From 12bce0691c1e291374904ceb0eb5726e25fd8b24 Mon Sep 17 00:00:00 2001 From: JaySon Date: Wed, 15 May 2024 11:40:14 +0800 Subject: [PATCH] This is an automated cherry-pick of #9041 Signed-off-by: ti-chi-bot --- .../DeltaMerge/StoragePool/StoragePool.cpp | 1040 +++++++++++++++++ .../V3/tests/gtest_page_storage_mix_mode.cpp | 97 +- 2 files changed, 1120 insertions(+), 17 deletions(-) create mode 100644 dbms/src/Storages/DeltaMerge/StoragePool/StoragePool.cpp diff --git a/dbms/src/Storages/DeltaMerge/StoragePool/StoragePool.cpp b/dbms/src/Storages/DeltaMerge/StoragePool/StoragePool.cpp new file mode 100644 index 00000000000..7614fd85115 --- /dev/null +++ b/dbms/src/Storages/DeltaMerge/StoragePool/StoragePool.cpp @@ -0,0 +1,1040 @@ +// Copyright 2024 PingCAP, Inc. +// +// Licensed 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. + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include + +namespace CurrentMetrics +{ +extern const Metric StoragePoolV2Only; +extern const Metric StoragePoolV3Only; +extern const Metric StoragePoolMixMode; +extern const Metric StoragePoolUniPS; +} // namespace CurrentMetrics + +namespace DB +{ +namespace ErrorCodes +{ +extern const int LOGICAL_ERROR; +} // namespace ErrorCodes + +namespace FailPoints +{ +extern const char force_set_dtfile_exist_when_acquire_id[]; +} // namespace FailPoints +} // namespace DB + +namespace DB::DM +{ + +StoragePool::StoragePool( + Context & global_ctx, + KeyspaceID keyspace_id_, + NamespaceID table_id_, + StoragePathPool & storage_path_pool_, + const String & name) + : StoragePool(global_ctx, keyspace_id_, table_id_, storage_path_pool_, global_ctx.getGlobalPageIdAllocator(), name) +{ + // +} + +StoragePool::StoragePool( + Context & global_ctx, + KeyspaceID keyspace_id_, + NamespaceID table_id_, + StoragePathPool & storage_path_pool_, + GlobalPageIdAllocatorPtr page_id_allocator_, + const String & name) + : logger(Logger::get(!name.empty() ? name : DB::toString(table_id_))) + , run_mode(global_ctx.getPageStorageRunMode()) + , keyspace_id(keyspace_id_) + , table_id(table_id_) + , storage_path_pool(storage_path_pool_) + , uni_ps(run_mode == PageStorageRunMode::UNI_PS ? global_ctx.getWriteNodePageStorage() : nullptr) + , global_context(global_ctx) + , global_id_allocator(std::move(page_id_allocator_)) + , storage_pool_metrics(CurrentMetrics::StoragePoolV3Only, 0) +{ + assert(global_id_allocator != nullptr); + const auto & global_storage_pool = global_context.getGlobalStoragePool(); + switch (run_mode) + { + case PageStorageRunMode::ONLY_V2: + { + log_storage_v2 = PageStorage::create( + name + ".log", + storage_path_pool.getPSDiskDelegatorMulti("log"), + extractConfig(global_context.getSettingsRef(), StorageType::Log), + global_context.getFileProvider(), + global_context); + data_storage_v2 = PageStorage::create( + name + ".data", + storage_path_pool.getPSDiskDelegatorSingle("data"), // keep for behavior not changed + extractConfig(global_context.getSettingsRef(), StorageType::Data), + global_context.getFileProvider(), + global_context); + meta_storage_v2 = PageStorage::create( + name + ".meta", + storage_path_pool.getPSDiskDelegatorMulti("meta"), + extractConfig(global_context.getSettingsRef(), StorageType::Meta), + global_context.getFileProvider(), + global_context); + log_storage_reader = std::make_shared( + run_mode, + keyspace_id, + StorageType::Log, + table_id, + log_storage_v2, + /*storage_v3_*/ nullptr, + /*uni_ps_*/ nullptr, + nullptr); + data_storage_reader = std::make_shared( + run_mode, + keyspace_id, + StorageType::Data, + table_id, + data_storage_v2, + /*storage_v3_*/ nullptr, + /*uni_ps_*/ nullptr, + nullptr); + meta_storage_reader = std::make_shared( + run_mode, + keyspace_id, + StorageType::Meta, + table_id, + meta_storage_v2, + /*storage_v3_*/ nullptr, + /*uni_ps_*/ nullptr, + nullptr); + + log_storage_writer = std::make_shared( + run_mode, + StorageType::Log, + log_storage_v2, + /*storage_v3_*/ nullptr, + /*uni_ps_*/ nullptr); + data_storage_writer = std::make_shared( + run_mode, + StorageType::Data, + data_storage_v2, + /*storage_v3_*/ nullptr, + /*uni_ps_*/ nullptr); + meta_storage_writer = std::make_shared( + run_mode, + StorageType::Meta, + meta_storage_v2, + /*storage_v3_*/ nullptr, + /*uni_ps_*/ nullptr); + break; + } + case PageStorageRunMode::ONLY_V3: + { + assert(global_storage_pool != nullptr); + log_storage_v3 = global_storage_pool->log_storage; + data_storage_v3 = global_storage_pool->data_storage; + meta_storage_v3 = global_storage_pool->meta_storage; + + log_storage_reader = std::make_shared( + run_mode, + keyspace_id, + StorageType::Log, + table_id, + /*storage_v2_*/ nullptr, + log_storage_v3, + /*uni_ps_*/ nullptr, + nullptr); + data_storage_reader = std::make_shared( + run_mode, + keyspace_id, + StorageType::Data, + table_id, + /*storage_v2_*/ nullptr, + data_storage_v3, + /*uni_ps_*/ nullptr, + nullptr); + meta_storage_reader = std::make_shared( + run_mode, + keyspace_id, + StorageType::Meta, + table_id, + /*storage_v2_*/ nullptr, + meta_storage_v3, + /*uni_ps_*/ nullptr, + nullptr); + + log_storage_writer = std::make_shared( + run_mode, + StorageType::Log, + /*storage_v2_*/ nullptr, + log_storage_v3, + /*uni_ps_*/ nullptr); + data_storage_writer = std::make_shared( + run_mode, + StorageType::Data, + /*storage_v2_*/ nullptr, + data_storage_v3, + /*uni_ps_*/ nullptr); + meta_storage_writer = std::make_shared( + run_mode, + StorageType::Meta, + /*storage_v2_*/ nullptr, + meta_storage_v3, + /*uni_ps_*/ nullptr); + break; + } + case PageStorageRunMode::MIX_MODE: + { + assert(global_storage_pool != nullptr); + log_storage_v3 = global_storage_pool->log_storage; + data_storage_v3 = global_storage_pool->data_storage; + meta_storage_v3 = global_storage_pool->meta_storage; + + if (storage_path_pool.isPSV2Deleted()) + { + LOG_INFO( + logger, + "PageStorage V2 is already mark deleted. Current pagestorage change from {} to {}, table_id={}", + magic_enum::enum_name(PageStorageRunMode::MIX_MODE), + magic_enum::enum_name(PageStorageRunMode::ONLY_V3), + table_id); + log_storage_v2 = nullptr; + data_storage_v2 = nullptr; + meta_storage_v2 = nullptr; + run_mode = PageStorageRunMode::ONLY_V3; + storage_path_pool.clearPSV2ObsoleteData(); + } + else + { + // Although there is no more write to ps v2 in mixed mode, the ps instances will keep running if there is some data in log storage when restart, + // so we keep its original config here. + // And we rely on the mechanism that writing file will be rotated if no valid pages in non writing files to reduce the disk space usage of these ps instances. + log_storage_v2 = PageStorage::create( + name + ".log", + storage_path_pool.getPSDiskDelegatorMulti("log"), + extractConfig(global_context.getSettingsRef(), StorageType::Log), + global_context.getFileProvider(), + global_context, + /* use_v3 */ false, + /* no_more_write_to_v2 */ true); + data_storage_v2 = PageStorage::create( + name + ".data", + storage_path_pool.getPSDiskDelegatorMulti("data"), + extractConfig(global_context.getSettingsRef(), StorageType::Data), + global_context.getFileProvider(), + global_context, + /* use_v3 */ false, + /* no_more_write_to_v2 */ true); + meta_storage_v2 = PageStorage::create( + name + ".meta", + storage_path_pool.getPSDiskDelegatorMulti("meta"), + extractConfig(global_context.getSettingsRef(), StorageType::Meta), + global_context.getFileProvider(), + global_context, + /* use_v3 */ false, + /* no_more_write_to_v2 */ true); + } + + log_storage_reader = std::make_shared( + run_mode, + keyspace_id, + StorageType::Log, + table_id, + log_storage_v2, + log_storage_v3, + /*uni_ps_*/ nullptr, + nullptr); + data_storage_reader = std::make_shared( + run_mode, + keyspace_id, + StorageType::Data, + table_id, + data_storage_v2, + data_storage_v3, + /*uni_ps_*/ nullptr, + nullptr); + meta_storage_reader = std::make_shared( + run_mode, + keyspace_id, + StorageType::Meta, + table_id, + meta_storage_v2, + meta_storage_v3, + /*uni_ps_*/ nullptr, + nullptr); + + log_storage_writer = std::make_shared( + run_mode, + StorageType::Log, + log_storage_v2, + log_storage_v3, + /*uni_ps_*/ nullptr); + data_storage_writer = std::make_shared( + run_mode, + StorageType::Data, + data_storage_v2, + data_storage_v3, + /*uni_ps_*/ nullptr); + meta_storage_writer = std::make_shared( + run_mode, + StorageType::Meta, + meta_storage_v2, + meta_storage_v3, + /*uni_ps_*/ nullptr); + break; + } + case PageStorageRunMode::UNI_PS: + { + log_storage_reader = std::make_shared( + run_mode, + keyspace_id, + StorageType::Log, + table_id, + /*storage_v2_*/ nullptr, + /*storage_v3_*/ nullptr, + uni_ps, + nullptr); + data_storage_reader = std::make_shared( + run_mode, + keyspace_id, + StorageType::Data, + table_id, + /*storage_v2_*/ nullptr, + /*storage_v3_*/ nullptr, + uni_ps, + nullptr); + meta_storage_reader = std::make_shared( + run_mode, + keyspace_id, + StorageType::Meta, + table_id, + /*storage_v2_*/ nullptr, + /*storage_v3_*/ nullptr, + uni_ps, + nullptr); + + log_storage_writer = std::make_shared( + run_mode, + StorageType::Log, + /*storage_v2_*/ nullptr, + /*storage_v3_*/ nullptr, + uni_ps); + data_storage_writer = std::make_shared( + run_mode, + StorageType::Data, + /*storage_v2_*/ nullptr, + /*storage_v3_*/ nullptr, + uni_ps); + meta_storage_writer = std::make_shared( + run_mode, + StorageType::Meta, + /*storage_v2_*/ nullptr, + /*storage_v3_*/ nullptr, + uni_ps); + break; + } + default: + throw Exception(ErrorCodes::LOGICAL_ERROR, "Unknown PageStorageRunMode {}", static_cast(run_mode)); + } +} + +void StoragePool::forceTransformMetaV2toV3() +{ + RUNTIME_CHECK_MSG( + run_mode == PageStorageRunMode::MIX_MODE, + "Transform meta must run under mix mode, run_mode={}", + static_cast(run_mode)); + + assert(meta_storage_v2 != nullptr); + assert(meta_storage_v3 != nullptr); + auto meta_transform_storage_writer = std::make_shared( + run_mode, + StorageType::Meta, + meta_storage_v2, + meta_storage_v3, + /*uni_ps_*/ nullptr); + auto meta_transform_storage_reader = std::make_shared( + run_mode, + keyspace_id, + StorageType::Meta, + table_id, + meta_storage_v2, + meta_storage_v3, + /*uni_ps_*/ nullptr, + nullptr); + + Pages pages_transform = {}; + auto meta_transform_acceptor = [&](const DB::Page & page) { + pages_transform.emplace_back(page); + }; + + meta_transform_storage_reader->traverse(meta_transform_acceptor, /*only_v2*/ true, /*only_v3*/ false); + + WriteBatch write_batch_transform{table_id}; + WriteBatch write_batch_del_v2{table_id}; + + for (const auto & page_transform : pages_transform) + { + // Check pages have not contain field offset + // Also get the tag of page_id + const auto & page_transform_entry = meta_transform_storage_reader->getPageEntry(page_transform.page_id); + if (!page_transform_entry.field_offsets.empty()) + { + throw Exception( + ErrorCodes::LOGICAL_ERROR, + "Can't transform meta from V2 to V3, page_id={} {}", // + page_transform.page_id, + page_transform_entry.toDebugString()); + } + + write_batch_transform.putPage( + page_transform.page_id, // + page_transform_entry.tag, + std::make_shared(page_transform.data.begin(), page_transform.data.size()), + page_transform.data.size()); + // Record del for V2 + write_batch_del_v2.delPage(page_transform.page_id); + } + + // Will rewrite into V3. + meta_transform_storage_writer->writeIntoV3(std::move(write_batch_transform), nullptr); + + // DEL must call after rewrite. + meta_transform_storage_writer->writeIntoV2(std::move(write_batch_del_v2), nullptr); +} + +static inline DB::PS::V2::PageEntriesVersionSetWithDelta::Snapshot * toV2ConcreteSnapshot( + const DB::PageStorage::SnapshotPtr & ptr) +{ + return dynamic_cast(ptr.get()); +} + +void StoragePool::forceTransformDataV2toV3() +{ + RUNTIME_CHECK_MSG( + run_mode == PageStorageRunMode::MIX_MODE, + "Transform data must run under mix mode, run_mode={}", + static_cast(run_mode)); + + assert(data_storage_v2 != nullptr); + assert(data_storage_v3 != nullptr); + auto data_transform_storage_writer = std::make_shared( + run_mode, + StorageType::Data, + data_storage_v2, + data_storage_v3, + /*uni_ps_*/ nullptr); + + auto snapshot = data_storage_v2->getSnapshot("transformDataV2toV3"); + auto * v2_snap = toV2ConcreteSnapshot(snapshot); + if (!snapshot || !v2_snap) + { + throw Exception("Can not allocate snapshot from pool.data v2", ErrorCodes::LOGICAL_ERROR); + } + + // Example + // 100 -> 100 + // 102 -> 100 + // 105 -> 100 + // 200 -> 200 + // 305 -> 300 + // Migration steps: + // collect v2 valid page id: 100, 102, 