rocksdb/env/file_system.cc

264 lines
8.6 KiB
C++
Raw Normal View History

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
// Copyright (c) 2019-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).
//
#include "rocksdb/file_system.h"
#include "env/composite_env_wrapper.h"
#include "env/env_chroot.h"
#include "env/env_encryption_ctr.h"
#include "env/fs_readonly.h"
#include "env/mock_env.h"
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 "options/db_options.h"
#include "rocksdb/convenience.h"
#include "rocksdb/utilities/customizable_util.h"
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 "rocksdb/utilities/object_registry.h"
#include "rocksdb/utilities/options_type.h"
#include "util/string_util.h"
#include "utilities/env_timed.h"
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
namespace ROCKSDB_NAMESPACE {
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
FileSystem::FileSystem() {}
FileSystem::~FileSystem() {}
Status FileSystem::Load(const std::string& value,
std::shared_ptr<FileSystem>* result) {
return CreateFromString(ConfigOptions(), value, result);
}
#ifndef ROCKSDB_LITE
static int RegisterBuiltinFileSystems(ObjectLibrary& library,
const std::string& /*arg*/) {
library.Register<FileSystem>(
TimedFileSystem::kClassName(),
[](const std::string& /*uri*/, std::unique_ptr<FileSystem>* guard,
std::string* /* errmsg */) {
guard->reset(new TimedFileSystem(nullptr));
return guard->get();
});
library.Register<FileSystem>(
ReadOnlyFileSystem::kClassName(),
[](const std::string& /*uri*/, std::unique_ptr<FileSystem>* guard,
std::string* /* errmsg */) {
guard->reset(new ReadOnlyFileSystem(nullptr));
return guard->get();
});
library.Register<FileSystem>(
EncryptedFileSystem::kClassName(),
[](const std::string& /*uri*/, std::unique_ptr<FileSystem>* guard,
std::string* errmsg) {
Status s = NewEncryptedFileSystemImpl(nullptr, nullptr, guard);
if (!s.ok()) {
*errmsg = s.ToString();
}
return guard->get();
});
library.Register<FileSystem>(
MockFileSystem::kClassName(),
[](const std::string& /*uri*/, std::unique_ptr<FileSystem>* guard,
std::string* /*errmsg*/) {
guard->reset(new MockFileSystem(SystemClock::Default()));
return guard->get();
});
#ifndef OS_WIN
library.Register<FileSystem>(
ChrootFileSystem::kClassName(),
[](const std::string& /*uri*/, std::unique_ptr<FileSystem>* guard,
std::string* /* errmsg */) {
guard->reset(new ChrootFileSystem(nullptr, ""));
return guard->get();
});
#endif // OS_WIN
size_t num_types;
return static_cast<int>(library.GetFactoryCount(&num_types));
}
#endif // ROCKSDB_LITE
Status FileSystem::CreateFromString(const ConfigOptions& config_options,
const std::string& value,
std::shared_ptr<FileSystem>* result) {
auto default_fs = FileSystem::Default();
if (default_fs->IsInstanceOf(value)) {
*result = default_fs;
return Status::OK();
} else {
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
#ifndef ROCKSDB_LITE
static std::once_flag once;
std::call_once(once, [&]() {
RegisterBuiltinFileSystems(*(ObjectLibrary::Default().get()), "");
});
#endif // ROCKSDB_LITE
return LoadSharedObject<FileSystem>(config_options, value, nullptr, result);
}
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
}
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
IOStatus FileSystem::ReuseWritableFile(const std::string& fname,
const std::string& old_fname,
const FileOptions& opts,
std::unique_ptr<FSWritableFile>* result,
IODebugContext* dbg) {
IOStatus s = RenameFile(old_fname, fname, opts.io_options, dbg);
if (!s.ok()) {
return s;
}
return NewWritableFile(fname, opts, result, dbg);
}
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
FileOptions FileSystem::OptimizeForLogRead(
const FileOptions& file_options) const {
FileOptions optimized_file_options(file_options);
optimized_file_options.use_direct_reads = false;
return optimized_file_options;
}
FileOptions FileSystem::OptimizeForManifestRead(
const FileOptions& file_options) const {
