ClickHouse/src/Storages/MergeTree/MergeTreeMarksLoader.cpp

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

190 lines
5.6 KiB
C++
Raw Normal View History

2019-11-20 13:33:41 +00:00
#include <Storages/MergeTree/MergeTreeMarksLoader.h>
2020-02-03 12:46:25 +00:00
#include <Storages/MergeTree/MergeTreeData.h>
#include <Common/MemoryTrackerBlockerInThread.h>
2020-02-03 12:46:25 +00:00
#include <IO/ReadBufferFromFile.h>
2023-01-19 15:55:15 +00:00
#include <Interpreters/threadPoolCallbackRunner.h>
2022-05-29 07:28:02 +00:00
#include <Compression/CompressedReadBufferFromFile.h>
2022-08-30 17:47:34 +00:00
#include <Common/setThreadName.h>
#include <Common/scope_guard_safe.h>
#include <Common/CurrentMetrics.h>
#include <Common/ThreadPool.h>
2019-11-20 13:33:41 +00:00
#include <utility>
2022-08-30 17:47:34 +00:00
namespace ProfileEvents
{
extern const Event WaitMarksLoadMicroseconds;
2022-09-04 17:10:46 +00:00
extern const Event BackgroundLoadingMarksTasks;
2022-08-30 17:47:34 +00:00
}
2019-11-20 13:33:41 +00:00
namespace DB
{
2020-02-03 12:46:25 +00:00
namespace ErrorCodes
{
2020-02-25 18:02:41 +00:00
extern const int CANNOT_READ_ALL_DATA;
2020-02-03 12:46:25 +00:00
extern const int CORRUPTED_DATA;
extern const int LOGICAL_ERROR;
}
2019-11-20 13:33:41 +00:00
MergeTreeMarksLoader::MergeTreeMarksLoader(
2022-04-12 18:59:49 +00:00
DataPartStoragePtr data_part_storage_,
2019-11-20 13:33:41 +00:00
MarkCache * mark_cache_,
const String & mrk_path_,
2020-02-03 12:46:25 +00:00
size_t marks_count_,
const MergeTreeIndexGranularityInfo & index_granularity_info_,
2019-11-20 13:33:41 +00:00
bool save_marks_in_cache_,
2022-07-18 12:09:57 +00:00
const ReadSettings & read_settings_,
ThreadPool * load_marks_threadpool_,
2020-02-03 12:46:25 +00:00
size_t columns_in_mark_)
2022-04-12 18:59:49 +00:00
: data_part_storage(std::move(data_part_storage_))
, mark_cache(mark_cache_)
2019-11-20 13:33:41 +00:00
, mrk_path(mrk_path_)
2020-02-03 12:46:25 +00:00
, marks_count(marks_count_)
, index_granularity_info(index_granularity_info_)
2019-11-20 13:33:41 +00:00
, save_marks_in_cache(save_marks_in_cache_)
2022-07-18 12:09:57 +00:00
, columns_in_mark(columns_in_mark_)
, read_settings(read_settings_)
, load_marks_threadpool(load_marks_threadpool_)
2022-07-18 12:09:57 +00:00
{
if (load_marks_threadpool)
2022-08-30 17:47:34 +00:00
{
future = loadMarksAsync();
}
2022-07-18 12:09:57 +00:00
}
2019-11-20 13:33:41 +00:00
2022-08-31 13:39:53 +00:00
MergeTreeMarksLoader::~MergeTreeMarksLoader()
{
if (future.valid())
{
future.wait();
}
2022-07-18 12:09:57 +00:00
}
2019-11-20 13:33:41 +00:00
const MarkInCompressedFile & MergeTreeMarksLoader::getMark(size_t row_index, size_t column_index)
{
if (!marks)
2022-08-30 17:47:34 +00:00
{
Stopwatch watch(CLOCK_MONOTONIC);
if (future.valid())
{
marks = future.get();
future = {};
}
else
{
marks = loadMarks();
}
watch.stop();
ProfileEvents::increment(ProfileEvents::WaitMarksLoadMicroseconds, watch.elapsedMicroseconds());
}
2020-02-03 12:46:25 +00:00
#ifndef NDEBUG
if (column_index >= columns_in_mark)
2022-08-28 20:33:42 +00:00
throw Exception(ErrorCodes::LOGICAL_ERROR, "Column index: {} is out of range [0, {})", column_index, columns_in_mark);
2020-02-03 12:46:25 +00:00
#endif
return (*marks)[row_index * columns_in_mark + column_index];
}
2020-02-03 12:46:25 +00:00
MarkCache::MappedPtr MergeTreeMarksLoader::loadMarksImpl()
{
/// Memory for marks must not be accounted as memory usage for query, because they are stored in shared cache.
MemoryTrackerBlockerInThread temporarily_disable_memory_tracker;
2020-02-03 12:46:25 +00:00