105, 200, 305 + // v3 put external 100, 200, 300; put ref 102, 105, 305 + // mark some id as deleted: v3 del 300 + // v2 delete 100, 102, 105, 200, 305 + + // The page ids that can be accessed by DeltaTree + const auto all_page_ids = v2_snap->view.validPageIds(); + + WriteBatch write_batch_transform{table_id}; + WriteBatch write_batch_del_v2{table_id}; + + PageIdU64Set created_dt_file_id; + for (const auto page_id : all_page_ids) + { + // resolve the page_id into dtfile id + const auto resolved_file_id = v2_snap->view.resolveRefId(page_id); + if (auto ins_result = created_dt_file_id.insert(resolved_file_id); /*created=*/ins_result.second) + { + // first see this file id, migrate to v3 + write_batch_transform.putExternal(resolved_file_id, 0); + } + // migrate the reference for v3 + if (page_id != resolved_file_id) + { + write_batch_transform.putRefPage(page_id, resolved_file_id); + } + // record del for V2 + write_batch_del_v2.delPage(page_id); + } + // If the file id is not existed in `all_page_ids`, it means the file id + // itself has been deleted. + for (const auto dt_file_id : created_dt_file_id) + { + if (all_page_ids.count(dt_file_id) == 0) + { + write_batch_transform.delPage(dt_file_id); + } + } + + // Will rewrite into V3. + data_transform_storage_writer->writeIntoV3(std::move(write_batch_transform), nullptr); + + // DEL must call after rewrite. + data_transform_storage_writer->writeIntoV2(std::move(write_batch_del_v2), nullptr); +} + +PageStorageRunMode StoragePool::restore() +{ + switch (run_mode) + { + case PageStorageRunMode::ONLY_V2: + { + // Restore the PSV2 instances from disk + log_storage_v2->restore(); + data_storage_v2->restore(); + meta_storage_v2->restore(); + + // ONLY_V2, make sure the page_ids is larger than that in PSV2 + global_id_allocator->raiseLogPageIdLowerBound(log_storage_v2->getMaxId()); + global_id_allocator->raiseDataPageIdLowerBound(data_storage_v2->getMaxId()); + global_id_allocator->raiseMetaPageIdLowerBound(meta_storage_v2->getMaxId()); + + storage_pool_metrics = CurrentMetrics::Increment{CurrentMetrics::StoragePoolV2Only}; + break; + } + case PageStorageRunMode::ONLY_V3: + { + // ONLY_V3 + // - StoragePool is simply a wrapper for the PS instances in GlobalStoragePool + // - Make sure the page_ids is larger than that in PSV2 + global_id_allocator->raiseLogPageIdLowerBound(log_storage_v3->getMaxId()); + global_id_allocator->raiseDataPageIdLowerBound(data_storage_v3->getMaxId()); + global_id_allocator->raiseMetaPageIdLowerBound(meta_storage_v3->getMaxId()); + + storage_pool_metrics = CurrentMetrics::Increment{CurrentMetrics::StoragePoolV3Only}; + break; + } + case PageStorageRunMode::MIX_MODE: + { + // Restore the PSV2 instances from disk + log_storage_v2->restore(); + data_storage_v2->restore(); + meta_storage_v2->restore(); + + // The pages on data and log can be rewritten to V3 and the old pages on V2 are deleted by `delta merge`. + // However, the pages on meta V2 can not be deleted. As the pages in meta are small, we perform a forceTransformMetaV2toV3 to convert pages before all. + if (const auto & meta_remain_pages = meta_storage_v2->getNumberOfPages(); meta_remain_pages != 0) + { + LOG_INFO( + logger, + "Current pool.meta transform to V3 begin, table_id={} pages_before_transform={}", + table_id, + meta_remain_pages); + forceTransformMetaV2toV3(); + const auto & meta_remain_pages_after_transform = meta_storage_v2->getNumberOfPages(); + LOG_INFO( + logger, + "Current pool.meta transform to V3 finished, table_id={} done={} pages_before_transform={} " + "pages_after_transform={}", // + table_id, + meta_remain_pages_after_transform == 0, + meta_remain_pages, + meta_remain_pages_after_transform); + } + else + { + LOG_INFO(logger, "Current pool.meta transform already done before restored, table_id={}", table_id); + } + + if (const auto & data_remain_pages = data_storage_v2->getNumberOfPages(); data_remain_pages != 0) + { + LOG_INFO( + logger, + "Current pool.data transform to V3 begin, table_id={} pages_before_transform={}", + table_id, + data_remain_pages); + forceTransformDataV2toV3(); + const auto & data_remain_pages_after_transform = data_storage_v2->getNumberOfPages(); + LOG_INFO( + logger, + "Current pool.data transform to V3 finished, table_id={} done={} pages_before_transform={} " + "pages_after_transform={}", // + table_id, + data_remain_pages_after_transform == 0, + data_remain_pages, + data_remain_pages_after_transform); + } + else + { + LOG_INFO(logger, "Current pool.data transform already done before restored, table_id={}", table_id); + } + + // Though all the pages may have been transformed into PageStoage V3 format, we still need + // to ensure the following allocated page_ids is larger than that in both v2 and v3. + // Because `PageStorageV3->getMaxId` is not accurate after the previous "meta" and "data" + // transformed from v2 to v3. + auto max_log_page_id = std::max(log_storage_v2->getMaxId(), log_storage_v3->getMaxId()); + auto max_data_page_id = std::max(data_storage_v2->getMaxId(), data_storage_v3->getMaxId()); + auto max_meta_page_id = std::max(meta_storage_v2->getMaxId(), meta_storage_v3->getMaxId()); + global_id_allocator->raiseLogPageIdLowerBound(max_log_page_id); + global_id_allocator->raiseDataPageIdLowerBound(max_data_page_id); + global_id_allocator->raiseMetaPageIdLowerBound(max_meta_page_id); + + // Check number of valid pages in v2 + // If V2 already have no any data in disk, Then change run_mode to ONLY_V3 + if (log_storage_v2->getNumberOfPages() == 0 && data_storage_v2->getNumberOfPages() == 0 + && meta_storage_v2->getNumberOfPages() == 0) + { + LOG_INFO( + logger, + "Current pagestorage change from {} to {}, table_id={}", + magic_enum::enum_name(PageStorageRunMode::MIX_MODE), + magic_enum::enum_name(PageStorageRunMode::ONLY_V3), + table_id); + + if (storage_path_pool.createPSV2DeleteMarkFile()) + { + log_storage_v2->drop(); + data_storage_v2->drop(); + meta_storage_v2->drop(); + } + log_storage_v2 = nullptr; + data_storage_v2 = nullptr; + meta_storage_v2 = nullptr; + + // Must init by PageStorageRunMode::ONLY_V3 + log_storage_reader = std::make_shared( + PageStorageRunMode::ONLY_V3, + keyspace_id, + StorageType::Log, + table_id, + /*storage_v2_*/ nullptr, + log_storage_v3, + /*uni_ps_*/ nullptr, + nullptr); + data_storage_reader = std::make_shared( + PageStorageRunMode::ONLY_V3, + keyspace_id, + StorageType::Data, + table_id, + /*storage_v2_*/ nullptr, + data_storage_v3, + /*uni_ps_*/ nullptr, + nullptr); + meta_storage_reader = std::make_shared( + PageStorageRunMode::ONLY_V3, + keyspace_id, + StorageType::Meta, + table_id, + /*storage_v2_*/ nullptr, + meta_storage_v3, + /*uni_ps_*/ nullptr, + nullptr); + + log_storage_writer = std::make_shared( + PageStorageRunMode::ONLY_V3, + StorageType::Log, + /*storage_v2_*/ nullptr, + log_storage_v3, + /*uni_ps_*/ nullptr); + data_storage_writer = std::make_shared( + PageStorageRunMode::ONLY_V3, + StorageType::Data, + /*storage_v2_*/ nullptr, + data_storage_v3, + /*uni_ps_*/ nullptr); + meta_storage_writer = std::make_shared( + PageStorageRunMode::ONLY_V3, + StorageType::Meta, + /*storage_v2_*/ nullptr, + meta_storage_v3, + /*uni_ps_*/ nullptr); + + run_mode = PageStorageRunMode::ONLY_V3; + storage_pool_metrics = CurrentMetrics::Increment{CurrentMetrics::StoragePoolV3Only}; + } + else // Still running Mix Mode + { + storage_pool_metrics = CurrentMetrics::Increment{CurrentMetrics::StoragePoolMixMode}; + } + break; + } + case PageStorageRunMode::UNI_PS: + { + // UNI_PS + // - StoragePool is simply a wrapper for the uni_ps + auto max_page_id = uni_ps->getMaxIdAfterRestart(); + global_id_allocator->raiseLogPageIdLowerBound(max_page_id); + global_id_allocator->raiseDataPageIdLowerBound(max_page_id); + global_id_allocator->raiseMetaPageIdLowerBound(max_page_id); + + storage_pool_metrics = CurrentMetrics::Increment{CurrentMetrics::StoragePoolUniPS}; + break; + } + default: + throw Exception(ErrorCodes::LOGICAL_ERROR, "Unknown PageStorageRunMode {}", static_cast(run_mode)); + } + + const auto [max_log_page_id, max_data_page_id, max_meta_page_id] = global_id_allocator->getCurrentIds(); + LOG_INFO( + logger, + "Finished StoragePool restore. run_mode={} keyspace_id={} table_id={}" + " max_log_page_id={} max_data_page_id={} max_meta_page_id={}", + magic_enum::enum_name(run_mode), + keyspace_id, + table_id, + max_log_page_id, + max_data_page_id, + max_meta_page_id); + return