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/env_posix.cc

521 lines
17 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
#include "port/lang.h"
#if !defined(OS_WIN)
#include <dirent.h>
#ifndef ROCKSDB_NO_DYNAMIC_EXTENSION
#include <dlfcn.h>
#endif
#include <errno.h>
#include <fcntl.h>
#if defined(ROCKSDB_IOURING_PRESENT)
#include <liburing.h>
#endif
#include <pthread.h>
#include <signal.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/mman.h>
#include <sys/stat.h>
#if defined(OS_LINUX) || defined(OS_SOLARIS) || defined(OS_ANDROID)
#include <sys/statfs.h>
#endif
Auto recovery from out of space errors (#4164) Summary: This commit implements automatic recovery from a Status::NoSpace() error during background operations such as write callback, flush and compaction. The broad design is as follows - 1. Compaction errors are treated as soft errors and don't put the database in read-only mode. A compaction is delayed until enough free disk space is available to accomodate the compaction outputs, which is estimated based on the input size. This means that users can continue to write, and we rely on the WriteController to delay or stop writes if the compaction debt becomes too high due to persistent low disk space condition 2. Errors during write callback and flush are treated as hard errors, i.e the database is put in read-only mode and goes back to read-write only fater certain recovery actions are taken. 3. Both types of recovery rely on the SstFileManagerImpl to poll for sufficient disk space. We assume that there is a 1-1 mapping between an SFM and the underlying OS storage container. For cases where multiple DBs are hosted on a single storage container, the user is expected to allocate a single SFM instance and use the same one for all the DBs. If no SFM is specified by the user, DBImpl::Open() will allocate one, but this will be one per DB and each DB will recover independently. The recovery implemented by SFM is as follows - a) On the first occurance of an out of space error during compaction, subsequent compactions will be delayed until the disk free space check indicates enough available space. The required space is computed as the sum of input sizes. b) The free space check requirement will be removed once the amount of free space is greater than the size reserved by in progress compactions when the first error occured c) If the out of space error is a hard error, a background thread in SFM will poll for sufficient headroom before triggering the recovery of the database and putting it in write-only mode. The headroom is calculated as the sum of the write_buffer_size of all the DB instances associated with the SFM 4. EventListener callbacks will be called at the start and completion of automatic recovery. Users can disable the auto recov ery in the start callback, and later initiate it manually by calling DB::Resume() Todo: 1. More extensive testing 2. Add disk full condition to db_stress (follow-on PR) Pull Request resolved: https://github.com/facebook/rocksdb/pull/4164 Differential Revision: D9846378 Pulled By: anand1976 fbshipit-source-id: 80ea875dbd7f00205e19c82215ff6e37da10da4a
6 years ago
#include <sys/statvfs.h>
#include <sys/time.h>
#include <sys/types.h>
#if defined(ROCKSDB_IOURING_PRESENT)
#include <sys/uio.h>
#endif
#include <time.h>
#include <unistd.h>
#include <algorithm>
// Get nano time includes
#if defined(OS_LINUX) || defined(OS_FREEBSD) || defined(OS_GNU_KFREEBSD)
#elif defined(__MACH__)
#include <Availability.h>
#include <mach/clock.h>
#include <mach/mach.h>
#else
#include <chrono>
#endif
#include <deque>
#include <set>
#include <vector>
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
5 years ago
#include "env/composite_env_wrapper.h"
#include "env/io_posix.h"
#include "monitoring/iostats_context_imp.h"
#include "monitoring/thread_status_updater.h"
#include "port/port.h"
#include "port/sys_time.h"
#include "rocksdb/env.h"
#include "rocksdb/options.h"
