mirror of
https://github.com/facebook/rocksdb.git
synced 2024-11-30 22:41:48 +00:00
9a034801ce
Summary:
**Context/Summary:**
- Similar to https://github.com/facebook/rocksdb/pull/11288 but for user read such as `Get(), MultiGet(), DBIterator::XXX(), Verify(File)Checksum()`.
- For this, I refactored some user-facing `MultiGet` calls in `TransactionBase` and various types of `DB` so that it does not call a user-facing `Get()` but `GetImpl()` for passing the `ReadOptions::io_activity` check (see PR conversation)
- New user read stats breakdown are guarded by `kExceptDetailedTimers` since measurement shows they have 4-5% regression to the upstream/main.
- Misc
- More refactoring: with https://github.com/facebook/rocksdb/pull/11288, we complete passing `ReadOptions/IOOptions` to FS level. So we can now replace the previously [added](https://github.com/facebook/rocksdb/pull/9424) `rate_limiter_priority` parameter in `RandomAccessFileReader`'s `Read/MultiRead/Prefetch()` with `IOOptions::rate_limiter_priority`
- Also, `ReadAsync()` call time is measured in `SST_READ_MICRO` now
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11444
Test Plan:
- CI fake db crash/stress test
- Microbenchmarking
**Build** `make clean && ROCKSDB_NO_FBCODE=1 DEBUG_LEVEL=0 make -jN db_basic_bench`
- google benchmark version: 604f6fd3f4
- db_basic_bench_base: upstream
- db_basic_bench_pr: db_basic_bench_base + this PR
- asyncread_db_basic_bench_base: upstream + [db basic bench patch for IteratorNext](https://github.com/facebook/rocksdb/compare/main...hx235:rocksdb:micro_bench_async_read)
- asyncread_db_basic_bench_pr: asyncread_db_basic_bench_base + this PR
**Test**
Get
```
TEST_TMPDIR=/dev/shm ./db_basic_bench_{null_stat|base|pr} --benchmark_filter=DBGet/comp_style:0/max_data:134217728/per_key_size:256/enable_statistics:1/negative_query:0/enable_filter:0/mmap:1/threads:1 --benchmark_repetitions=1000
```
Result
```
Coming soon
```
AsyncRead
```
TEST_TMPDIR=/dev/shm ./asyncread_db_basic_bench_{base|pr} --benchmark_filter=IteratorNext/comp_style:0/max_data:134217728/per_key_size:256/enable_statistics:1/async_io:1/include_detailed_timers:0 --benchmark_repetitions=1000 > syncread_db_basic_bench_{base|pr}.out
```
Result
```
Base:
1956,1956,1968,1977,1979,1986,1988,1988,1988,1990,1991,1991,1993,1993,1993,1993,1994,1996,1997,1997,1997,1998,1999,2001,2001,2002,2004,2007,2007,2008,
PR (2.3% regression, due to measuring `SST_READ_MICRO` that wasn't measured before):
1993,2014,2016,2022,2024,2027,2027,2028,2028,2030,2031,2031,2032,2032,2038,2039,2042,2044,2044,2047,2047,2047,2048,2049,2050,2052,2052,2052,2053,2053,
```
Reviewed By: ajkr
Differential Revision: D45918925
Pulled By: hx235
fbshipit-source-id: 58a54560d9ebeb3a59b6d807639692614dad058a
282 lines
9.5 KiB
C++
282 lines
9.5 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).
