespace
void AssignEnvOptions(EnvOptions* env_options, const DBOptions& options) {
env_options->use_mmap_reads = options.allow_mmap_reads;
env_options->use_mmap_writes = options.allow_mmap_writes;
env_options->use_direct_reads = options.use_direct_reads;
env_options->set_fd_cloexec = options.is_fd_close_on_exec;
env_options->bytes_per_sync = options.bytes_per_sync;
env_options->compaction_readahead_size = options.compaction_readahead_size;
env_options->random_access_max_buffer_size =
options.random_access_max_buffer_size;
env_options->rate_limiter = options.rate_limiter.get();
env_options->writable_file_max_buffer_size =
options.writable_file_max_buffer_size;
env_options->allow_fallocate = options.allow_fallocate;
Optionally wait on bytes_per_sync to smooth I/O (#5183) Summary: The existing implementation does not guarantee bytes reach disk every `bytes_per_sync` when writing SST files, or every `wal_bytes_per_sync` when writing WALs. This can cause confusing behavior for users who enable this feature to avoid large syncs during flush and compaction, but then end up hitting them anyways. My understanding of the existing behavior is we used `sync_file_range` with `SYNC_FILE_RANGE_WRITE` to submit ranges for async writeback, such that we could continue processing the next range of bytes while that I/O is happening. I believe we can preserve that benefit while also limiting how far the processing can get ahead of the I/O, which prevents huge syncs from happening when the file finishes. Consider this `sync_file_range` usage: `sync_file_range(fd_, 0, static_cast<off_t>(offset + nbytes), SYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE)`. Expanding the range to start at 0 and adding the `SYNC_FILE_RANGE_WAIT_BEFORE` flag causes any pending writeback (like from a previous call to `sync_file_range`) to finish before it proceeds to submit the latest `nbytes` for writeback. The latest `nbytes` are still written back asynchronously, unless processing exceeds I/O speed, in which case the following `sync_file_range` will need to wait on it. There is a second change in this PR to use `fdatasync` when `sync_file_range` is unavailable (determined statically) or has some known problem with the underlying filesystem (determined dynamically). The above two changes only apply when the user enables a new option, `strict_bytes_per_sync`. Pull Request resolved: https://github.com/facebook/rocksdb/pull/5183 Differential Revision: D14953553 Pulled By: siying fbshipit-source-id: 445c3862e019fb7b470f9c7f314fc231b62706e9
2019-04-22 20:48:45 +02:00
env_options->strict_bytes_per_sync = options.strict_bytes_per_sync;
options.env->SanitizeEnvOptions(env_options);
}
}
EnvOptions Env::OptimizeForLogWrite(const EnvOptions& env_options,
const DBOptions& db_options) const {
EnvOptions optimized_env_options(env_options);
optimized_env_options.bytes_per_sync = db_options.wal_bytes_per_sync;
optimized_env_options.writable_file_max_buffer_size =
db_options.writable_file_max_buffer_size;
return optimized_env_options;
}
EnvOptions Env::OptimizeForManifestWrite(const EnvOptions& env_options) const {
return env_options;
}
EnvOptions Env::OptimizeForLogRead(const EnvOptions& env_options) const {
EnvOptions optimized_env_options(env_options);
optimized_env_options.use_direct_reads = false;
return optimized_env_options;
}
EnvOptions Env::OptimizeForManifestRead(const EnvOptions& env_options) const {
EnvOptions optimized_env_options(env_options);
optimized_env_options.use_direct_reads = false;
return optimized_env_options;
}
EnvOptions Env::OptimizeForCompactionTableWrite(
const EnvOptions& env_options, const ImmutableDBOptions& db_options) const {
EnvOptions optimized_env_options(env_options);
optimized_env_options.use_direct_writes =
db_options.use_direct_io_for_flush_and_compaction;
return optimized_env_options;
}
EnvOptions Env::OptimizeForCompactionTableRead(
const EnvOptions& env_options, const ImmutableDBOptions& db_options) const {
EnvOptions optimized_env_options(env_options);
optimized_env_options.use_direct_reads = db_options.use_direct_reads;
return optimized_env_options;
}
EnvOptions Env::OptimizeForBlobFileRead(
const EnvOptions& env_options, const ImmutableDBOptions& db_options) const {
EnvOptions optimized_env_options(env_options);
optimized_env_options.use_direct_reads = db_options.use_direct_reads;
return optimized_env_options;
}
EnvOptions::EnvOptions(const DBOptions& options) {
AssignEnvOptions(this, options);
}
EnvOptions::EnvOptions() {
DBOptions options;
AssignEnvOptions(this, options);
}
Status NewEnvLogger(const std::string& fname, Env* env,
std::shared_ptr<Logger>* result) {
FileOptions options;
// TODO: Tune the buffer size.