#include "rocksdb/slice.h"
#include "rocksdb/system_clock.h"
#include "test_util/sync_point.h"
#include "util/coding.h"
#include "util/compression_context_cache.h"
#include "util/random.h"
#include "util/string_util.h"
#include "util/thread_local.h"
#include "util/threadpool_imp.h"
#if !defined(TMPFS_MAGIC)
#define TMPFS_MAGIC 0x01021994
#endif
#if !defined(XFS_SUPER_MAGIC)
#define XFS_SUPER_MAGIC 0x58465342
#endif
#if !defined(EXT4_SUPER_MAGIC)
#define EXT4_SUPER_MAGIC 0xEF53
#endif
namespace ROCKSDB_NAMESPACE {
#if defined(OS_WIN)
static const std::string kSharedLibExt = ".dll";
static const char kPathSeparator = ';';
#else
static const char kPathSeparator = ':';
#if defined(OS_MACOSX)
static const std::string kSharedLibExt = ".dylib";
#else
static const std::string kSharedLibExt = ".so";
#endif
#endif
namespace {
ThreadStatusUpdater* CreateThreadStatusUpdater() {
return new ThreadStatusUpdater();
}
#ifndef ROCKSDB_NO_DYNAMIC_EXTENSION
class PosixDynamicLibrary : public DynamicLibrary {
public:
PosixDynamicLibrary(const std::string& name, void* handle)
: name_(name), handle_(handle) {}
~PosixDynamicLibrary() override { dlclose(handle_); }
Fix tsan error (#5414) Summary: Previous code has a warning when compile with tsan, leading to an error since we have -Werror. Compilation result ``` In file included from ./env/env_chroot.h:12, from env/env_test.cc:40: ./include/rocksdb/env.h: In instantiation of ‘rocksdb::Status rocksdb::DynamicLibrary::LoadFunction(const string&, std::function<T>*) [with T = void*(void*, const char*); std::__cxx11::string = std::__cxx11::basic_string<char>]’: env/env_test.cc:260:5: required from here ./include/rocksdb/env.h:1010:17: error: cast between incompatible function types from ‘rocksdb::DynamicLibrary::FunctionPtr’ {aka ‘void* (*)()’} to ‘void* (*)(void*, const char*)’ [-Werror=cast-function-type] *function = reinterpret_cast<T*>(ptr); ^~~~~~~~~~~~~~~~~~~~~~~~~ cc1plus: all warnings being treated as errors make: *** [env/env_test.o] Error 1 ``` It also has another error reported by clang ``` env/env_posix.cc:141:11: warning: Value stored to 'err' during its initialization is never read char* err = dlerror(); // Clear any old error ^~~ ~~~~~~~~~ 1 warning generated. ``` Test plan (on my devserver). ``` $make clean $OPT=-g ROCKSDB_FBCODE_BUILD_WITH_PLATFORM007=1 COMPILE_WITH_TSAN=1 make -j32 $ $make clean $USE_CLANG=1 TEST_TMPDIR=/dev/shm/rocksdb OPT=-g make -j1 analyze ``` Both should pass. Pull Request resolved: https://github.com/facebook/rocksdb/pull/5414 Differential Revision: D15637315 Pulled By: riversand963 fbshipit-source-id: 8e307483761019a4d5998cab92d49516d7edffbf
5 years ago
Status LoadSymbol(const std::string& sym_name, void** func) override {
assert(nullptr != func);
dlerror(); // Clear any old error
*func = dlsym(handle_, sym_name.c_str());
if (*func != nullptr) {
return Status::OK();
} else {
Fix tsan error (#5414) Summary: Previous code has a warning when compile with tsan, leading to an error since we have -Werror. Compilation result ``` In file included from ./env/env_chroot.h:12, from env/env_test.cc:40: ./include/rocksdb/env.h: In instantiation of ‘rocksdb::Status rocksdb::DynamicLibrary::LoadFunction(const string&, std::function<T>*) [with T = void*(void*, const char*); std::__cxx11::string = std::__cxx11::basic_string<char>]’: env/env_test.cc:260:5: required from here ./include/rocksdb/env.h:1010:17: error: cast between incompatible function types from ‘rocksdb::DynamicLibrary::FunctionPtr’ {aka ‘void* (*)()’} to ‘void* (*)(void*, const char*)’ [-Werror=cast-function-type] *function = reinterpret_cast<T*>(ptr); ^~~~~~~~~~~~~~~~~~~~~~~~~ cc1plus: all warnings being treated as errors make: *** [env/env_test.o] Error 1 ``` It also has another error reported by clang ``` env/env_posix.cc:141:11: warning: Value stored to 'err' during its initialization is never read char* err = dlerror(); // Clear any old error ^~~ ~~~~~~~~~ 1 warning generated. ``` Test plan (on my devserver). ``` $make clean $OPT=-g ROCKSDB_FBCODE_BUILD_WITH_PLATFORM007=1 COMPILE_WITH_TSAN=1 make -j32 $ $make clean $USE_CLANG=1 TEST_TMPDIR=/dev/shm/rocksdb OPT=-g make -j1 analyze ``` Both should pass. Pull Request resolved: https://github.com/facebook/rocksdb/pull/5414 Differential Revision: D15637315 Pulled By: riversand963 fbshipit-source-id: 8e307483761019a4d5998cab92d49516d7edffbf