|
|
//
|
|
#include "file/file_util.h"
|
|
|
|
#include <algorithm>
|
|
#include <string>
|
|
|
|
#include "file/random_access_file_reader.h"
|
|
#include "file/sequence_file_reader.h"
|
|
#include "file/sst_file_manager_impl.h"
|
|
#include "file/writable_file_writer.h"
|
|
#include "rocksdb/env.h"
|
|
|
|
namespace ROCKSDB_NAMESPACE {
|
|
|
|
// Utility function to copy a file up to a specified length
|
|
IOStatus CopyFile(FileSystem* fs, const std::string& source,
|
|
std::unique_ptr<WritableFileWriter>& dest_writer,
|
|
uint64_t size, bool use_fsync,
|
|
const std::shared_ptr<IOTracer>& io_tracer,
|
|
const Temperature temperature) {
|
|
FileOptions soptions;
|
|
IOStatus io_s;
|
|
std::unique_ptr<SequentialFileReader> src_reader;
|
|
|
|
{
|
|
soptions.temperature = temperature;
|
|
std::unique_ptr<FSSequentialFile> srcfile;
|
|
io_s = fs->NewSequentialFile(source, soptions, &srcfile, nullptr);
|
|
if (!io_s.ok()) {
|
|
return io_s;
|
|
}
|
|
|
|
if (size == 0) {
|
|
// default argument means copy everything
|
|
io_s = fs->GetFileSize(source, IOOptions(), &size, nullptr);
|
|
if (!io_s.ok()) {
|
|
return io_s;
|
|
}
|
|
}
|
|
src_reader.reset(
|
|
new SequentialFileReader(std::move(srcfile), source, io_tracer));
|
|
}
|
|
|
|
char buffer[4096];
|
|
Slice slice;
|
|
while (size > 0) {
|
|
size_t bytes_to_read = std::min(sizeof(buffer), static_cast<size_t>(size));
|
|
// TODO: rate limit copy file
|
|
io_s = status_to_io_status(
|
|
src_reader->Read(bytes_to_read, &slice, buffer,
|
|
Env::IO_TOTAL /* rate_limiter_priority */));
|
|
if (!io_s.ok()) {
|
|
return io_s;
|
|
}
|
|
if (slice.size() == 0) {
|
|
return IOStatus::Corruption("file too small");
|
|
}
|
|
io_s = dest_writer->Append(slice);
|
|
if (!io_s.ok()) {
|
|
return io_s;
|
|
}
|
|
size -= slice.size();
|
|
}
|
|
return dest_writer->Sync(use_fsync);
|
|
}
|
|
|
|
IOStatus CopyFile(FileSystem* fs, const std::string& source,
|
|
const std::string& destination, uint64_t size, bool use_fsync,
|
|
const std::shared_ptr<IOTracer>& io_tracer,
|
|
const Temperature temperature) {
|
|
FileOptions options;
|
|
IOStatus io_s;
|
|
std::unique_ptr<WritableFileWriter> dest_writer;
|
|
|
|
{
|
|
options.temperature = temperature;
|
|
std::unique_ptr<FSWritableFile> destfile;
|
|
io_s = fs->NewWritableFile(destination, options, &destfile, nullptr);
|
|
if (!io_s.ok()) {
|
|
return io_s;
|
|
}
|
|
|
|
dest_writer.reset(
|
|
new WritableFileWriter(std::move(destfile), destination, options));
|
|
}
|
|
|
|
return CopyFile(fs, source, dest_writer, size, use_fsync, io_tracer,
|
|
temperature);
|
|
}
|
|
|
|
// Utility function to create a file with the provided contents
|
|
IOStatus CreateFile(FileSystem* fs, const std::string& destination,
|
|
const std::string& contents, bool use_fsync) {
|
|
const EnvOptions soptions;
|
|
IOStatus io_s;
|
|
std::unique_ptr<WritableFileWriter> dest_writer;
|
|
|
|
std::unique_ptr<FSWritableFile> destfile;
|
|
io_s = fs->NewWritableFile(destination, soptions, &destfile, nullptr);
|
|
if (!io_s.ok()) {
|
|
return io_s;
|
|
}
|
|
dest_writer.reset(
|
|
new WritableFileWriter(std::move(destfile), destination, soptions));
|
|
io_s = dest_writer->Append(Slice(contents));
|
|
if (!io_s.ok()) {
|
|
return io_s;
|
|
}
|
|
return dest_writer->Sync(use_fsync);
|
|
}
|
|
|
|
Status DeleteDBFile(const ImmutableDBOptions* db_options,
|
|
const std::string& fname, const std::string& dir_to_sync,
|
|
const bool force_bg, const bool force_fg) {
|
|
SstFileManagerImpl* sfm =
|
|
static_cast<SstFileManagerImpl*>(db_options->sst_file_manager.get());
|
|
if (sfm && !force_fg) {
|
|
return sfm->ScheduleFileDeletion(fname, dir_to_sync, force_bg);
|
|
} else {
|
|
return db_options->env->DeleteFile(fname);
|
|
}
|
|
}
|
|
|
|
// requested_checksum_func_name brings the function name of the checksum
|
|
// generator in checksum_factory. Empty string is permitted, in which case the
|
|
// name of the generator created by the factory is unchecked. When
|
|
// `requested_checksum_func_name` is non-empty, however, the created generator's
|
|
// name must match it, otherwise an `InvalidArgument` error is returned.
