mem tracker can be logically divided into 4 layers: 1)process 2)type 3)query/load/compation task etc. 4)exec node etc.
type includes
enum Type {
GLOBAL = 0, // Life cycle is the same as the process, e.g. Cache and default Orphan
QUERY = 1, // Count the memory consumption of all Query tasks.
LOAD = 2, // Count the memory consumption of all Load tasks.
COMPACTION = 3, // Count the memory consumption of all Base and Cumulative tasks.
SCHEMA_CHANGE = 4, // Count the memory consumption of all SchemaChange tasks.
CLONE = 5, // Count the memory consumption of all EngineCloneTask. Note: Memory that does not contain make/release snapshots.
BATCHLOAD = 6, // Count the memory consumption of all EngineBatchLoadTask.
CONSISTENCY = 7 // Count the memory consumption of all EngineChecksumTask.
}
Object pointers are no longer saved between each layer, and the values of process and each type are periodically aggregated.
other fix:
In [fix](memtracker) Fix transmit_tracker null pointer because phamp is not thread safe #13528, I tried to separate the memory that was manually abandoned in the query from the orphan mem tracker. But in the actual test, the accuracy of this part of the memory cannot be guaranteed, so put it back to the orphan mem tracker again.
414 lines
15 KiB
C++
414 lines
15 KiB
C++
// Licensed to the Apache Software Foundation (ASF) under one
|
|
// or more contributor license agreements. See the NOTICE file
|
|
// distributed with this work for additional information
|
|
// regarding copyright ownership. The ASF licenses this file
|
|
// to you under the Apache License, Version 2.0 (the
|
|
// "License"); you may not use this file except in compliance
|
|
// with the License. You may obtain a copy of the License at
|
|
//
|
|
// http://www.apache.org/licenses/LICENSE-2.0
|
|
//
|
|
// Unless required by applicable law or agreed to in writing,
|
|
// software distributed under the License is distributed on an
|
|
// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
|
|
// KIND, either express or implied. See the License for the
|
|
// specific language governing permissions and limitations
|
|
// under the License.
|
|
// This file is copied from
|
|
// https://github.com/apache/impala/blob/branch-2.9.0/be/src/runtime/runtime-state.cpp
|
|
// and modified by Doris
|
|
|
|
#include "runtime/runtime_state.h"
|
|
|
|
#include <fmt/format.h>
|
|
|
|
#include <boost/algorithm/string/join.hpp>
|
|
#include <sstream>
|
|
#include <string>
|
|
|
|
#include "common/logging.h"
|
|
#include "common/object_pool.h"
|
|
#include "common/status.h"
|
|
#include "exec/exec_node.h"
|
|
#include "runtime/buffered_block_mgr2.h"
|
|
#include "runtime/exec_env.h"
|
|
#include "runtime/load_path_mgr.h"
|
|
#include "runtime/memory/mem_tracker.h"
|
|
#include "runtime/runtime_filter_mgr.h"
|
|
#include "util/file_utils.h"
|
|
#include "util/load_error_hub.h"
|
|
#include "util/pretty_printer.h"
|
|
#include "util/timezone_utils.h"
|
|
#include "util/uid_util.h"
|
|
|
|