5 years ago
char* err = dlerror();
return Status::NotFound("Error finding symbol: " + sym_name, err);
}
}
const char* Name() const override { return name_.c_str(); }
private:
std::string name_;
void* handle_;
};
#endif // !ROCKSDB_NO_DYNAMIC_EXTENSION
class PosixClock : public SystemClock {
public:
static const char* kClassName() { return "PosixClock"; }
const char* Name() const override { return kDefaultName(); }
const char* NickName() const override { return kClassName(); }
uint64_t NowMicros() override {
port::TimeVal tv;
port::GetTimeOfDay(&tv, nullptr);
return static_cast<uint64_t>(tv.tv_sec) * 1000000 + tv.tv_usec;
}
uint64_t NowNanos() override {
#if defined(OS_LINUX) || defined(OS_FREEBSD) || defined(OS_GNU_KFREEBSD) || \
defined(OS_AIX)
struct timespec ts;
clock_gettime(CLOCK_MONOTONIC, &ts);
return static_cast<uint64_t>(ts.tv_sec) * 1000000000 + ts.tv_nsec;
#elif defined(OS_SOLARIS)
return gethrtime();
#elif defined(__MACH__)
clock_serv_t cclock;
mach_timespec_t ts;
host_get_clock_service(mach_host_self(), CALENDAR_CLOCK, &cclock);
clock_get_time(cclock, &ts);
mach_port_deallocate(mach_task_self(), cclock);
return static_cast<uint64_t>(ts.tv_sec) * 1000000000 + ts.tv_nsec;
#else
return std::chrono::duration_cast<std::chrono::nanoseconds>(
std::chrono::steady_clock::now().time_since_epoch())
.count();
#endif
}
uint64_t CPUMicros() override {
#if defined(OS_LINUX) || defined(OS_FREEBSD) || defined(OS_GNU_KFREEBSD) || \
defined(OS_AIX) || (defined(__MACH__) && defined(__MAC_10_12))
struct timespec ts;
clock_gettime(CLOCK_THREAD_CPUTIME_ID, &ts);
return (static_cast<uint64_t>(ts.tv_sec) * 1000000000 + ts.tv_nsec) / 1000;
#endif
return 0;
}
uint64_t CPUNanos() override {
#if defined(OS_LINUX) || defined(OS_FREEBSD) || defined(OS_GNU_KFREEBSD) || \
defined(OS_AIX) || (defined(__MACH__) && defined(__MAC_10_12))
struct timespec ts;
clock_gettime(CLOCK_THREAD_CPUTIME_ID, &ts);
return static_cast<uint64_t>(ts.tv_sec) * 1000000000 + ts.tv_nsec;
#endif
return 0;
}
void SleepForMicroseconds(int micros) override { usleep(micros); }
Status GetCurrentTime(int64_t* unix_time) override {
time_t ret = time(nullptr);
if (ret == (time_t)-1) {
return IOError("GetCurrentTime", "", errno);
}
*unix_time = (int64_t)ret;
return Status::OK();
}
std::string TimeToString(uint64_t secondsSince1970) override {
const time_t seconds = (time_t)secondsSince1970;
struct tm t;
int maxsize = 64;
std::string dummy;
dummy.reserve(maxsize);
dummy.resize(maxsize);
char* p = &dummy[0];
port::LocalTimeR(&seconds, &t);
snprintf(p, maxsize, "%04d/%02d/%02d-%02d:%02d:%02d ", t.tm_year + 1900,
t.tm_mon + 1, t.tm_mday, t.tm_hour, t.tm_min, t.tm_sec);
return dummy;
}
};
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
4 years ago
class PosixEnv : public CompositeEnv {
public:
static const char* kClassName() { return "PosixEnv"; }
const char* Name() const override { return kClassName(); }
const char* NickName() const override { return kDefaultName(); }
~PosixEnv() override {
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
4 years ago
if (this == Env::Default()) {
for (const auto tid : threads_to_join_) {
pthread_join(tid, nullptr);
}
for (int pool_id = 0; pool_id < Env::Priority::TOTAL; ++pool_id) {
thread_pools_[pool_id].JoinAllThreads();
}
// Do not delete the thread_status_updater_ in order to avoid the
// free after use when Env::Default() is destructed while some other
// child threads are still trying to update thread status. All
// PosixEnv instances use the same thread_status_updater_, so never
// explicitly delete it.