|
|
IOStatus GenerateOneFileChecksum(
|
|
FileSystem* fs, const std::string& file_path,
|
|
FileChecksumGenFactory* checksum_factory,
|
|
const std::string& requested_checksum_func_name, std::string* file_checksum,
|
|
std::string* file_checksum_func_name,
|
|
size_t verify_checksums_readahead_size, bool /*allow_mmap_reads*/,
|
|
std::shared_ptr<IOTracer>& io_tracer, RateLimiter* rate_limiter,
|
|
const ReadOptions& read_options) {
|
|
if (checksum_factory == nullptr) {
|
|
return IOStatus::InvalidArgument("Checksum factory is invalid");
|
|
}
|
|
assert(file_checksum != nullptr);
|
|
assert(file_checksum_func_name != nullptr);
|
|
|
|
FileChecksumGenContext gen_context;
|
|
gen_context.requested_checksum_func_name = requested_checksum_func_name;
|
|
gen_context.file_name = file_path;
|
|
std::unique_ptr<FileChecksumGenerator> checksum_generator =
|
|
checksum_factory->CreateFileChecksumGenerator(gen_context);
|
|
if (checksum_generator == nullptr) {
|
|
std::string msg =
|
|
"Cannot get the file checksum generator based on the requested "
|
|
"checksum function name: " +
|
|
requested_checksum_func_name +
|
|
" from checksum factory: " + checksum_factory->Name();
|
|
return IOStatus::InvalidArgument(msg);
|
|
} else {
|
|
// For backward compatibility and use in file ingestion clients where there
|
|
// is no stored checksum function name, `requested_checksum_func_name` can
|
|
// be empty. If we give the requested checksum function name, we expect it
|
|
// is the same name of the checksum generator.
|
|
if (!requested_checksum_func_name.empty() &&
|
|
checksum_generator->Name() != requested_checksum_func_name) {
|
|
std::string msg = "Expected file checksum generator named '" +
|
|
requested_checksum_func_name +
|
|
"', while the factory created one "
|
|
"named '" +
|
|
checksum_generator->Name() + "'";
|
|
return IOStatus::InvalidArgument(msg);
|
|
}
|
|
}
|
|
|
|
uint64_t size;
|
|
IOStatus io_s;
|
|
std::unique_ptr<RandomAccessFileReader> reader;
|
|
{
|
|
std::unique_ptr<FSRandomAccessFile> r_file;
|
|
io_s = fs->NewRandomAccessFile(file_path, FileOptions(), &r_file, nullptr);
|
|
if (!io_s.ok()) {
|
|
return io_s;
|
|
}
|
|
io_s = fs->GetFileSize(file_path, IOOptions(), &size, nullptr);
|
|
if (!io_s.ok()) {
|
|
return io_s;
|
|
}
|
|
reader.reset(new RandomAccessFileReader(
|
|
std::move(r_file), file_path, nullptr /*Env*/, io_tracer, nullptr,
|
|
Histograms::HISTOGRAM_ENUM_MAX, nullptr, rate_limiter));
|
|
}
|
|
|
|
// Found that 256 KB readahead size provides the best performance, based on
|
|
// experiments, for auto readahead. Experiment data is in PR #3282.
