mirror of
https://github.com/facebook/rocksdb.git
synced 2024-11-28 05:43:50 +00:00
151242ce46
Summary: **Context:** The existing stat rocksdb.sst.read.micros does not reflect each of compaction and flush cases but aggregate them, which is not so helpful for us to understand IO read behavior of each of them. **Summary** - Update `StopWatch` and `RandomAccessFileReader` to record `rocksdb.sst.read.micros` and `rocksdb.file.{flush/compaction}.read.micros` - Fixed the default histogram in `RandomAccessFileReader` - New field `ReadOptions/IOOptions::io_activity`; Pass `ReadOptions` through paths under db open, flush and compaction to where we can prepare `IOOptions` and pass it to `RandomAccessFileReader` - Use `thread_status_util` for assertion in `DbStressFSWrapper` for continuous testing on we are passing correct `io_activity` under db open, flush and compaction Pull Request resolved: https://github.com/facebook/rocksdb/pull/11288 Test Plan: - **Stress test** - **Db bench 1: rocksdb.sst.read.micros COUNT ≈ sum of rocksdb.file.read.flush.micros's and rocksdb.file.read.compaction.micros's.** (without blob) - May not be exactly the same due to `HistogramStat::Add` only guarantees atomic not accuracy across threads. ``` ./db_bench -db=/dev/shm/testdb/ -statistics=true -benchmarks="fillseq" -key_size=32 -value_size=512 -num=50000 -write_buffer_size=655 -target_file_size_base=655 -disable_auto_compactions=false -compression_type=none -bloom_bits=3 (-use_plain_table=1 -prefix_size=10) ``` ``` // BlockBasedTable rocksdb.sst.read.micros P50 : 2.009374 P95 : 4.968548 P99 : 8.110362 P100 : 43.000000 COUNT : 40456 SUM : 114805 rocksdb.file.read.flush.micros P50 : 1.871841 P95 : 3.872407 P99 : 5.540541 P100 : 43.000000 COUNT : 2250 SUM : 6116 rocksdb.file.read.compaction.micros P50 : 2.023109 P95 : 5.029149 P99 : 8.196910 P100 : 26.000000 COUNT : 38206 SUM : 108689 // PlainTable Does not apply ``` - **Db bench 2: performance** **Read** SETUP: db with 900 files ``` ./db_bench -db=/dev/shm/testdb/ -benchmarks="fillseq" -key_size=32 -value_size=512 -num=50000 -write_buffer_size=655 -disable_auto_compactions=true -target_file_size_base=655 -compression_type=none ```run till convergence ``` ./db_bench -seed=1678564177044286 -use_existing_db=true -db=/dev/shm/testdb -benchmarks=readrandom[-X60] -statistics=true -num=1000000 -disable_auto_compactions=true -compression_type=none -bloom_bits=3 ``` Pre-change `readrandom [AVG 60 runs] : 21568 (± 248) ops/sec` Post-change (no regression, -0.3%) `readrandom [AVG 60 runs] : 21486 (± 236) ops/sec` **Compaction/Flush**run till convergence ``` ./db_bench -db=/dev/shm/testdb2/ -seed=1678564177044286 -benchmarks="fillseq[-X60]" -key_size=32 -value_size=512 -num=50000 -write_buffer_size=655 -disable_auto_compactions=false -target_file_size_base=655 -compression_type=none rocksdb.sst.read.micros COUNT : 33820 rocksdb.sst.read.flush.micros COUNT : 1800 rocksdb.sst.read.compaction.micros COUNT : 32020 ``` Pre-change `fillseq [AVG 46 runs] : 1391 (± 214) ops/sec; 0.7 (± 0.1) MB/sec` Post-change (no regression, ~-0.4%) `fillseq [AVG 46 runs] : 1385 (± 216) ops/sec; 0.7 (± 0.1) MB/sec` Reviewed By: ajkr Differential Revision: D44007011 Pulled By: hx235 fbshipit-source-id: a54c89e4846dfc9a135389edf3f3eedfea257132