run_mode; +} + +StoragePool::~StoragePool() +{ + shutdown(); +} + +void StoragePool::startup(ExternalPageCallbacks && callbacks) +{ + switch (run_mode) + { + case PageStorageRunMode::ONLY_V2: + { + // For V2, we need a per physical table gc handle to perform the gc of its PageStorage instances. + data_storage_v2->registerExternalPagesCallbacks(callbacks); + gc_handle + = global_context.getBackgroundPool().addTask([this] { return this->gc(global_context.getSettingsRef()); }); + break; + } + case PageStorageRunMode::ONLY_V3: + { + // For V3, the GC is handled by `GlobalStoragePool::gc`, just register callbacks is OK. + data_storage_v3->registerExternalPagesCallbacks(callbacks); + break; + } + case PageStorageRunMode::MIX_MODE: + { + // For V3, the GC is handled by `GlobalStoragePool::gc`. + // Since we have transformed all external pages from V2 to V3 in `StoragePool::restore`, + // just register callbacks to V3 is OK + data_storage_v3->registerExternalPagesCallbacks(callbacks); + // we still need a gc_handle to reclaim the V2 disk space. + gc_handle + = global_context.getBackgroundPool().addTask([this] { return this->gc(global_context.getSettingsRef()); }); + break; + } + case PageStorageRunMode::UNI_PS: + { + // For uni ps, the GC is handled by `UniversalPageStorageService`, register callbacks with prefix for this table + UniversalExternalPageCallbacks us_callbacks; + us_callbacks.remover = std::move(callbacks.remover); + us_callbacks.scanner = std::move(callbacks.scanner); + us_callbacks.prefix = UniversalPageIdFormat::toFullPrefix(keyspace_id, StorageType::Data, table_id); + uni_ps->registerUniversalExternalPagesCallbacks(us_callbacks); + break; + } + default: + throw Exception(ErrorCodes::LOGICAL_ERROR, "Unknown PageStorageRunMode {}", static_cast(run_mode)); + } +} + +void StoragePool::shutdown() +{ + // Note: Should reset the gc_handle before unregistering the pages callbacks + if (gc_handle) + { + global_context.getBackgroundPool().removeTask(gc_handle); + gc_handle = nullptr; + } + + switch (run_mode) + { + case PageStorageRunMode::ONLY_V2: + { + meta_storage_v2->shutdown(); + log_storage_v2->shutdown(); + data_storage_v2->shutdown(); + data_storage_v2->unregisterExternalPagesCallbacks(table_id); + break; + } + case PageStorageRunMode::ONLY_V3: + { + data_storage_v3->unregisterExternalPagesCallbacks(table_id); + break; + } + case PageStorageRunMode::MIX_MODE: + { + meta_storage_v2->shutdown(); + log_storage_v2->shutdown(); + data_storage_v2->shutdown(); + // We have transformed all external pages from V2 to V3 in `restore`, so + // only need to unregister callbacks for V3. + data_storage_v3->unregisterExternalPagesCallbacks(table_id); + break; + } + case PageStorageRunMode::UNI_PS: + { + uni_ps->unregisterUniversalExternalPagesCallbacks( + UniversalPageIdFormat::toFullPrefix(keyspace_id, StorageType::Data, table_id)); + break; + } + default: + throw Exception(ErrorCodes::LOGICAL_ERROR, "Unknown PageStorageRunMode {}", static_cast(run_mode)); + } +} + +bool StoragePool::doV2Gc(const Settings & settings) const +{ + bool done_anything = false; + auto write_limiter = global_context.getWriteLimiter(); + auto read_limiter = global_context.getReadLimiter(); + + auto config = extractConfig(settings, StorageType::Meta); + meta_storage_v2->reloadSettings(config); + done_anything |= meta_storage_v2->gc(/*not_skip*/ false, write_limiter, read_limiter); + + config = extractConfig(settings, StorageType::Data); + data_storage_v2->reloadSettings(config); + done_anything |= data_storage_v2->gc(/*not_skip*/ false, write_limiter, read_limiter); + + config = extractConfig(settings, StorageType::Log); + log_storage_v2->reloadSettings(config); + done_anything |= log_storage_v2->gc(/*not_skip*/ false, write_limiter, read_limiter); + return done_anything; +} + +bool StoragePool::gc(const Settings & settings, const Seconds & try_gc_period) +{ + if (run_mode == PageStorageRunMode::ONLY_V3 || run_mode == PageStorageRunMode::UNI_PS) + return false; + + { + std::lock_guard lock(mutex); + // Just do gc for owned storage, otherwise