namespace doris {
|
|
|
|
// for ut only
|
|
RuntimeState::RuntimeState(const TUniqueId& fragment_instance_id,
|
|
const TQueryOptions& query_options, const TQueryGlobals& query_globals,
|
|
ExecEnv* exec_env)
|
|
: _profile("Fragment " + print_id(fragment_instance_id)),
|
|
_obj_pool(new ObjectPool()),
|
|
_runtime_filter_mgr(new RuntimeFilterMgr(TUniqueId(), this)),
|
|
_data_stream_recvrs_pool(new ObjectPool()),
|
|
_unreported_error_idx(0),
|
|
_is_cancelled(false),
|
|
_per_fragment_instance_idx(0),
|
|
_root_node_id(-1),
|
|
_num_rows_load_total(0),
|
|
_num_rows_load_filtered(0),
|
|
_num_rows_load_unselected(0),
|
|
_num_print_error_rows(0),
|
|
_num_bytes_load_total(0),
|
|
_load_job_id(-1),
|
|
_normal_row_number(0),
|
|
_error_row_number(0),
|
|
_error_log_file_path(""),
|
|
_error_log_file(nullptr) {
|
|
Status status = init(fragment_instance_id, query_options, query_globals, exec_env);
|
|
DCHECK(status.ok());
|
|
}
|
|
|
|
RuntimeState::RuntimeState(const TPlanFragmentExecParams& fragment_exec_params,
|
|
const TQueryOptions& query_options, const TQueryGlobals& query_globals,
|
|
ExecEnv* exec_env)
|
|
: _profile("Fragment " + print_id(fragment_exec_params.fragment_instance_id)),
|
|
_obj_pool(new ObjectPool()),
|
|
_runtime_filter_mgr(new RuntimeFilterMgr(fragment_exec_params.query_id, this)),
|
|
_data_stream_recvrs_pool(new ObjectPool()),
|
|
_unreported_error_idx(0),
|
|
_query_id(fragment_exec_params.query_id),
|
|
_is_cancelled(false),
|
|
_per_fragment_instance_idx(0),
|
|
_root_node_id(-1),
|
|
_num_rows_load_total(0),
|
|
_num_rows_load_filtered(0),
|
|
_num_rows_load_unselected(0),
|
|
_num_print_error_rows(0),
|
|
_num_bytes_load_total(0),
|
|
_normal_row_number(0),
|
|
_error_row_number(0),
|
|
_error_log_file_path(""),
|
|
_error_log_file(nullptr) {
|
|
if (fragment_exec_params.__isset.runtime_filter_params) {
|
|
_runtime_filter_mgr->set_runtime_filter_params(fragment_exec_params.runtime_filter_params);
|
|
}
|
|
Status status =
|
|
init(fragment_exec_params.fragment_instance_id, query_options, query_globals, exec_env);
|
|
DCHECK(status.ok());
|
|
}
|
|
|
|
RuntimeState::RuntimeState(const TQueryGlobals& query_globals)
|
|
: _profile("<unnamed>"),
|
|
_obj_pool(new ObjectPool()),
|
|
_data_stream_recvrs_pool(new ObjectPool()),
|
|
_unreported_error_idx(0),
|
|
_is_cancelled(false),
|
|
_per_fragment_instance_idx(0) {
|
|
_query_options.batch_size = DEFAULT_BATCH_SIZE;
|
|
if (query_globals.__isset.time_zone && query_globals.__isset.nano_seconds) {
|
|
_timezone = query_globals.time_zone;
|
|
_timestamp_ms = query_globals.timestamp_ms;
|
|
_nano_seconds = query_globals.nano_seconds;
|
|
} else if (query_globals.__isset.time_zone) {
|
|
_timezone = query_globals.time_zone;
|
|
_timestamp_ms = query_globals.timestamp_ms;
|
|
_nano_seconds = 0;
|
|
} else if (!query_globals.now_string.empty()) {
|
|
_timezone = TimezoneUtils::default_time_zone;
|
|
DateTimeValue dt;
|
|
dt.from_date_str(query_globals.now_string.c_str(), query_globals.now_string.size());
