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

262 lines
8.7 KiB

// 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.
#pragma once
#include <errno.h>
#include <unistd.h>
#include <atomic>
#include <string>
#include "rocksdb/env.h"
// For non linux platform, the following macros are used only as place
// holder.
#if !(defined OS_LINUX) && !(defined CYGWIN) && !(defined OS_AIX)
#define POSIX_FADV_NORMAL 0 /* [MC1] no further special treatment */
#define POSIX_FADV_RANDOM 1 /* [MC1] expect random page refs */
#define POSIX_FADV_SEQUENTIAL 2 /* [MC1] expect sequential page refs */
#define POSIX_FADV_WILLNEED 3 /* [MC1] will need these pages */
#define POSIX_FADV_DONTNEED 4 /* [MC1] dont need these pages */
#endif
namespace rocksdb {
static std::string IOErrorMsg(const std::string& context,
const std::string& file_name) {
if (file_name.empty()) {
return context;
}
return context + ": " + file_name;
}
// file_name can be left empty if it is not unkown.
static Status IOError(const std::string& context, const std::string& file_name,
int err_number) {
switch (err_number) {
case ENOSPC:
return Status::NoSpace(IOErrorMsg(context, file_name),
strerror(err_number));
case ESTALE:
return Status::IOError(Status::kStaleFile);
Support for single-primary, multi-secondary instances (#4899) Summary: This PR allows RocksDB to run in single-primary, multi-secondary process mode. The writer is a regular RocksDB (e.g. an `DBImpl`) instance playing the role of a primary. Multiple `DBImplSecondary` processes (secondaries) share the same set of SST files, MANIFEST, WAL files with the primary. Secondaries tail the MANIFEST of the primary and apply updates to their own in-memory state of the file system, e.g. `VersionStorageInfo`. This PR has several components: 1. (Originally in #4745). Add a `PathNotFound` subcode to `IOError` to denote the failure when a secondary tries to open a file which has been deleted by the primary. 2. (Similar to #4602). Add `FragmentBufferedReader` to handle partially-read, trailing record at the end of a log from where future read can continue. 3. (Originally in #4710 and #4820). Add implementation of the secondary, i.e. `DBImplSecondary`. 3.1 Tail the primary's MANIFEST during recovery. 3.2 Tail the primary's MANIFEST during normal processing by calling `ReadAndApply`. 3.3 Tailing WAL will be in a future PR. 4. Add an example in 'examples/multi_processes_example.cc' to demonstrate the usage of secondary RocksDB instance in a multi-process setting. Instructions to run the example can be found at the beginning of the source code. Pull Request resolved: https://github.com/facebook/rocksdb/pull/4899 Differential Revision: D14510945 Pulled By: riversand963 fbshipit-source-id: 4ac1c5693e6012ad23f7b4b42d3c374fecbe8886
5 years ago
case ENOENT:
return Status::PathNotFound(IOErrorMsg(context, file_name),
strerror(err_number));
default:
return Status::IOError(IOErrorMsg(context, file_name),
strerror(err_number));
}
}
class PosixHelper {
public:
static size_t GetUniqueIdFromFile(int fd, char* id, size_t max_size);
};
class PosixSequentialFile : public SequentialFile {
private:
std::string filename_;
FILE* file_;
int fd_;
bool use_direct_io_;
size_t logical_sector_size_;
public:
PosixSequentialFile(const std::string& fname, FILE* file, int fd,
const EnvOptions& options);
virtual ~PosixSequentialFile();
virtual Status Read(size_t n, Slice* result, char* scratch) override;
virtual Status PositionedRead(uint64_t offset, size_t n, Slice* result,
char* scratch) override;
virtual Status Skip(uint64_t n) override;
virtual Status InvalidateCache(size_t offset, size_t length) override;
virtual bool use_direct_io() const override { return use_direct_io_; }
virtual size_t GetRequiredBufferAlignment() const override {
return logical_sector_size_;
}
};
class PosixRandomAccessFile : public RandomAccessFile {
protected:
std::string filename_;
int fd_;
bool use_direct_io_;
size_t logical_sector_size_;
public:
PosixRandomAccessFile(const std::string& fname, int fd,
const EnvOptions& options);
virtual ~PosixRandomAccessFile();
virtual Status Read(uint64_t offset, size_t n, Slice* result,
char* scratch) const override;
virtual Status Prefetch(uint64_t offset, size_t n) override;
#if defined(OS_LINUX) || defined(OS_MACOSX) || defined(OS_AIX)
virtual size_t GetUniqueId(char* id, size_t max_size) const override;
#endif
virtual void Hint(AccessPattern pattern) override;
virtual Status InvalidateCache(size_t offset, size_t length) override;
virtual bool use_direct_io() const override { return use_direct_io_; }
virtual size_t GetRequiredBufferAlignment() const override {
return logical_sector_size_;
}
};
class PosixWritableFile : public WritableFile {
protected:
const std::string filename_;
const bool use_direct_io_;
int fd_;
uint64_t filesize_;
size_t logical_sector_size_;
#ifdef ROCKSDB_FALLOCATE_PRESENT
bool allow_fallocate_;
bool fallocate_with_keep_size_;
#endif
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
5 years ago
#ifdef ROCKSDB_RANGESYNC_PRESENT
// Even if the syscall is present, the filesystem may still not properly
// support it, so we need to do a dynamic check too.