|
|
size_t default_max_read_ahead_size = 256 * 1024;
|
|
size_t readahead_size = (verify_checksums_readahead_size != 0)
|
|
? verify_checksums_readahead_size
|
|
: default_max_read_ahead_size;
|
|
std::unique_ptr<char[]> buf;
|
|
if (reader->use_direct_io()) {
|
|
size_t alignment = reader->file()->GetRequiredBufferAlignment();
|
|
readahead_size = (readahead_size + alignment - 1) & ~(alignment - 1);
|
|
}
|
|
buf.reset(new char[readahead_size]);
|
|
|
|
Slice slice;
|
|
uint64_t offset = 0;
|
|
IOOptions opts;
|
|
io_s = reader->PrepareIOOptions(read_options, opts);
|
|
if (!io_s.ok()) {
|
|
return io_s;
|
|
}
|
|
while (size > 0) {
|
|
size_t bytes_to_read =
|
|
static_cast<size_t>(std::min(uint64_t{readahead_size}, size));
|
|
io_s =
|
|
reader->Read(opts, offset, bytes_to_read, &slice, buf.get(), nullptr);
|
|
if (!io_s.ok()) {
|
|
return IOStatus::Corruption("file read failed with error: " +
|
|
io_s.ToString());
|
|
}
|
|
if (slice.size() == 0) {
|
|
return IOStatus::Corruption("file too small");
|
|
}
|
|
checksum_generator->Update(slice.data(), slice.size());
|
|
size -= slice.size();
|
|
offset += slice.size();
|
|
|
|
TEST_SYNC_POINT("GenerateOneFileChecksum::Chunk:0");
|
|
}
|
|
checksum_generator->Finalize();
|
|
*file_checksum = checksum_generator->GetChecksum();
|
|
*file_checksum_func_name = checksum_generator->Name();
|
|
return IOStatus::OK();
|
|
}
|
|
|
|
Status DestroyDir(Env* env, const std::string& dir) {
|
|
Status s;
|
|
if (env->FileExists(dir).IsNotFound()) {
|
|
return s;
|
|
}
|
|
std::vector<std::string> files_in_dir;
|
|
s = env->GetChildren(dir, &files_in_dir);
|
|
if (s.ok()) {
|
|
for (auto& file_in_dir : files_in_dir) {
|
|
std::string path = dir + "/" + file_in_dir;
|
|
bool is_dir = false;
|
|
s = env->IsDirectory(path, &is_dir);
|
|
if (s.ok()) {
|
|
if (is_dir) {
|
|
s = DestroyDir(env, path);
|
|
} else {
|
|
s = env->DeleteFile(path);
|
|
}
|
|
} else if (s.IsNotSupported()) {
|
|
s = Status::OK();
|
|
}
|
|
if (!s.ok()) {
|
|
// IsDirectory, etc. might not report NotFound
|
|
if (s.IsNotFound() || env->FileExists(path).IsNotFound()) {
|
|
// Allow files to be deleted externally
|
|
s = Status::OK();
|
|
} else {
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
if (s.ok()) {
|
|
s = env->DeleteDir(dir);
|
|
// DeleteDir might or might not report NotFound
|
|
if (!s.ok() && (s.IsNotFound() || env->FileExists(dir).IsNotFound())) {
|
|
// Allow to be deleted externally
|
|
s = Status::OK();
|
|
}
|
|
}
|
|
return s;
|
|
}
|
|
|
|
} // namespace ROCKSDB_NAMESPACE
|