|
|
int64_t timestamp;
|
|
dt.unix_timestamp(×tamp, _timezone);
|
|
_timestamp_ms = timestamp * 1000;
|
|
_nano_seconds = 0;
|
|
} else {
|
|
//Unit test may set into here
|
|
_timezone = TimezoneUtils::default_time_zone;
|
|
_timestamp_ms = 0;
|
|
_nano_seconds = 0;
|
|
}
|
|
TimezoneUtils::find_cctz_time_zone(_timezone, _timezone_obj);
|
|
}
|
|
|
|
RuntimeState::RuntimeState()
|
|
: _profile("<unnamed>"),
|
|
_obj_pool(new ObjectPool()),
|
|
_data_stream_recvrs_pool(new ObjectPool()),
|
|
_unreported_error_idx(0),
|
|
_is_cancelled(false),
|
|
_per_fragment_instance_idx(0) {
|
|
_query_options.batch_size = DEFAULT_BATCH_SIZE;
|
|
_timezone = TimezoneUtils::default_time_zone;
|
|
_timestamp_ms = 0;
|
|
_nano_seconds = 0;
|
|
TimezoneUtils::find_cctz_time_zone(_timezone, _timezone_obj);
|
|
_exec_env = ExecEnv::GetInstance();
|
|
}
|
|
|
|
RuntimeState::~RuntimeState() {
|
|
_block_mgr2.reset();
|
|
// close error log file
|
|
if (_error_log_file != nullptr && _error_log_file->is_open()) {
|
|
_error_log_file->close();
|
|
delete _error_log_file;
|
|
_error_log_file = nullptr;
|
|
}
|
|
|
|
if (_error_hub != nullptr) {
|
|
_error_hub->close();
|
|
}
|
|
|
|
_obj_pool->clear();
|
|
_runtime_filter_mgr.reset();
|
|
}
|
|
|
|
Status RuntimeState::init(const TUniqueId& fragment_instance_id, const TQueryOptions& query_options,
|
|
const TQueryGlobals& query_globals, ExecEnv* exec_env) {
|
|
_fragment_instance_id = fragment_instance_id;
|
|
_query_options = query_options;
|
|
if (query_globals.__isset.time_zone && query_globals.__isset.nano_seconds) {
|
|
_timezone = query_globals.time_zone;
|
|
_timestamp_ms = query_globals.timestamp_ms;
|
|
_nano_seconds = query_globals.nano_seconds;
|
|
} else if (query_globals.__isset.time_zone) {
|
|
_timezone = query_globals.time_zone;
|
|
_timestamp_ms = query_globals.timestamp_ms;
|
|
_nano_seconds = 0;
|
|
} else if (!query_globals.now_string.empty()) {
|
|
_timezone = TimezoneUtils::default_time_zone;
|
|
DateTimeValue dt;
|
|
dt.from_date_str(query_globals.now_string.c_str(), query_globals.now_string.size());
|
|
int64_t timestamp;
|
|
dt.unix_timestamp(×tamp, _timezone);
|
|
_timestamp_ms = timestamp * 1000;
|
|
_nano_seconds = 0;
|
|
} else {
|
|
//Unit test may set into here
|
|
_timezone = TimezoneUtils::default_time_zone;
|
|
_timestamp_ms = 0;
|
|
_nano_seconds = 0;
|
|
}
|
|
TimezoneUtils::find_cctz_time_zone(_timezone, _timezone_obj);
|
|
|
|
if (query_globals.__isset.load_zero_tolerance) {
|
|
_load_zero_tolerance = query_globals.load_zero_tolerance;
|
|
}
|
|
|
|
_exec_env = exec_env;
|
|
|
|
if (_query_options.max_errors <= 0) {
|
|
// TODO: fix linker error and uncomment this
|
|
//_query_options.max_errors = config::max_errors;
|
|
_query_options.max_errors = 100;
|
|
}
|
|
|
|
if (_query_options.batch_size <= 0) {
|
|
_query_options.batch_size = DEFAULT_BATCH_SIZE;
|
|
}
|
|
|
|
_db_name = "insert_stmt";
|
|
_import_label = print_id(fragment_instance_id);
|
|