363 lines
13 KiB
C++
363 lines
13 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 <condition_variable>
|
|
#include <mutex>
|
|
|
|
#include "monitoring/thread_status_updater.h"
|
|
#include "rocksdb/db.h"
|
|
#include "test_util/testharness.h"
|
|
|
|
#ifdef ROCKSDB_USING_THREAD_STATUS
|
|
|
|
namespace ROCKSDB_NAMESPACE {
|
|
|
|
class SimulatedBackgroundTask {
|
|
public:
|
|
SimulatedBackgroundTask(
|
|
const void* db_key, const std::string& db_name, const void* cf_key,
|
|
const std::string& cf_name,
|
|
const ThreadStatus::OperationType operation_type =
|
|
ThreadStatus::OP_UNKNOWN,
|
|
const ThreadStatus::StateType state_type = ThreadStatus::STATE_UNKNOWN)
|
|
: db_key_(db_key),
|
|
db_name_(db_name),
|
|
cf_key_(cf_key),
|
|
cf_name_(cf_name),
|
|
operation_type_(operation_type),
|
|
state_type_(state_type),
|
|
should_run_(true),
|
|
running_count_(0) {
|
|
Env::Default()->GetThreadStatusUpdater()->NewColumnFamilyInfo(
|
|
db_key_, db_name_, cf_key_, cf_name_);
|
|
}
|
|
|
|
~SimulatedBackgroundTask() {
|
|
Env::Default()->GetThreadStatusUpdater()->EraseDatabaseInfo(db_key_);
|
|
}
|
|
|
|
void Run() {
|
|
std::unique_lock<std::mutex> l(mutex_);
|
|
running_count_++;
|
|
bg_cv_.notify_all();
|
|
assert(cf_key_);
|
|
Env::Default()->GetThreadStatusUpdater()->SetEnableTracking(true);
|
|
Env::Default()->GetThreadStatusUpdater()->SetColumnFamilyInfoKey(cf_key_);
|
|
Env::Default()->GetThreadStatusUpdater()->SetThreadOperation(
|
|
operation_type_);
|
|
Env::Default()->GetThreadStatusUpdater()->SetThreadState(state_type_);
|
|
while (should_run_) {
|
|
bg_cv_.wait(l);
|
|
}
|
|
Env::Default()->GetThreadStatusUpdater()->ClearThreadState();
|
|
Env::Default()->GetThreadStatusUpdater()->ClearThreadOperation();
|
|
Env::Default()->GetThreadStatusUpdater()->SetColumnFamilyInfoKey(nullptr);
|
|
running_count_--;
|
|
bg_cv_.notify_all();
|
|
}
|
|
|
|
void FinishAllTasks() {
|
|
std::unique_lock<std::mutex> l(mutex_);
|
|
should_run_ = false;
|
|
bg_cv_.notify_all();
|
|
}
|
|
|
|
void WaitUntilScheduled(int job_count) {
|
|
std::unique_lock<std::mutex> l(mutex_);
|
|
while (running_count_ < job_count) {
|
|
bg_cv_.wait(l);
|
|
}
|
|
}
|
|
|
|
void WaitUntilDone() {
|
|
std::unique_lock<std::mutex> l(mutex_);
|
|
while (running_count_ > 0) {
|
|
bg_cv_.wait(l);
|
|
}
|
|
}
|
|
|
|
static void DoSimulatedTask(void* arg) {
|
|
reinterpret_cast<SimulatedBackgroundTask*>(arg)->Run();
|
|
}
|
|
|
|
private:
|
|
const void* db_key_;
|
|
const std::string db_name_;
|
|
const void* cf_key_;
|
|
const std::string cf_name_;
|
|
const ThreadStatus::OperationType operation_type_;
|
|
const ThreadStatus::StateType state_type_;
|
|
std::mutex mutex_;
|
|
std::condition_variable bg_cv_;
|
|
bool should_run_;
|
|
std::atomic<int> running_count_;
|
|
};
|
|
|
|
class ThreadListTest : public testing::Test {
|
|
public:
|
|
ThreadListTest() {}
|
|
};
|
|
|
|
TEST_F(ThreadListTest, GlobalTables) {
|
|
// verify the global tables for operations and states are properly indexed.