}
}
void SetFD_CLOEXEC(int fd, const EnvOptions* options) {
if ((options == nullptr || options->set_fd_cloexec) && fd > 0) {
fcntl(fd, F_SETFD, fcntl(fd, F_GETFD) | FD_CLOEXEC);
}
}
#ifndef ROCKSDB_NO_DYNAMIC_EXTENSION
Fix tsan error (#5414) Summary: Previous code has a warning when compile with tsan, leading to an error since we have -Werror. Compilation result ``` In file included from ./env/env_chroot.h:12, from env/env_test.cc:40: ./include/rocksdb/env.h: In instantiation of ‘rocksdb::Status rocksdb::DynamicLibrary::LoadFunction(const string&, std::function<T>*) [with T = void*(void*, const char*); std::__cxx11::string = std::__cxx11::basic_string<char>]’: env/env_test.cc:260:5: required from here ./include/rocksdb/env.h:1010:17: error: cast between incompatible function types from ‘rocksdb::DynamicLibrary::FunctionPtr’ {aka ‘void* (*)()’} to ‘void* (*)(void*, const char*)’ [-Werror=cast-function-type] *function = reinterpret_cast<T*>(ptr); ^~~~~~~~~~~~~~~~~~~~~~~~~ cc1plus: all warnings being treated as errors make: *** [env/env_test.o] Error 1 ``` It also has another error reported by clang ``` env/env_posix.cc:141:11: warning: Value stored to 'err' during its initialization is never read char* err = dlerror(); // Clear any old error ^~~ ~~~~~~~~~ 1 warning generated. ``` Test plan (on my devserver). ``` $make clean $OPT=-g ROCKSDB_FBCODE_BUILD_WITH_PLATFORM007=1 COMPILE_WITH_TSAN=1 make -j32 $ $make clean $USE_CLANG=1 TEST_TMPDIR=/dev/shm/rocksdb OPT=-g make -j1 analyze ``` Both should pass. Pull Request resolved: https://github.com/facebook/rocksdb/pull/5414 Differential Revision: D15637315 Pulled By: riversand963 fbshipit-source-id: 8e307483761019a4d5998cab92d49516d7edffbf
5 years ago
// Loads the named library into the result.
// If the input name is empty, the current executable is loaded
// On *nix systems, a "lib" prefix is added to the name if one is not supplied
// Comparably, the appropriate shared library extension is added to the name
// if not supplied. If search_path is not specified, the shared library will
// be loaded using the default path (LD_LIBRARY_PATH) If search_path is
// specified, the shared library will be searched for in the directories
// provided by the search path
Status LoadLibrary(const std::string& name, const std::string& path,
std::shared_ptr<DynamicLibrary>* result) override {
assert(result != nullptr);
if (name.empty()) {
void* hndl = dlopen(NULL, RTLD_NOW);
if (hndl != nullptr) {
result->reset(new PosixDynamicLibrary(name, hndl));
return Status::OK();
}
} else {
std::string library_name = name;
if (library_name.find(kSharedLibExt) == std::string::npos) {
library_name = library_name + kSharedLibExt;
}
#if !defined(OS_WIN)
if (library_name.find('/') == std::string::npos &&
library_name.compare(0, 3, "lib") != 0) {
library_name = "lib" + library_name;
}
#endif
if (path.empty()) {
void* hndl = dlopen(library_name.c_str(), RTLD_NOW);
if (hndl != nullptr) {
result->reset(new PosixDynamicLibrary(library_name, hndl));
return Status::OK();
}
} else {
std::string local_path;
std::stringstream ss(path);
while (getline(ss, local_path, kPathSeparator)) {
if (!path.empty()) {
std::string full_name = local_path + "/" + library_name;
void* hndl = dlopen(full_name.c_str(), RTLD_NOW);
if (hndl != nullptr) {
result->reset(new PosixDynamicLibrary(full_name, hndl));
return Status::OK();
}
}
}
}
}
return Status::IOError(
IOErrorMsg("Failed to open shared library: xs", name), dlerror());
}
#endif // !ROCKSDB_NO_DYNAMIC_EXTENSION
void Schedule(void (*function)(void* arg1), void* arg, Priority pri = LOW,
void* tag = nullptr,
void (*unschedFunction)(void* arg) = nullptr) override;
int UnSchedule(void* arg, Priority pri) override;
void StartThread(void (*function)(void* arg), void* arg) override;
void WaitForJoin() override;
unsigned int GetThreadPoolQueueLen(Priority pri = LOW) const override;
int ReserveThreads(int threads_to_be_reserved, Priority pri) override;
int ReleaseThreads(int threads_to_be_released, Priority pri) override;
Status GetThreadList(std::vector<ThreadStatus>* thread_list) override {
assert(thread_status_updater_);
return thread_status_updater_->GetThreadList(thread_list);
}
uint64_t GetThreadID() const override {
uint64_t thread_id = 0;
#if defined(_GNU_SOURCE) && defined(__GLIBC_PREREQ)
#if __GLIBC_PREREQ(2, 30)
thread_id = ::gettid();
#else // __GLIBC_PREREQ(2, 30)
pthread_t tid = pthread_self();
memcpy(&thread_id, &tid, std::min(sizeof(thread_id), sizeof(tid)));
#endif // __GLIBC_PREREQ(2, 30)
#else // defined(_GNU_SOURCE) && defined(__GLIBC_PREREQ)
pthread_t tid = pthread_self();
memcpy(&thread_id, &tid, std::min(sizeof(thread_id), sizeof(tid)));
#endif // defined(_GNU_SOURCE) && defined(__GLIBC_PREREQ)
return thread_id;
}
Status GetHostName(char* name, uint64_t len) override {
int ret = gethostname(name, static_cast<size_t>(len));
if (ret < 0) {
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
5 years ago
if (errno == EFAULT || errno == EINVAL) {