bool sync_file_range_supported_;
#endif // ROCKSDB_RANGESYNC_PRESENT
public:
explicit PosixWritableFile(const std::string& fname, int fd,
const EnvOptions& options);
virtual ~PosixWritableFile();
// Need to implement this so the file is truncated correctly
// with direct I/O
virtual Status Truncate(uint64_t size) override;
virtual Status Close() override;
virtual Status Append(const Slice& data) override;
virtual Status PositionedAppend(const Slice& data, uint64_t offset) override;
virtual Status Flush() override;
virtual Status Sync() override;
virtual Status Fsync() override;
virtual bool IsSyncThreadSafe() const override;
virtual bool use_direct_io() const override { return use_direct_io_; }
virtual void SetWriteLifeTimeHint(Env::WriteLifeTimeHint hint) override;
virtual uint64_t GetFileSize() override;
virtual Status InvalidateCache(size_t offset, size_t length) override;
virtual size_t GetRequiredBufferAlignment() const override {
return logical_sector_size_;
}
#ifdef ROCKSDB_FALLOCATE_PRESENT
virtual Status Allocate(uint64_t offset, uint64_t len) override;
#endif
virtual Status RangeSync(uint64_t offset, uint64_t nbytes) override;
#ifdef OS_LINUX
virtual size_t GetUniqueId(char* id, size_t max_size) const override;
#endif
};
// mmap() based random-access
class PosixMmapReadableFile : public RandomAccessFile {
private:
int fd_;
std::string filename_;
void* mmapped_region_;
size_t length_;
public:
PosixMmapReadableFile(const int fd, const std::string& fname, void* base,
size_t length, const EnvOptions& options);
virtual ~PosixMmapReadableFile();
virtual Status Read(uint64_t offset, size_t n, Slice* result,
char* scratch) const override;
virtual Status InvalidateCache(size_t offset, size_t length) override;
};
class PosixMmapFile : public WritableFile {
private:
std::string filename_;
int fd_;
size_t page_size_;
size_t map_size_; // How much extra memory to map at a time
char* base_; // The mapped region
char* limit_; // Limit of the mapped region
char* dst_; // Where to write next (in range [base_,limit_])
char* last_sync_; // Where have we synced up to
uint64_t file_offset_; // Offset of base_ in file
#ifdef ROCKSDB_FALLOCATE_PRESENT
bool allow_fallocate_; // If false, fallocate calls are bypassed
bool fallocate_with_keep_size_;
#endif
// Roundup x to a multiple of y
static size_t Roundup(size_t x, size_t y) { return ((x + y - 1) / y) * y; }
size_t TruncateToPageBoundary(size_t s) {
s -= (s & (page_size_ - 1));
assert((s % page_size_) == 0);
return s;
}
Status MapNewRegion();
Status UnmapCurrentRegion();
Status Msync();
public:
PosixMmapFile(const std::string& fname, int fd, size_t page_size,
const EnvOptions& options);
~PosixMmapFile();
// Means Close() will properly take care of truncate
// and it does not need any additional information
virtual Status Truncate(uint64_t /*size*/) override { return Status::OK(); }
virtual Status Close() override;
virtual Status Append(const Slice& data) override;
virtual Status Flush() override;
virtual Status Sync() override;
virtual Status Fsync() override;
virtual uint64_t GetFileSize() override;
virtual Status InvalidateCache(size_t offset, size_t length) override;
#ifdef ROCKSDB_FALLOCATE_PRESENT
virtual Status Allocate(uint64_t offset, uint64_t len) override;
#endif
};
class PosixRandomRWFile : public RandomRWFile {
public:
explicit PosixRandomRWFile(const std::string& fname, int fd,
const EnvOptions& options);
virtual ~PosixRandomRWFile();
virtual Status Write(uint64_t offset, const Slice& data) override;
virtual Status Read(uint64_t offset, size_t n, Slice* result,
char* scratch) const override;
virtual Status Flush() override;
virtual Status Sync() override;
virtual Status Fsync() override;
virtual Status Close() override;
private:
const std::string filename_;
int fd_;
};
struct PosixMemoryMappedFileBuffer : public MemoryMappedFileBuffer {
PosixMemoryMappedFileBuffer(void* _base, size_t _length)
: MemoryMappedFileBuffer(_base, _length) {}
virtual ~PosixMemoryMappedFileBuffer();
};
class PosixDirectory : public Directory {
public:
explicit PosixDirectory(int fd) : fd_(fd) {}
~PosixDirectory();
virtual Status Fsync() override;
private:
int fd_;
};
} // namespace rocksdb