|
|
for (int type = 0; type != ThreadStatus::NUM_OP_TYPES; ++type) {
|
|
ASSERT_EQ(global_operation_table[type].type, type);
|
|
ASSERT_EQ(
|
|
global_operation_table[type].name,
|
|
ThreadStatus::GetOperationName(ThreadStatus::OperationType(type)));
|
|
}
|
|
|
|
for (int type = 0; type != ThreadStatus::NUM_STATE_TYPES; ++type) {
|
|
ASSERT_EQ(global_state_table[type].type, type);
|
|
ASSERT_EQ(global_state_table[type].name,
|
|
ThreadStatus::GetStateName(ThreadStatus::StateType(type)));
|
|
}
|
|
|
|
for (int stage = 0; stage != ThreadStatus::NUM_OP_STAGES; ++stage) {
|
|
ASSERT_EQ(global_op_stage_table[stage].stage, stage);
|
|
ASSERT_EQ(global_op_stage_table[stage].name,
|
|
ThreadStatus::GetOperationStageName(
|
|
ThreadStatus::OperationStage(stage)));
|
|
}
|
|
}
|
|
|
|
TEST_F(ThreadListTest, SimpleColumnFamilyInfoTest) {
|
|
Env* env = Env::Default();
|
|
const int kHighPriorityThreads = 3;
|
|
const int kLowPriorityThreads = 5;
|
|
const int kSimulatedHighPriThreads = kHighPriorityThreads - 1;
|
|
const int kSimulatedLowPriThreads = kLowPriorityThreads / 3;
|
|
const int kDelayMicros = 1000000;
|
|
env->SetBackgroundThreads(kHighPriorityThreads, Env::HIGH);
|
|
env->SetBackgroundThreads(kLowPriorityThreads, Env::LOW);
|
|
// Wait 1 second so that threads start
|
|
Env::Default()->SleepForMicroseconds(kDelayMicros);
|
|
SimulatedBackgroundTask running_task(reinterpret_cast<void*>(1234), "running",
|
|
reinterpret_cast<void*>(5678),
|
|
"pikachu");
|
|
|
|
for (int test = 0; test < kSimulatedHighPriThreads; ++test) {
|
|
env->Schedule(&SimulatedBackgroundTask::DoSimulatedTask, &running_task,
|
|
Env::Priority::HIGH);
|
|
}
|
|
|
|
for (int test = 0; test < kSimulatedLowPriThreads; ++test) {
|
|
env->Schedule(&SimulatedBackgroundTask::DoSimulatedTask, &running_task,
|
|
Env::Priority::LOW);
|
|
}
|
|
running_task.WaitUntilScheduled(kSimulatedHighPriThreads +
|
|
kSimulatedLowPriThreads);
|
|
// We can only reserve limited number of waiting threads
|
|
ASSERT_EQ(kHighPriorityThreads - kSimulatedHighPriThreads,
|
|
env->ReserveThreads(kHighPriorityThreads, Env::Priority::HIGH));
|
|
ASSERT_EQ(kLowPriorityThreads - kSimulatedLowPriThreads,
|
|
env->ReserveThreads(kLowPriorityThreads, Env::Priority::LOW));
|
|
|
|
// Reservation shall not affect the existing thread list
|
|
std::vector<ThreadStatus> thread_list;
|
|
|
|
// Verify the number of running threads in each pool.
|
|
ASSERT_OK(env->GetThreadList(&thread_list));
|
|
int running_count[ThreadStatus::NUM_THREAD_TYPES] = {0};
|
|
for (auto thread_status : thread_list) {
|
|
if (thread_status.cf_name == "pikachu" &&
|
|
thread_status.db_name == "running") {
|
|
running_count[thread_status.thread_type]++;
|
|
}
|
|
}
|
|
// Cannot reserve more threads
|
|
ASSERT_EQ(0, env->ReserveThreads(kHighPriorityThreads, Env::Priority::HIGH));
|
|
ASSERT_EQ(0, env->ReserveThreads(kLowPriorityThreads, Env::Priority::LOW));
|
|
|
|
ASSERT_EQ(running_count[ThreadStatus::HIGH_PRIORITY],
|
|
kSimulatedHighPriThreads);
|
|
ASSERT_EQ(running_count[ThreadStatus::LOW_PRIORITY], kSimulatedLowPriThreads);
|
|
ASSERT_EQ(running_count[ThreadStatus::USER], 0);
|
|
|
|
running_task.FinishAllTasks();
|
|
running_task.WaitUntilDone();
|
|
|
|
ASSERT_EQ(kHighPriorityThreads - kSimulatedHighPriThreads,
|
|
env->ReleaseThreads(kHighPriorityThreads, Env::Priority::HIGH));
|
|
ASSERT_EQ(kLowPriorityThreads - kSimulatedLowPriThreads,