return Status::InvalidArgument(errnoStr(errno).c_str());
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
5 years ago
} else {
return IOError("GetHostName", name, errno);
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
5 years ago
}
}
return Status::OK();
}
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
5 years ago
ThreadStatusUpdater* GetThreadStatusUpdater() const override {
return Env::GetThreadStatusUpdater();
}
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
5 years ago
std::string GenerateUniqueId() override { return Env::GenerateUniqueId(); }
// Allow increasing the number of worker threads.
void SetBackgroundThreads(int num, Priority pri) override {
Introduce bottom-pri thread pool for large universal compactions Summary: When we had a single thread pool for compactions, a thread could be busy for a long time (minutes) executing a compaction involving the bottom level. In multi-instance setups, the entire thread pool could be consumed by such bottom-level compactions. Then, top-level compactions (e.g., a few L0 files) would be blocked for a long time ("head-of-line blocking"). Such top-level compactions are critical to prevent compaction stalls as they can quickly reduce number of L0 files / sorted runs. This diff introduces a bottom-priority queue for universal compactions including the bottom level. This alleviates the head-of-line blocking situation for fast, top-level compactions. - Added `Env::Priority::BOTTOM` thread pool. This feature is only enabled if user explicitly configures it to have a positive number of threads. - Changed `ThreadPoolImpl`'s default thread limit from one to zero. This change is invisible to users as we call `IncBackgroundThreadsIfNeeded` on the low-pri/high-pri pools during `DB::Open` with values of at least one. It is necessary, though, for bottom-pri to start with zero threads so the feature is disabled by default. - Separated `ManualCompaction` into two parts in `PrepickedCompaction`. `PrepickedCompaction` is used for any compaction that's picked outside of its execution thread, either manual or automatic. - Forward universal compactions involving last level to the bottom pool (worker thread's entry point is `BGWorkBottomCompaction`). - Track `bg_bottom_compaction_scheduled_` so we can wait for bottom-level compactions to finish. We don't count them against the background jobs limits. So users of this feature will get an extra compaction for free. Closes https://github.com/facebook/rocksdb/pull/2580 Differential Revision: D5422916 Pulled By: ajkr fbshipit-source-id: a74bd11f1ea4933df3739b16808bb21fcd512333
7 years ago
assert(pri >= Priority::BOTTOM && pri <= Priority::HIGH);
thread_pools_[pri].SetBackgroundThreads(num);
}
int GetBackgroundThreads(Priority pri) override {
Introduce bottom-pri thread pool for large universal compactions Summary: When we had a single thread pool for compactions, a thread could be busy for a long time (minutes) executing a compaction involving the bottom level. In multi-instance setups, the entire thread pool could be consumed by such bottom-level compactions. Then, top-level compactions (e.g., a few L0 files) would be blocked for a long time ("head-of-line blocking"). Such top-level compactions are critical to prevent compaction stalls as they can quickly reduce number of L0 files / sorted runs. This diff introduces a bottom-priority queue for universal compactions including the bottom level. This alleviates the head-of-line blocking situation for fast, top-level compactions. - Added `Env::Priority::BOTTOM` thread pool. This feature is only enabled if user explicitly configures it to have a positive number of threads. - Changed `ThreadPoolImpl`'s default thread limit from one to zero. This change is invisible to users as we call `IncBackgroundThreadsIfNeeded` on the low-pri/high-pri pools during `DB::Open` with values of at least one. It is necessary, though, for bottom-pri to start with zero threads so the feature is disabled by default. - Separated `ManualCompaction` into two parts in `PrepickedCompaction`. `PrepickedCompaction` is used for any compaction that's picked outside of its execution thread, either manual or automatic. - Forward universal compactions involving last level to the bottom pool (worker thread's entry point is `BGWorkBottomCompaction`). - Track `bg_bottom_compaction_scheduled_` so we can wait for bottom-level compactions to finish. We don't count them against the background jobs limits. So users of this feature will get an extra compaction for free. Closes https://github.com/facebook/rocksdb/pull/2580 Differential Revision: D5422916 Pulled By: ajkr fbshipit-source-id: a74bd11f1ea4933df3739b16808bb21fcd512333
7 years ago
assert(pri >= Priority::BOTTOM && pri <= Priority::HIGH);
return thread_pools_[pri].GetBackgroundThreads();
}
Status SetAllowNonOwnerAccess(bool allow_non_owner_access) override {
allow_non_owner_access_ = allow_non_owner_access;
return Status::OK();
}
// Allow increasing the number of worker threads.