FileOptions optimized_file_options(file_options);
optimized_file_options.use_direct_reads = false;
return optimized_file_options;
}
FileOptions FileSystem::OptimizeForLogWrite(const FileOptions& file_options,
const DBOptions& db_options) const {
FileOptions optimized_file_options(file_options);
optimized_file_options.bytes_per_sync = db_options.wal_bytes_per_sync;
optimized_file_options.writable_file_max_buffer_size =
db_options.writable_file_max_buffer_size;
return optimized_file_options;
}
FileOptions FileSystem::OptimizeForManifestWrite(
const FileOptions& file_options) const {
return file_options;
}
FileOptions FileSystem::OptimizeForCompactionTableWrite(
const FileOptions& file_options,
const ImmutableDBOptions& db_options) const {
FileOptions optimized_file_options(file_options);
optimized_file_options.use_direct_writes =
db_options.use_direct_io_for_flush_and_compaction;
return optimized_file_options;
}
FileOptions FileSystem::OptimizeForCompactionTableRead(
const FileOptions& file_options,
const ImmutableDBOptions& db_options) const {
FileOptions optimized_file_options(file_options);
optimized_file_options.use_direct_reads = db_options.use_direct_reads;
return optimized_file_options;
}
FileOptions FileSystem::OptimizeForBlobFileRead(
const FileOptions& file_options,
const ImmutableDBOptions& db_options) const {
FileOptions optimized_file_options(file_options);
optimized_file_options.use_direct_reads = db_options.use_direct_reads;
return optimized_file_options;
}
IOStatus WriteStringToFile(FileSystem* fs, const Slice& data,
const std::string& fname, bool should_sync) {
std::unique_ptr<FSWritableFile> file;
EnvOptions soptions;
IOStatus s = fs->NewWritableFile(fname, soptions, &file, nullptr);
if (!s.ok()) {
return s;
}
s = file->Append(data, IOOptions(), nullptr);
if (s.ok() && should_sync) {
s = file->Sync(IOOptions(), nullptr);
}
if (!s.ok()) {
fs->DeleteFile(fname, IOOptions(), nullptr);
}
return s;
}
IOStatus ReadFileToString(FileSystem* fs, const std::string& fname,
std::string* data) {
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
FileOptions soptions;
data->clear();
std::unique_ptr<FSSequentialFile> file;
IOStatus s = status_to_io_status(
fs->NewSequentialFile(fname, soptions, &file, nullptr));
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
if (!s.ok()) {
return s;
}
static const int kBufferSize = 8192;
char* space = new char[kBufferSize];
while (true) {
Slice fragment;
s = file->Read(kBufferSize, IOOptions(), &fragment, space,
nullptr);
if (!s.ok()) {
break;
}
data->append(fragment.data(), fragment.size());
if (fragment.empty()) {
break;
}
}
delete[] space;
return s;
}
namespace {
static std::unordered_map<std::string, OptionTypeInfo> fs_wrapper_type_info = {
#ifndef ROCKSDB_LITE
{"target",
OptionTypeInfo::AsCustomSharedPtr<FileSystem>(
0, OptionVerificationType::kByName, OptionTypeFlags::kDontSerialize)},
#endif // ROCKSDB_LITE
};
} // namespace
FileSystemWrapper::FileSystemWrapper(const std::shared_ptr<FileSystem>& t)
: target_(t) {
RegisterOptions("", &target_, &fs_wrapper_type_info);
}
Status FileSystemWrapper::PrepareOptions(const ConfigOptions& options) {
if (target_ == nullptr) {
target_ = FileSystem::Default();
}
return FileSystem::PrepareOptions(options);
}
#ifndef ROCKSDB_LITE
std::string FileSystemWrapper::SerializeOptions(
const ConfigOptions& config_options, const std::string& header) const {
auto parent = FileSystem::SerializeOptions(config_options, "");
if (config_options.IsShallow() || target_ == nullptr ||
target_->IsInstanceOf(FileSystem::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
DirFsyncOptions::DirFsyncOptions() { reason = kDefault; }
DirFsyncOptions::DirFsyncOptions(std::string file_renamed_new_name) {
reason = kFileRenamed;
renamed_new_name = file_renamed_new_name;
}
DirFsyncOptions::DirFsyncOptions(FsyncReason fsync_reason) {
assert(fsync_reason != kFileRenamed);
reason = fsync_reason;
}
} // namespace ROCKSDB_NAMESPACE