2022-04-12 18:59:49 +00:00
size_t file_size = data_part_storage->getFileSize(mrk_path);
2020-02-03 12:46:25 +00:00
size_t mark_size = index_granularity_info.getMarkSizeInBytes(columns_in_mark);
size_t expected_uncompressed_size = mark_size * marks_count;
2020-02-03 12:46:25 +00:00
auto res = std::make_shared<MarksInCompressedFile>(marks_count * columns_in_mark);
2022-09-06 15:41:39 +00:00
if (!index_granularity_info.mark_type.compressed && expected_uncompressed_size != file_size)
2020-02-03 12:46:25 +00:00
throw Exception(
2022-04-12 18:59:49 +00:00
ErrorCodes::CORRUPTED_DATA,
"Bad size of marks file '{}': {}, must be: {}",
std::string(fs::path(data_part_storage->getFullPath()) / mrk_path),
file_size, expected_uncompressed_size);
auto buffer = data_part_storage->readFile(mrk_path, read_settings.adjustBufferSize(file_size), file_size, std::nullopt);
std::unique_ptr<ReadBuffer> reader;
2022-09-06 15:41:39 +00:00
if (!index_granularity_info.mark_type.compressed)
reader = std::move(buffer);
2022-08-29 17:09:58 +00:00
else
reader = std::make_unique<CompressedReadBufferFromFile>(std::move(buffer));
2020-02-03 12:46:25 +00:00
2022-09-05 05:26:58 +00:00
if (!index_granularity_info.mark_type.adaptive)
2020-02-03 12:46:25 +00:00
{
/// Read directly to marks.
reader->readStrict(reinterpret_cast<char *>(res->data()), expected_uncompressed_size);
2020-02-03 12:46:25 +00:00
if (!reader->eof())
2022-08-28 20:33:42 +00:00
throw Exception(ErrorCodes::CANNOT_READ_ALL_DATA,
"Cannot read all marks from file {}, is eof: {}, buffer size: {}, file size: {}",
mrk_path, reader->eof(), reader->buffer().size(), file_size);
2020-02-03 12:46:25 +00:00
}
else
{
size_t i = 0;
size_t granularity;
while (!reader->eof())
2020-02-03 12:46:25 +00:00
{
res->read(*reader, i * columns_in_mark, columns_in_mark);
readIntBinary(granularity, *reader);
2020-02-03 12:46:25 +00:00
++i;
}
2022-08-28 18:42:11 +00:00
if (i * mark_size != expected_uncompressed_size)
throw Exception(ErrorCodes::CANNOT_READ_ALL_DATA, "Cannot read all marks from file {}", mrk_path);
2020-02-03 12:46:25 +00:00
}
2020-02-03 12:46:25 +00:00
res->protect();
return res;
2019-11-20 13:33:41 +00:00
}
2022-08-30 17:47:34 +00:00
MarkCache::MappedPtr MergeTreeMarksLoader::loadMarks()
2019-11-20 13:33:41 +00:00
{
2022-08-30 17:47:34 +00:00
MarkCache::MappedPtr loaded_marks;
2019-11-20 13:33:41 +00:00
if (mark_cache)
{
2022-04-22 20:29:14 +00:00
auto key = mark_cache->hash(fs::path(data_part_storage->getFullPath()) / mrk_path);
2019-11-20 13:33:41 +00:00
if (save_marks_in_cache)
{
2020-04-22 06:22:14 +00:00
auto callback = [this]{ return loadMarksImpl(); };
2022-08-30 17:47:34 +00:00
loaded_marks = mark_cache->getOrSet(key, callback);
2019-11-20 13:33:41 +00:00
}
else
{
2022-08-30 17:47:34 +00:00
loaded_marks = mark_cache->get(key);
if (!loaded_marks)
loaded_marks = loadMarksImpl();
2019-11-20 13:33:41 +00:00
}
}
else
2022-08-30 17:47:34 +00:00
loaded_marks = loadMarksImpl();
2019-11-20 13:33:41 +00:00
2022-08-30 17:47:34 +00:00
if (!loaded_marks)
{
throw Exception(
ErrorCodes::LOGICAL_ERROR, "Failed to load marks: {}",
(fs::path(data_part_storage->getFullPath()) / mrk_path).string());
}
return loaded_marks;
}
std::future<MarkCache::MappedPtr> MergeTreeMarksLoader::loadMarksAsync()
{
2023-01-19 15:55:15 +00:00
return scheduleFromThreadPool<MarkCache::MappedPtr>([this]() -> MarkCache::MappedPtr
{
ProfileEvents::increment(ProfileEvents::BackgroundLoadingMarksTasks);
2023-01-19 17:44:29 +00:00
return loadMarks();
2023-01-19 15:55:15 +00:00
}, *load_marks_threadpool, "LoadMarksThread");
2019-11-20 13:33:41 +00:00
}
}