|
|
return Status::OK();
|
|
}
|
|
|
|
Status RuntimeState::init_mem_trackers(const TUniqueId& query_id) {
|
|
_query_mem_tracker = std::make_shared<MemTrackerLimiter>(
|
|
MemTrackerLimiter::Type::QUERY, fmt::format("TestQuery#Id={}", print_id(query_id)));
|
|
_scanner_mem_tracker =
|
|
std::make_shared<MemTracker>(fmt::format("TestScanner#QueryId={}", print_id(query_id)));
|
|
return Status::OK();
|
|
}
|
|
|
|
Status RuntimeState::create_block_mgr() {
|
|
DCHECK(_block_mgr2.get() == nullptr);
|
|
RETURN_IF_ERROR(BufferedBlockMgr2::create(this, runtime_profile(), _exec_env->tmp_file_mgr(),
|
|
_exec_env->disk_io_mgr()->max_read_buffer_size(),
|
|
&_block_mgr2));
|
|
return Status::OK();
|
|
}
|
|
|
|
bool RuntimeState::error_log_is_empty() {
|
|
std::lock_guard<std::mutex> l(_error_log_lock);
|
|
return (_error_log.size() > 0);
|
|
}
|
|
|
|
std::string RuntimeState::error_log() {
|
|
std::lock_guard<std::mutex> l(_error_log_lock);
|
|
return boost::algorithm::join(_error_log, "\n");
|
|
}
|
|
|
|
bool RuntimeState::log_error(const std::string& error) {
|
|
std::lock_guard<std::mutex> l(_error_log_lock);
|
|
|
|
if (_error_log.size() < _query_options.max_errors) {
|
|
_error_log.push_back(error);
|
|
return true;
|
|
}
|
|
|
|
return false;
|
|
}
|
|
|
|
void RuntimeState::log_error(const Status& status) {
|
|
if (status.ok()) {
|
|
return;
|
|
}
|
|
|
|
log_error(status.get_error_msg());
|
|
}
|
|
|
|
void RuntimeState::get_unreported_errors(std::vector<std::string>* new_errors) {
|
|
std::lock_guard<std::mutex> l(_error_log_lock);
|
|
|
|
if (_unreported_error_idx < _error_log.size()) {
|
|
new_errors->assign(_error_log.begin() + _unreported_error_idx, _error_log.end());
|
|
_unreported_error_idx = _error_log.size();
|
|
}
|
|
}
|
|
|
|
Status RuntimeState::set_mem_limit_exceeded(const std::string& msg) {
|
|
{
|
|
std::lock_guard<std::mutex> l(_process_status_lock);
|
|
if (_process_status.ok()) {
|
|
_process_status = Status::MemoryLimitExceeded(msg);
|
|
}
|
|
}
|
|
DCHECK(_process_status.is_mem_limit_exceeded());
|
|
return _process_status;
|
|
}
|
|
|
|
Status RuntimeState::check_query_state(const std::string& msg) {
|
|
// TODO: it would be nice if this also checked for cancellation, but doing so breaks
|
|
// cases where we use Status::Cancelled("Cancelled") to indicate that the limit was reached.
|
|
if (thread_context()->_thread_mem_tracker_mgr->limiter_mem_tracker()->limit_exceeded()) {
|
|
RETURN_LIMIT_EXCEEDED(this, msg);
|
|
}
|
|
return query_status();
|
|
}
|
|
|
|
const std::string ERROR_FILE_NAME = "error_log";
|
|
const int64_t MAX_ERROR_NUM = 50;
|
|
|
|
Status RuntimeState::create_load_dir() {
|
|
if (!_load_dir.empty()) {
|
|
return Status::OK();
|
|
}
|
|
RETURN_IF_ERROR(_exec_env->load_path_mgr()->allocate_dir(_db_name, _import_label, &_load_dir));
|
|
_load_dir += "/output";
|
|
return FileUtils::create_dir(_load_dir);
|
|
}
|
|
|
|
Status RuntimeState::create_error_log_file() {
|
|
// Make sure that load dir exists.