|
|
env->ReleaseThreads(kLowPriorityThreads, Env::Priority::LOW));
|
|
// Verify none of the threads are running
|
|
ASSERT_OK(env->GetThreadList(&thread_list));
|
|
|
|
for (int i = 0; i < ThreadStatus::NUM_THREAD_TYPES; ++i) {
|
|
running_count[i] = 0;
|
|
}
|
|
for (auto thread_status : thread_list) {
|
|
if (thread_status.cf_name == "pikachu" &&
|
|
thread_status.db_name == "running") {
|
|
running_count[thread_status.thread_type]++;
|
|
}
|
|
}
|
|
|
|
ASSERT_EQ(running_count[ThreadStatus::HIGH_PRIORITY], 0);
|
|
ASSERT_EQ(running_count[ThreadStatus::LOW_PRIORITY], 0);
|
|
ASSERT_EQ(running_count[ThreadStatus::USER], 0);
|
|
}
|
|
|
|
namespace {
|
|
void UpdateStatusCounts(const std::vector<ThreadStatus>& thread_list,
|
|
int operation_counts[], int state_counts[]) {
|
|
for (auto thread_status : thread_list) {
|
|
operation_counts[thread_status.operation_type]++;
|
|
state_counts[thread_status.state_type]++;
|
|
}
|
|
}
|
|
|
|
void VerifyAndResetCounts(const int correct_counts[], int collected_counts[],
|
|
int size) {
|
|
for (int i = 0; i < size; ++i) {
|
|
ASSERT_EQ(collected_counts[i], correct_counts[i]);
|
|
collected_counts[i] = 0;
|
|
}
|
|
}
|
|
|
|
void UpdateCount(int operation_counts[], int from_event, int to_event,
|
|
int amount) {
|
|
operation_counts[from_event] -= amount;
|
|
operation_counts[to_event] += amount;
|
|
}
|
|
} // namespace
|
|
|
|
TEST_F(ThreadListTest, SimpleEventTest) {
|
|
Env* env = Env::Default();
|
|
|
|
// simulated tasks
|
|
const int kFlushWriteTasks = 3;
|
|
SimulatedBackgroundTask flush_write_task(
|
|
reinterpret_cast<void*>(1234), "running", reinterpret_cast<void*>(5678),
|
|
"pikachu", ThreadStatus::OP_FLUSH);
|
|
|
|
const int kCompactionWriteTasks = 4;
|
|
SimulatedBackgroundTask compaction_write_task(
|
|
reinterpret_cast<void*>(1234), "running", reinterpret_cast<void*>(5678),
|
|
"pikachu", ThreadStatus::OP_COMPACTION);
|
|
|
|
const int kCompactionReadTasks = 5;
|
|
SimulatedBackgroundTask compaction_read_task(
|
|
reinterpret_cast<void*>(1234), "running", reinterpret_cast<void*>(5678),
|
|
"pikachu", ThreadStatus::OP_COMPACTION);
|
|
|
|
const int kCompactionWaitTasks = 6;
|
|
SimulatedBackgroundTask compaction_wait_task(
|
|
reinterpret_cast<void*>(1234), "running", reinterpret_cast<void*>(5678),
|
|
"pikachu", ThreadStatus::OP_COMPACTION);
|
|
|
|
// setup right answers
|
|
int correct_operation_counts[ThreadStatus::NUM_OP_TYPES] = {0};
|
|
correct_operation_counts[ThreadStatus::OP_FLUSH] = kFlushWriteTasks;
|
|
correct_operation_counts[ThreadStatus::OP_COMPACTION] =
|
|
kCompactionWriteTasks + kCompactionReadTasks + kCompactionWaitTasks;
|
|
|
|
env->SetBackgroundThreads(correct_operation_counts[ThreadStatus::OP_FLUSH],
|
|
Env::HIGH);
|
|
env->SetBackgroundThreads(
|
|
correct_operation_counts[ThreadStatus::OP_COMPACTION], Env::LOW);
|
|
|
|
// schedule the simulated tasks
|
|
for (int t = 0; t < kFlushWriteTasks; ++t) {
|
|
env->Schedule(&SimulatedBackgroundTask::DoSimulatedTask, &flush_write_task,
|
|
Env::Priority::HIGH);
|
|
}
|
|
flush_write_task.WaitUntilScheduled(kFlushWriteTasks);
|
|
|
|
for (int t = 0; t < kCompactionWriteTasks; ++t) {
|
|
env->Schedule(&SimulatedBackgroundTask::DoSimulatedTask,
|
|
&compaction_write_task, Env::Priority::LOW);
|
|
}
|
|
compaction_write_task.WaitUntilScheduled(kCompactionWriteTasks);
|
|
|
|