void IncBackgroundThreadsIfNeeded(int num, Priority pri) override {
Introduce bottom-pri thread pool for large universal compactions Summary: When we had a single thread pool for compactions, a thread could be busy for a long time (minutes) executing a compaction involving the bottom level. In multi-instance setups, the entire thread pool could be consumed by such bottom-level compactions. Then, top-level compactions (e.g., a few L0 files) would be blocked for a long time ("head-of-line blocking"). Such top-level compactions are critical to prevent compaction stalls as they can quickly reduce number of L0 files / sorted runs. This diff introduces a bottom-priority queue for universal compactions including the bottom level. This alleviates the head-of-line blocking situation for fast, top-level compactions. - Added `Env::Priority::BOTTOM` thread pool. This feature is only enabled if user explicitly configures it to have a positive number of threads. - Changed `ThreadPoolImpl`'s default thread limit from one to zero. This change is invisible to users as we call `IncBackgroundThreadsIfNeeded` on the low-pri/high-pri pools during `DB::Open` with values of at least one. It is necessary, though, for bottom-pri to start with zero threads so the feature is disabled by default. - Separated `ManualCompaction` into two parts in `PrepickedCompaction`. `PrepickedCompaction` is used for any compaction that's picked outside of its execution thread, either manual or automatic. - Forward universal compactions involving last level to the bottom pool (worker thread's entry point is `BGWorkBottomCompaction`). - Track `bg_bottom_compaction_scheduled_` so we can wait for bottom-level compactions to finish. We don't count them against the background jobs limits. So users of this feature will get an extra compaction for free. Closes https://github.com/facebook/rocksdb/pull/2580 Differential Revision: D5422916 Pulled By: ajkr fbshipit-source-id: a74bd11f1ea4933df3739b16808bb21fcd512333
7 years ago
assert(pri >= Priority::BOTTOM && pri <= Priority::HIGH);
thread_pools_[pri].IncBackgroundThreadsIfNeeded(num);
}
void LowerThreadPoolIOPriority(Priority pool) override {
Introduce bottom-pri thread pool for large universal compactions Summary: When we had a single thread pool for compactions, a thread could be busy for a long time (minutes) executing a compaction involving the bottom level. In multi-instance setups, the entire thread pool could be consumed by such bottom-level compactions. Then, top-level compactions (e.g., a few L0 files) would be blocked for a long time ("head-of-line blocking"). Such top-level compactions are critical to prevent compaction stalls as they can quickly reduce number of L0 files / sorted runs. This diff introduces a bottom-priority queue for universal compactions including the bottom level. This alleviates the head-of-line blocking situation for fast, top-level compactions. - Added `Env::Priority::BOTTOM` thread pool. This feature is only enabled if user explicitly configures it to have a positive number of threads. - Changed `ThreadPoolImpl`'s default thread limit from one to zero. This change is invisible to users as we call `IncBackgroundThreadsIfNeeded` on the low-pri/high-pri pools during `DB::Open` with values of at least one. It is necessary, though, for bottom-pri to start with zero threads so the feature is disabled by default. - Separated `ManualCompaction` into two parts in `PrepickedCompaction`. `PrepickedCompaction` is used for any compaction that's picked outside of its execution thread, either manual or automatic. - Forward universal compactions involving last level to the bottom pool (worker thread's entry point is `BGWorkBottomCompaction`). - Track `bg_bottom_compaction_scheduled_` so we can wait for bottom-level compactions to finish. We don't count them against the background jobs limits. So users of this feature will get an extra compaction for free. Closes https://github.com/facebook/rocksdb/pull/2580 Differential Revision: D5422916 Pulled By: ajkr fbshipit-source-id: a74bd11f1ea4933df3739b16808bb21fcd512333
7 years ago
assert(pool >= Priority::BOTTOM && pool <= Priority::HIGH);
#ifdef OS_LINUX
thread_pools_[pool].LowerIOPriority();
#else
(void)pool;
#endif
}
void LowerThreadPoolCPUPriority(Priority pool) override {
assert(pool >= Priority::BOTTOM && pool <= Priority::HIGH);
thread_pools_[pool].LowerCPUPriority(CpuPriority::kLow);
}
Status LowerThreadPoolCPUPriority(Priority pool, CpuPriority pri) override {
assert(pool >= Priority::BOTTOM && pool <= Priority::HIGH);
thread_pools_[pool].LowerCPUPriority(pri);
return Status::OK();
}
private:
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
4 years ago
friend Env* Env::Default();
// Constructs the default Env, a singleton
PosixEnv();
// The below 4 members are only used by the default PosixEnv instance.
