mirror of
https://github.com/facebook/rocksdb.git
synced 2024-11-27 20:43:57 +00:00
4da34b97ee
Summary: ### Context: Background compactions and flush generate large reads and writes, and can be long running, especially for universal compaction. In some cases, this can impact foreground reads and writes by users. ### Solution User, Flush, and Compaction reads share some code path. For this task, we update the rate_limiter_priority in ReadOptions for code paths (e.g. FindTable (mainly in BlockBasedTable::Open()) and various iterators), and eventually update the rate_limiter_priority in IOOptions for FSRandomAccessFile. **This PR is for the Read path.** The **Read:** dynamic priority for different state are listed as follows: | State | Normal | Delayed | Stalled | | ----- | ------ | ------- | ------- | | Flush (verification read in BuildTable()) | IO_USER | IO_USER | IO_USER | | Compaction | IO_LOW | IO_USER | IO_USER | | User | User provided | User provided | User provided | We will respect the read_options that the user provided and will not set it. The only sst read for Flush is the verification read in BuildTable(). It claims to be "regard as user read". **Details** 1. Set read_options.rate_limiter_priority dynamically: - User: Do not update the read_options. Use the read_options that the user provided. - Compaction: Update read_options in CompactionJob::ProcessKeyValueCompaction(). - Flush: Update read_options in BuildTable(). 2. Pass the rate limiter priority to FSRandomAccessFile functions: - After calling the FindTable(), read_options is passed through GetTableReader(table_cache.cc), BlockBasedTableFactory::NewTableReader(block_based_table_factory.cc), and BlockBasedTable::Open(). The Open() needs some updates for the ReadOptions variable and the updates are also needed for the called functions, including PrefetchTail(), PrepareIOOptions(), ReadFooterFromFile(), ReadMetaIndexblock(), ReadPropertiesBlock(), PrefetchIndexAndFilterBlocks(), and ReadRangeDelBlock(). - In RandomAccessFileReader, the functions to be updated include Read(), MultiRead(), ReadAsync(), and Prefetch(). - Update the downstream functions of NewIndexIterator(), NewDataBlockIterator(), and BlockBasedTableIterator(). ### Test Plans Add unit tests. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9996 Reviewed By: anand1976 Differential Revision: D36452483 Pulled By: gitbw95 fbshipit-source-id: 60978204a4f849bb9261cb78d9bc1cb56d6008cf
198 lines
7.3 KiB
C++
198 lines
7.3 KiB
C++
// 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 <atomic>
|
|
#include <sstream>
|
|
#include <string>
|
|
|
|
#include "env/file_system_tracer.h"
|
|
#include "port/port.h"
|
|
#include "rocksdb/file_system.h"
|
|
#include "rocksdb/listener.h"
|
|
#include "rocksdb/options.h"
|
|
#include "rocksdb/rate_limiter.h"
|
|
#include "util/aligned_buffer.h"
|
|
|
|
namespace ROCKSDB_NAMESPACE {
|
|
class Statistics;
|
|
class HistogramImpl;
|
|
class SystemClock;
|
|
|
|
using AlignedBuf = std::unique_ptr<char[]>;
|
|
|
|
// Align the request r according to alignment and return the aligned result.
|
|
FSReadRequest Align(const FSReadRequest& r, size_t alignment);
|
|
|
|
// Try to merge src to dest if they have overlap.
|
|
//
|
|
// Each request represents an inclusive interval [offset, offset + len].
|
|
// If the intervals have overlap, update offset and len to represent the
|
|
// merged interval, and return true.
|
|
// Otherwise, do nothing and return false.
|
|
bool TryMerge(FSReadRequest* dest, const FSReadRequest& src);
|
|
|
|
// RandomAccessFileReader is a wrapper on top of Env::RandomAccessFile. It is
|
|
// responsible for:
|
|
// - Handling Buffered and Direct reads appropriately.
|
|
// - Rate limiting compaction reads.
|
|
// - Notifying any interested listeners on the completion of a read.
|
|
// - Updating IO stats.