|
|
// create_load_dir();
|
|
|
|
_exec_env->load_path_mgr()->get_load_error_file_name(
|
|
_db_name, _import_label, _fragment_instance_id, &_error_log_file_path);
|
|
// std::stringstream ss;
|
|
// ss << load_dir() << "/" << ERROR_FILE_NAME
|
|
// << "_" << std::hex << fragment_instance_id().hi
|
|
// << "_" << fragment_instance_id().lo;
|
|
// _error_log_file_path = ss.str();
|
|
std::string error_log_absolute_path =
|
|
_exec_env->load_path_mgr()->get_load_error_absolute_path(_error_log_file_path);
|
|
_error_log_file = new std::ofstream(error_log_absolute_path, std::ifstream::out);
|
|
if (!_error_log_file->is_open()) {
|
|
std::stringstream error_msg;
|
|
error_msg << "Fail to open error file: [" << _error_log_file_path << "].";
|
|
LOG(WARNING) << error_msg.str();
|
|
return Status::InternalError(error_msg.str());
|
|
}
|
|
VLOG_FILE << "create error log file: " << _error_log_file_path;
|
|
|
|
return Status::OK();
|
|
}
|
|
|
|
Status RuntimeState::append_error_msg_to_file(std::function<std::string()> line,
|
|
std::function<std::string()> error_msg,
|
|
bool* stop_processing, bool is_summary) {
|
|
*stop_processing = false;
|
|
if (query_type() != TQueryType::LOAD) {
|
|
return Status::OK();
|
|
}
|
|
// If file haven't been opened, open it here
|
|
if (_error_log_file == nullptr) {
|
|
Status status = create_error_log_file();
|
|
if (!status.ok()) {
|
|
LOG(WARNING) << "Create error file log failed. because: " << status.get_error_msg();
|
|
if (_error_log_file != nullptr) {
|
|
_error_log_file->close();
|
|
delete _error_log_file;
|
|
_error_log_file = nullptr;
|
|
}
|
|
return status;
|
|
}
|
|
}
|
|
|
|
// if num of printed error row exceeds the limit, and this is not a summary message,
|
|
// if _load_zero_tolerance, return Error to stop the load process immediately.
|
|
if (_num_print_error_rows.fetch_add(1, std::memory_order_relaxed) > MAX_ERROR_NUM &&
|
|
!is_summary) {
|
|
if (_load_zero_tolerance) {
|
|
*stop_processing = true;
|
|
}
|
|
return Status::OK();
|
|
}
|
|
|
|
fmt::memory_buffer out;
|
|
if (is_summary) {
|
|
fmt::format_to(out, "Summary: {}", error_msg());
|
|
} else {
|
|
if (_error_row_number < MAX_ERROR_NUM) {
|
|
// Note: export reason first in case src line too long and be truncated.
|
|
fmt::format_to(out, "Reason: {}. src line [{}]; ", error_msg(), line());
|
|
} else if (_error_row_number == MAX_ERROR_NUM) {
|
|
fmt::format_to(out, "TOO MUCH ERROR! already reach {}. show no more next error.",
|
|
MAX_ERROR_NUM);
|
|
}
|
|
}
|
|
|
|
if (out.size() > 0) {
|
|
(*_error_log_file) << fmt::to_string(out) << std::endl;
|
|
export_load_error(fmt::to_string(out));
|
|
}
|
|
return Status::OK();
|
|
}
|
|
|
|
const int64_t HUB_MAX_ERROR_NUM = 10;
|
|
|
|
void RuntimeState::export_load_error(const std::string& err_msg) {
|
|
if (_error_hub == nullptr) {
|
|
std::lock_guard<std::mutex> lock(_create_error_hub_lock);
|
|
if (_error_hub == nullptr) {
|
|
if (_load_error_hub_info == nullptr) {
|
|
return;
|
|
}
|
|
|
|
Status st = LoadErrorHub::create_hub(_exec_env, _load_error_hub_info.get(),
|
|
_error_log_file_path, &_error_hub);
|
|
if (!st.ok()) {
|
|
LOG(WARNING) << "failed to create load error hub: " << st.get_error_msg();
|
|
return;
|
|
}
|
|
}
|
|
}
|
|
|
|
if (_error_row_number <= HUB_MAX_ERROR_NUM) {
|
|
LoadErrorHub::ErrorMsg err(_load_job_id, err_msg);
|
|
// TODO(lingbin): think if should check return value?
|
|
_error_hub->export_error(err);
|
|
}
|
|
}
|
|
|
|
int64_t RuntimeState::get_load_mem_limit() {
|
|
// TODO: the code is abandoned, it can be deleted after v1.3
|
|
if (_query_options.__isset.load_mem_limit && _query_options.load_mem_limit > 0) {
|
|
return _query_options.load_mem_limit;
|
|
} else {
|
|
return _query_mem_tracker->limit();
|
|
}
|
|
}
|
|
|
|
} // end namespace doris
|