// Non-default instances simply maintain references to the backing
// members in te default instance
std::vector<ThreadPoolImpl> thread_pools_storage_;
pthread_mutex_t mu_storage_;
std::vector<pthread_t> threads_to_join_storage_;
bool allow_non_owner_access_storage_;
std::vector<ThreadPoolImpl>& thread_pools_;
pthread_mutex_t& mu_;
std::vector<pthread_t>& threads_to_join_;
// If true, allow non owner read access for db files. Otherwise, non-owner
// has no access to db files.
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
4 years ago
bool& allow_non_owner_access_;
};
PosixEnv::PosixEnv()
: CompositeEnv(FileSystem::Default(), SystemClock::Default()),
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
4 years ago
thread_pools_storage_(Priority::TOTAL),
allow_non_owner_access_storage_(true),
thread_pools_(thread_pools_storage_),
mu_(mu_storage_),
threads_to_join_(threads_to_join_storage_),
allow_non_owner_access_(allow_non_owner_access_storage_) {
ThreadPoolImpl::PthreadCall("mutex_init", pthread_mutex_init(&mu_, nullptr));
for (int pool_id = 0; pool_id < Env::Priority::TOTAL; ++pool_id) {
thread_pools_[pool_id].SetThreadPriority(
static_cast<Env::Priority>(pool_id));
// This allows later initializing the thread-local-env of each thread.
thread_pools_[pool_id].SetHostEnv(this);
}
thread_status_updater_ = CreateThreadStatusUpdater();
}
void PosixEnv::Schedule(void (*function)(void* arg1), void* arg, Priority pri,
void* tag, void (*unschedFunction)(void* arg)) {
Introduce bottom-pri thread pool for large universal compactions Summary: When we had a single thread pool for compactions, a thread could be busy for a long time (minutes) executing a compaction involving the bottom level. In multi-instance setups, the entire thread pool could be consumed by such bottom-level compactions. Then, top-level compactions (e.g., a few L0 files) would be blocked for a long time ("head-of-line blocking"). Such top-level compactions are critical to prevent compaction stalls as they can quickly reduce number of L0 files / sorted runs. This diff introduces a bottom-priority queue for universal compactions including the bottom level. This alleviates the head-of-line blocking situation for fast, top-level compactions. - Added `Env::Priority::BOTTOM` thread pool. This feature is only enabled if user explicitly configures it to have a positive number of threads. - Changed `ThreadPoolImpl`'s default thread limit from one to zero. This change is invisible to users as we call `IncBackgroundThreadsIfNeeded` on the low-pri/high-pri pools during `DB::Open` with values of at least one. It is necessary, though, for bottom-pri to start with zero threads so the feature is disabled by default. - Separated `ManualCompaction` into two parts in `PrepickedCompaction`. `PrepickedCompaction` is used for any compaction that's picked outside of its execution thread, either manual or automatic. - Forward universal compactions involving last level to the bottom pool (worker thread's entry point is `BGWorkBottomCompaction`). - Track `bg_bottom_compaction_scheduled_` so we can wait for bottom-level compactions to finish. We don't count them against the background jobs limits. So users of this feature will get an extra compaction for free. Closes https://github.com/facebook/rocksdb/pull/2580 Differential Revision: D5422916 Pulled By: ajkr fbshipit-source-id: a74bd11f1ea4933df3739b16808bb21fcd512333
7 years ago
assert(pri >= Priority::BOTTOM && pri <= Priority::HIGH);
thread_pools_[pri].Schedule(function, arg, tag, unschedFunction);
}
int PosixEnv::UnSchedule(void* arg, Priority pri) {
return thread_pools_[pri].UnSchedule(arg);
}
unsigned int PosixEnv::GetThreadPoolQueueLen(Priority pri) const {