for (int t = 0; t < kCompactionReadTasks; ++t) {
|
|
env->Schedule(&SimulatedBackgroundTask::DoSimulatedTask,
|
|
&compaction_read_task, Env::Priority::LOW);
|
|
}
|
|
compaction_read_task.WaitUntilScheduled(kCompactionReadTasks);
|
|
|
|
for (int t = 0; t < kCompactionWaitTasks; ++t) {
|
|
env->Schedule(&SimulatedBackgroundTask::DoSimulatedTask,
|
|
&compaction_wait_task, Env::Priority::LOW);
|
|
}
|
|
compaction_wait_task.WaitUntilScheduled(kCompactionWaitTasks);
|
|
|
|
// verify the thread-status
|
|
int operation_counts[ThreadStatus::NUM_OP_TYPES] = {0};
|
|
int state_counts[ThreadStatus::NUM_STATE_TYPES] = {0};
|
|
|
|
std::vector<ThreadStatus> thread_list;
|
|
ASSERT_OK(env->GetThreadList(&thread_list));
|
|
UpdateStatusCounts(thread_list, operation_counts, state_counts);
|
|
VerifyAndResetCounts(correct_operation_counts, operation_counts,
|
|
ThreadStatus::NUM_OP_TYPES);
|
|
|
|
// terminate compaction-wait tasks and see if the thread-status
|
|
// reflects this update
|
|
compaction_wait_task.FinishAllTasks();
|
|
compaction_wait_task.WaitUntilDone();
|
|
UpdateCount(correct_operation_counts, ThreadStatus::OP_COMPACTION,
|
|
ThreadStatus::OP_UNKNOWN, kCompactionWaitTasks);
|
|
|
|
ASSERT_OK(env->GetThreadList(&thread_list));
|
|
UpdateStatusCounts(thread_list, operation_counts, state_counts);
|
|
VerifyAndResetCounts(correct_operation_counts, operation_counts,
|
|
ThreadStatus::NUM_OP_TYPES);
|
|
|
|
// terminate flush-write tasks and see if the thread-status
|
|
// reflects this update
|
|
flush_write_task.FinishAllTasks();
|
|
flush_write_task.WaitUntilDone();
|
|
UpdateCount(correct_operation_counts, ThreadStatus::OP_FLUSH,
|
|
ThreadStatus::OP_UNKNOWN, kFlushWriteTasks);
|
|
|
|
ASSERT_OK(env->GetThreadList(&thread_list));
|
|
UpdateStatusCounts(thread_list, operation_counts, state_counts);
|
|
VerifyAndResetCounts(correct_operation_counts, operation_counts,
|
|
ThreadStatus::NUM_OP_TYPES);
|
|
|
|
// terminate compaction-write tasks and see if the thread-status
|
|
// reflects this update
|
|
compaction_write_task.FinishAllTasks();
|
|
compaction_write_task.WaitUntilDone();
|
|
UpdateCount(correct_operation_counts, ThreadStatus::OP_COMPACTION,
|
|
ThreadStatus::OP_UNKNOWN, kCompactionWriteTasks);
|
|
|
|
ASSERT_OK(env->GetThreadList(&thread_list));
|
|
UpdateStatusCounts(thread_list, operation_counts, state_counts);
|
|
VerifyAndResetCounts(correct_operation_counts, operation_counts,
|
|
ThreadStatus::NUM_OP_TYPES);
|
|
|
|
// terminate compaction-write tasks and see if the thread-status
|
|
// reflects this update
|
|
compaction_read_task.FinishAllTasks();
|
|
compaction_read_task.WaitUntilDone();
|
|
UpdateCount(correct_operation_counts, ThreadStatus::OP_COMPACTION,
|
|
ThreadStatus::OP_UNKNOWN, kCompactionReadTasks);
|
|
|
|
ASSERT_OK(env->GetThreadList(&thread_list));
|
|
UpdateStatusCounts(thread_list, operation_counts, state_counts);
|
|
VerifyAndResetCounts(correct_operation_counts, operation_counts,
|
|
ThreadStatus::NUM_OP_TYPES);
|
|
}
|
|
|
|
} // namespace ROCKSDB_NAMESPACE
|
|
|
|
int main(int argc, char** argv) {
|
|
ROCKSDB_NAMESPACE::port::InstallStackTraceHandler();
|
|
::testing::InitGoogleTest(&argc, argv);
|
|
return RUN_ALL_TESTS();
|
|
}
|
|
|
|
#else
|
|
|
|
int main(int argc, char** argv) {
|
|
::testing::InitGoogleTest(&argc, argv);
|
|
return 0;
|
|
}
|
|
|
|
#endif // ROCKSDB_USING_THREAD_STATUS
|