the gc will be handled globally + + Timepoint now = Clock::now(); + if (now < (last_try_gc_time.load() + try_gc_period)) + return false; + + last_try_gc_time = now; + } + + // Only do the v2 GC + return doV2Gc(settings); +} + +void StoragePool::drop() +{ + shutdown(); + + if (run_mode == PageStorageRunMode::ONLY_V2 || run_mode == PageStorageRunMode::MIX_MODE) + { + meta_storage_v2->drop(); + data_storage_v2->drop(); + log_storage_v2->drop(); + } +} + +PageIdU64 StoragePool::newDataPageIdForDTFile(StableDiskDelegator & delegator, const char * who) const +{ + // In case that there is a DTFile created on disk but TiFlash crashes without persisting the ID. + // After TiFlash process restored, the ID will be inserted into the stable delegator, but we may + // get a duplicated ID from the `storage_pool.data`. (tics#2756) + PageIdU64 dtfile_id; + do + { + dtfile_id = global_id_allocator->newDataPageIdForDTFile(); + + auto existed_path = delegator.getDTFilePath(dtfile_id, /*throw_on_not_exist=*/false); + fiu_do_on(FailPoints::force_set_dtfile_exist_when_acquire_id, { + static std::atomic fail_point_called(0); + if (existed_path.empty() && fail_point_called.load() % 10 == 0) + { + existed_path = ""; + } + fail_point_called++; + }); + if (likely(existed_path.empty())) + { + break; + } + // else there is a DTFile with that id, continue to acquire a new ID. + LOG_WARNING( + logger, + "The DTFile is already exists, continute to acquire another ID. call={} path={} file_id={}", + who, + existed_path, + dtfile_id); + } while (true); + return dtfile_id; +} + +PageIdU64 StoragePool::newLogPageId() const +{ + return global_id_allocator->newLogPageId(); +} + +PageIdU64 StoragePool::newMetaPageId() const +{ + return global_id_allocator->newMetaPageId(); +} + +template +inline static PageReaderPtr newReader( + const PageStorageRunMode run_mode, + KeyspaceID keyspace_id, + StorageType tag, + const NamespaceID table_id, + T & storage_v2, + T & storage_v3, + UniversalPageStoragePtr uni_ps, + ReadLimiterPtr read_limiter, + bool snapshot_read, + const String & tracing_id) +{ + switch (run_mode) + { + case PageStorageRunMode::ONLY_V2: + return std::make_shared( + run_mode, + keyspace_id, + tag, + table_id, + storage_v2, + nullptr, + /*uni_ps*/ nullptr, + snapshot_read ? storage_v2->getSnapshot(tracing_id) : nullptr, + read_limiter); + case PageStorageRunMode::ONLY_V3: + return std::make_shared( + run_mode, + keyspace_id, + tag, + table_id, + nullptr, + storage_v3, + /*uni_ps*/ nullptr, + snapshot_read ? storage_v3->getSnapshot(tracing_id) : nullptr, + read_limiter); + case PageStorageRunMode::MIX_MODE: + return std::make_shared( + run_mode, + keyspace_id, + tag, + table_id, + storage_v2, + storage_v3, + /*uni_ps*/ nullptr, + snapshot_read ? std::make_shared( + storage_v2->getSnapshot(fmt::format("{}-v2", tracing_id)), + storage_v3->getSnapshot(fmt::format("{}-v3", tracing_id))) + : nullptr, + read_limiter); + case PageStorageRunMode::UNI_PS: + return std::make_shared( + run_mode, + keyspace_id, + tag, + table_id, + nullptr, + nullptr, + uni_ps, + snapshot_read ? uni_ps->getSnapshot(tracing_id) : nullptr, + read_limiter); + default: + throw Exception(ErrorCodes::LOGICAL_ERROR, "Unknown PageStorageRunMode {}", static_cast(run_mode)); + } +} + +PageReaderPtr StoragePool::newLogReader(ReadLimiterPtr read_limiter, bool snapshot_read, const String & tracing_id) +{ + return newReader( + run_mode, + keyspace_id, + StorageType::Log, + table_id, + log_storage_v2, + log_storage_v3, + uni_ps, + read_limiter, + snapshot_read, + tracing_id); +} + +PageReaderPtr StoragePool::newLogReader(ReadLimiterPtr read_limiter, PageStorage::SnapshotPtr & snapshot) +{ + return std::make_shared( + run_mode, + keyspace_id, + StorageType::Log, + table_id, + log_storage_v2, + log_storage_v3, + uni_ps, + snapshot, + read_limiter); +} + +PageReaderPtr StoragePool::newDataReader(ReadLimiterPtr read_limiter, bool snapshot_read, const String & tracing_id) +{ + return newReader( + run_mode, + keyspace_id, + StorageType::Data, + table_id, + data_storage_v2, + data_storage_v3, + uni_ps, + read_limiter, + snapshot_read, + tracing_id); +} + +PageReaderPtr