Introduce bottom-pri thread pool for large universal compactions Summary: When we had a single thread pool for compactions, a thread could be busy for a long time (minutes) executing a compaction involving the bottom level. In multi-instance setups, the entire thread pool could be consumed by such bottom-level compactions. Then, top-level compactions (e.g., a few L0 files) would be blocked for a long time ("head-of-line blocking"). Such top-level compactions are critical to prevent compaction stalls as they can quickly reduce number of L0 files / sorted runs. This diff introduces a bottom-priority queue for universal compactions including the bottom level. This alleviates the head-of-line blocking situation for fast, top-level compactions. - Added `Env::Priority::BOTTOM` thread pool. This feature is only enabled if user explicitly configures it to have a positive number of threads. - Changed `ThreadPoolImpl`'s default thread limit from one to zero. This change is invisible to users as we call `IncBackgroundThreadsIfNeeded` on the low-pri/high-pri pools during `DB::Open` with values of at least one. It is necessary, though, for bottom-pri to start with zero threads so the feature is disabled by default. - Separated `ManualCompaction` into two parts in `PrepickedCompaction`. `PrepickedCompaction` is used for any compaction that's picked outside of its execution thread, either manual or automatic. - Forward universal compactions involving last level to the bottom pool (worker thread's entry point is `BGWorkBottomCompaction`). - Track `bg_bottom_compaction_scheduled_` so we can wait for bottom-level compactions to finish. We don't count them against the background jobs limits. So users of this feature will get an extra compaction for free. Closes https://github.com/facebook/rocksdb/pull/2580 Differential Revision: D5422916 Pulled By: ajkr fbshipit-source-id: a74bd11f1ea4933df3739b16808bb21fcd512333
7 years ago
assert(pri >= Priority::BOTTOM && pri <= Priority::HIGH);
return thread_pools_[pri].GetQueueLen();
}
int PosixEnv::ReserveThreads(int threads_to_reserved, Priority pri) {
assert(pri >= Priority::BOTTOM && pri <= Priority::HIGH);
return thread_pools_[pri].ReserveThreads(threads_to_reserved);
}
int PosixEnv::ReleaseThreads(int threads_to_released, Priority pri) {
assert(pri >= Priority::BOTTOM && pri <= Priority::HIGH);
return thread_pools_[pri].ReleaseThreads(threads_to_released);
}
struct StartThreadState {
void (*user_function)(void*);
void* arg;
};
static void* StartThreadWrapper(void* arg) {
StartThreadState* state = reinterpret_cast<StartThreadState*>(arg);
state->user_function(state->arg);
delete state;
return nullptr;
}
void PosixEnv::StartThread(void (*function)(void* arg), void* arg) {
pthread_t t;
StartThreadState* state = new StartThreadState;
state->user_function = function;
state->arg = arg;
ThreadPoolImpl::PthreadCall(
"start thread", pthread_create(&t, nullptr, &StartThreadWrapper, state));
ThreadPoolImpl::PthreadCall("lock", pthread_mutex_lock(&mu_));
threads_to_join_.push_back(t);
ThreadPoolImpl::PthreadCall("unlock", pthread_mutex_unlock(&mu_));
}
void PosixEnv::WaitForJoin() {
for (const auto tid : threads_to_join_) {
pthread_join(tid, nullptr);
}
threads_to_join_.clear();
}
} // namespace
//
// Default Posix Env
//
Env* Env::Default() {
// The following function call initializes the singletons of ThreadLocalPtr
// right before the static default_env. This guarantees default_env will
// always being destructed before the ThreadLocalPtr singletons get
// destructed as C++ guarantees that the destructions of static variables
// is in the reverse order of their constructions.
//
// Since static members are destructed in the reverse order
// of their construction, having this call here guarantees that
// the destructor of static PosixEnv will go first, then the
// the singletons of ThreadLocalPtr.
ThreadLocalPtr::InitSingletons();
CompressionContextCache::InitSingleton();
INIT_SYNC_POINT_SINGLETONS();
// ~PosixEnv must be called on exit
//**TODO: Can we make this a STATIC_AVOID_DESTRUCTION?
static PosixEnv default_env;
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
4 years ago
return &default_env;
}
//
// Default Posix SystemClock
//
const std::shared_ptr<SystemClock>& SystemClock::Default() {
STATIC_AVOID_DESTRUCTION(std::shared_ptr<SystemClock>, instance)
(std::make_shared<PosixClock>());
return instance;
}
} // namespace ROCKSDB_NAMESPACE
#endif