|
|
class RandomAccessFileReader {
|
|
private:
|
|
#ifndef ROCKSDB_LITE
|
|
void NotifyOnFileReadFinish(
|
|
uint64_t offset, size_t length,
|
|
const FileOperationInfo::StartTimePoint& start_ts,
|
|
const FileOperationInfo::FinishTimePoint& finish_ts,
|
|
const Status& status) const {
|
|
FileOperationInfo info(FileOperationType::kRead, file_name_, start_ts,
|
|
finish_ts, status, file_temperature_);
|
|
info.offset = offset;
|
|
info.length = length;
|
|
|
|
for (auto& listener : listeners_) {
|
|
listener->OnFileReadFinish(info);
|
|
}
|
|
info.status.PermitUncheckedError();
|
|
}
|
|
|
|
void NotifyOnIOError(const IOStatus& io_status, FileOperationType operation,
|
|
const std::string& file_path, size_t length,
|
|
uint64_t offset) const {
|
|
if (listeners_.empty()) {
|
|
return;
|
|
}
|
|
IOErrorInfo io_error_info(io_status, operation, file_path, length, offset);
|
|
|
|
for (auto& listener : listeners_) {
|
|
listener->OnIOError(io_error_info);
|
|
}
|
|
io_status.PermitUncheckedError();
|
|
}
|
|
|
|
#endif // ROCKSDB_LITE
|
|
|
|
bool ShouldNotifyListeners() const { return !listeners_.empty(); }
|
|
|
|
FSRandomAccessFilePtr file_;
|
|
std::string file_name_;
|
|
SystemClock* clock_;
|
|
Statistics* stats_;
|
|
uint32_t hist_type_;
|
|
HistogramImpl* file_read_hist_;
|
|
RateLimiter* rate_limiter_;
|
|
std::vector<std::shared_ptr<EventListener>> listeners_;
|
|
const Temperature file_temperature_;
|
|
const bool is_last_level_;
|
|
|
|
struct ReadAsyncInfo {
|
|
#ifndef ROCKSDB_LITE
|
|
FileOperationInfo::StartTimePoint fs_start_ts_;
|
|
#endif
|
|
uint64_t start_time_;
|
|
std::function<void(const FSReadRequest&, void*)> cb_;
|
|
void* cb_arg_;
|
|
};
|
|
|
|
public:
|
|
explicit RandomAccessFileReader(
|
|
std::unique_ptr<FSRandomAccessFile>&& raf, const std::string& _file_name,
|
|
SystemClock* clock = nullptr,
|
|
const std::shared_ptr<IOTracer>& io_tracer = nullptr,
|
|
Statistics* stats = nullptr, uint32_t hist_type = 0,
|
|
HistogramImpl* file_read_hist = nullptr,
|
|
RateLimiter* rate_limiter = nullptr,
|
|
const std::vector<std::shared_ptr<EventListener>>& listeners = {},
|
|
Temperature file_temperature = Temperature::kUnknown,
|
|
bool is_last_level = false)
|
|
: file_(std::move(raf), io_tracer, _file_name),
|
|
file_name_(std::move(_file_name)),
|
|
clock_(clock),
|
|
stats_(stats),
|
|
hist_type_(hist_type),
|
|
file_read_hist_(file_read_hist),
|
|
rate_limiter_(rate_limiter),
|
|
listeners_(),
|
|
file_temperature_(file_temperature),
|
|
is_last_level_(is_last_level) {
|
|
#ifndef ROCKSDB_LITE
|
|
std::for_each(listeners.begin(), listeners.end(),
|
|
[this](const std::shared_ptr<EventListener>& e) {
|
|
if (e->ShouldBeNotifiedOnFileIO()) {
|
|
listeners_.emplace_back(e);
|
|
}
|
|
});
|
|
#else // !ROCKSDB_LITE
|
|
(void)listeners;
|
|
#endif
|
|
}
|
|
|
|
static IOStatus Create(const std::shared_ptr<FileSystem>& fs,
|
|
const std::string& fname, const FileOptions& file_opts,
|
|
std::unique_ptr<RandomAccessFileReader>* reader,
|
|
IODebugContext* dbg);
|
|
RandomAccessFileReader(const RandomAccessFileReader&) = delete;
|
|
RandomAccessFileReader& operator=(const RandomAccessFileReader&) = delete;
|
|
|
|
// In non-direct IO mode,
|
|
// 1. if using mmap, result is stored in a buffer other than scratch;
|
|
// 2. if not using mmap, result is stored in the buffer starting from scratch.
|
|
//
|
|
// In direct IO mode, an aligned buffer is allocated internally.
|
|
// 1. If aligned_buf is null, then results are copied to the buffer
|
|
// starting from scratch;
|
|
// 2. Otherwise, scratch is not used and can be null, the aligned_buf owns
|
|
// the internally allocated buffer on return, and the result refers to a
|
|
// region in aligned_buf.
|
|
//
|
|
// `rate_limiter_priority` is used to charge the internal rate limiter when
|
|
// enabled. The special value `Env::IO_TOTAL` makes this operation bypass the
|
|
// rate limiter.
|
|
IOStatus Read(const IOOptions& opts, uint64_t offset, size_t n, Slice* result,
|
|
char* scratch, AlignedBuf* aligned_buf,
|
|
Env::IOPriority rate_limiter_priority) const;
|
|
|
|
// REQUIRES:
|
|
// num_reqs > 0, reqs do not overlap, and offsets in reqs are increasing.
|
|
// In non-direct IO mode, aligned_buf should be null;
|
|
// In direct IO mode, aligned_buf stores the aligned buffer allocated inside
|
|
// MultiRead, the result Slices in reqs refer to aligned_buf.
|
|
//
|
|
// `rate_limiter_priority` will be used to charge the internal rate limiter.
|
|
// It is not yet supported so the client must provide the special value
|
|
// `Env::IO_TOTAL` to bypass the rate limiter.
|
|
IOStatus MultiRead(const IOOptions& opts, FSReadRequest* reqs,
|
|
size_t num_reqs, AlignedBuf* aligned_buf,
|
|
Env::IOPriority rate_limiter_priority) const;
|
|
|
|
IOStatus Prefetch(uint64_t offset, size_t n,
|
|
const Env::IOPriority rate_limiter_priority) const {
|
|
IOOptions opts;
|
|
opts.rate_limiter_priority = rate_limiter_priority;
|
|
return file_->Prefetch(offset, n, opts, nullptr);
|
|
}
|
|
|
|
FSRandomAccessFile* file() { return file_.get(); }
|
|
|
|
const std::string& file_name() const { return file_name_; }
|
|
|
|
bool use_direct_io() const { return file_->use_direct_io(); }
|
|
|
|
IOStatus PrepareIOOptions(const ReadOptions& ro, IOOptions& opts);
|
|
|
|
IOStatus ReadAsync(FSReadRequest& req, const IOOptions& opts,
|
|
std::function<void(const FSReadRequest&, void*)> cb,
|
|
void* cb_arg, void** io_handle, IOHandleDeleter* del_fn,
|
|
Env::IOPriority rate_limiter_priority);
|
|
|
|
void ReadAsyncCallback(const FSReadRequest& req, void* cb_arg);
|
|
};
|
|
} // namespace ROCKSDB_NAMESPACE
|