options.writable_file_max_buffer_size = 1024 * 1024;
std::unique_ptr<FSWritableFile> writable_file;
const auto status = env->GetFileSystem()->NewWritableFile(
fname, options, &writable_file, nullptr);
if (!status.ok()) {
return status;
}
*result = std::make_shared<EnvLogger>(std::move(writable_file), fname,
options, env);
return Status::OK();
}
Simplify migration to FileSystem API (#6552) Summary: The current Env/FileSystem API separation has a couple of issues - 1. It requires the user to specify 2 options - ```Options::env``` and ```Options::file_system``` - which means they have to make code changes to benefit from the new APIs. Furthermore, there is a risk of accessing the same APIs in two different ways, through Env in the old way and through FileSystem in the new way. The two may not always match, for example, if env is ```PosixEnv``` and FileSystem is a custom implementation. Any stray RocksDB calls to env will use the ```PosixEnv``` implementation rather than the file_system implementation. 2. There needs to be a simple way for the FileSystem developer to instantiate an Env for backward compatibility purposes. This PR solves the above issues and simplifies the migration in the following ways - 1. Embed a shared_ptr to the ```FileSystem``` in the ```Env```, and remove ```Options::file_system``` as a configurable option. This way, no code changes will be required in application code to benefit from the new API. The default Env constructor uses a ```LegacyFileSystemWrapper``` as the embedded ```FileSystem```. 1a. - This also makes it more robust by ensuring that even if RocksDB has some stray calls to Env APIs rather than FileSystem, they will go through the same object and thus there is no risk of getting out of sync. 2. Provide a ```NewCompositeEnv()``` API that can be used to construct a PosixEnv with a custom FileSystem implementation. This eliminates an indirection to call Env APIs, and relieves the FileSystem developer of the burden of having to implement wrappers for the Env APIs. 3. Add a couple of missing FileSystem APIs - ```SanitizeEnvOptions()``` and ```NewLogger()``` Tests: 1. New unit tests 2. make check and make asan_check Pull Request resolved: https://github.com/facebook/rocksdb/pull/6552 Reviewed By: riversand963 Differential Revision: D20592038 Pulled By: anand1976 fbshipit-source-id: c3801ad4153f96d21d5a3ae26c92ba454d1bf1f7
2020-03-24 05:50:42 +01:00
const std::shared_ptr<FileSystem>& Env::GetFileSystem() const {
return file_system_;
}
const std::shared_ptr<SystemClock>& Env::GetSystemClock() const {
return system_clock_;
}
namespace {
static std::unordered_map<std::string, OptionTypeInfo> sc_wrapper_type_info = {
#ifndef ROCKSDB_LITE
{"target",
OptionTypeInfo::AsCustomSharedPtr<SystemClock>(
0, OptionVerificationType::kByName, OptionTypeFlags::kDontSerialize)},
#endif // ROCKSDB_LITE
};
} // namespace
SystemClockWrapper::SystemClockWrapper(const std::shared_ptr<SystemClock>& t)
: target_(t) {
RegisterOptions("", &target_, &sc_wrapper_type_info);
}
Status SystemClockWrapper::PrepareOptions(const ConfigOptions& options) {
if (target_ == nullptr) {
target_ = SystemClock::Default();
}
return SystemClock::PrepareOptions(options);
}
#ifndef ROCKSDB_LITE
std::string SystemClockWrapper::SerializeOptions(
const ConfigOptions& config_options, const std::string& header) const {
auto parent = SystemClock::SerializeOptions(config_options, "");
if (config_options.IsShallow() || target_ == nullptr ||
target_->IsInstanceOf(SystemClock::kDefaultName())) {
return parent;
} else {
std::string result = header;
if (!StartsWith(parent, OptionTypeInfo::kIdPropName())) {
result.append(OptionTypeInfo::kIdPropName()).append("=");
}
result.append(parent);
if (!EndsWith(result, config_options.delimiter)) {
result.append(config_options.delimiter);
}
result.append("target=").append(target_->ToString(config_options));
return result;
}
}
#endif // ROCKSDB_LITE
#ifndef ROCKSDB_LITE
static int RegisterBuiltinSystemClocks(ObjectLibrary& library,
const std::string& /*arg*/) {
library.AddFactory<SystemClock>(
EmulatedSystemClock::kClassName(),
[](const std::string& /*uri*/, std::unique_ptr<SystemClock>* guard,
std::string* /* errmsg */) {
guard->reset(new EmulatedSystemClock(SystemClock::Default()));
return guard->get();
});
size_t num_types;
return static_cast<int>(library.GetFactoryCount(&num_types));
}
#endif // ROCKSDB_LITE
Status SystemClock::CreateFromString(const ConfigOptions& config_options,
const std::string& value,
std::shared_ptr<SystemClock>* result) {
auto clock = SystemClock::Default();
if (clock->IsInstanceOf(value)) {
*result = clock;
return Status::OK();
} else {
#ifndef ROCKSDB_LITE
static std::once_flag once;
std::call_once(once, [&]() {
RegisterBuiltinSystemClocks(*(ObjectLibrary::Default().get()), "");
});
#endif // ROCKSDB_LITE
return LoadSharedObject<SystemClock>(config_options, value, nullptr,
result);
}
}
} // namespace ROCKSDB_NAMESPACE