StoragePool::newMetaReader(ReadLimiterPtr read_limiter, bool snapshot_read, const String & tracing_id) +{ + return newReader( + run_mode, + keyspace_id, + StorageType::Meta, + table_id, + meta_storage_v2, + meta_storage_v3, + uni_ps, + read_limiter, + snapshot_read, + tracing_id); +} + +} // namespace DB::DM diff --git a/dbms/src/Storages/Page/V3/tests/gtest_page_storage_mix_mode.cpp b/dbms/src/Storages/Page/V3/tests/gtest_page_storage_mix_mode.cpp index ee02f56ecd3..043d2c4c4cb 100644 --- a/dbms/src/Storages/Page/V3/tests/gtest_page_storage_mix_mode.cpp +++ b/dbms/src/Storages/Page/V3/tests/gtest_page_storage_mix_mode.cpp @@ -13,7 +13,12 @@ // limitations under the License. #include +<<<<<<< HEAD #include +======= +#include +#include +>>>>>>> 961f9dded3 (Storage: Fix page_id being mis-reuse when upgrade from cluster < 6.5 (#9041)) #include #include #include @@ -24,6 +29,11 @@ #include #include +namespace DB::FailPoints +{ +extern const char force_set_dtfile_exist_when_acquire_id[]; +} // namespace DB::FailPoints + namespace DB { using namespace tests; @@ -688,6 +698,52 @@ try } CATCH +TEST_F(PageStorageMixedTest, GetMaxIdAfterUpgraded) +try +{ + const size_t buf_sz = 1024; + char c_buff[buf_sz] = {0}; + + const PageIdU64 max_data_page_id_allocated = (1 << 30) + 1; + const PageIdU64 max_meta_page_id_allocated = (1 << 28) + 1; + { + // Prepare a StoragePool with + // - 0 pages in "log" (must be 0) + // - some pages in "data" with `max_data_page_id_allocated` + // - some pages in "meta" with `max_meta_page_id_allocated` + { + WriteBatch batch; + ReadBufferPtr buff = std::make_shared(c_buff, sizeof(c_buff)); + batch.putPage(1, 0, buff, buf_sz); + batch.putRefPage(2, 1); + ReadBufferPtr buff2 = std::make_shared(c_buff, sizeof(c_buff)); + batch.putPage(1 << 30, 0, buff2, buf_sz); + batch.putRefPage(max_data_page_id_allocated, 1 << 30); + storage_pool_v2->dataWriter()->write(std::move(batch), nullptr); + } + { + WriteBatch batch; + ReadBufferPtr buff = std::make_shared(c_buff, sizeof(c_buff)); + batch.putPage(max_meta_page_id_allocated, 0, buff, buf_sz); + storage_pool_v2->metaWriter()->write(std::move(batch), nullptr); + } + } + + // Mock that after upgraded, the run_mode is transformed to `ONLY_V3` + ASSERT_EQ(reloadMixedStoragePool(), PageStorageRunMode::ONLY_V3); + + // Disable some failpoint to avoid it affect the allocated id + DB::FailPointHelper::disableFailPoint(DB::FailPoints::force_set_dtfile_exist_when_acquire_id); + SCOPE_EXIT({ DB::FailPointHelper::enableFailPoint(DB::FailPoints::force_set_dtfile_exist_when_acquire_id); }); + + // Allocate new id for "data" should be larger than `max_data_page_id_allocated` + auto d = storage_path_pool_v2->getStableDiskDelegator(); + EXPECT_GT(storage_pool_mix->newDataPageIdForDTFile(d, "GetMaxIdAfterUpgraded"), max_data_page_id_allocated); + + // Allocate new id for "meta" should be larger than `max_meta_page_id_allocated` + EXPECT_GT(storage_pool_mix->newMetaPageId(), max_meta_page_id_allocated); +} +CATCH TEST_F(PageStorageMixedTest, ReuseV2ID) try @@ -696,40 +752,47 @@ try char c_buff[buf_sz] = {0}; { - WriteBatch batch; - ReadBufferPtr buff = std::make_shared(c_buff, sizeof(c_buff)); - batch.putPage(1, 0, buff, buf_sz); - page_writer_v2->write(std::move(batch), nullptr); + // prepare "put 1" && "del 1" + { + WriteBatch batch; + ReadBufferPtr buff = std::make_shared(c_buff, sizeof(c_buff)); + batch.putPage(1, 0, buff, buf_sz); + page_writer_v2->write(std::move(batch), nullptr); + } + { + WriteBatch batch; + batch.delPage(1); + page_writer_v2->write(std::move(batch), nullptr); + } } - { - WriteBatch batch; - batch.delPage(1); - page_writer_v2->write(std::move(batch), nullptr); - } - - { - ASSERT_EQ(reloadMixedStoragePool(), PageStorageRunMode::ONLY_V3); - ASSERT_EQ(storage_pool_mix->newLogPageId(), 1); - } + // All pages in v2 are removed, it run on V3 after restart + ASSERT_EQ(reloadMixedStoragePool(), PageStorageRunMode::ONLY_V3); + ASSERT_EQ(storage_pool_mix->newLogPageId(), 2); // new allocated page id won't reuse page 1 + // Mock that some new pages are written { WriteBatch batch; ReadBufferPtr buff = std::make_shared(c_buff, sizeof(c_buff)); - batch.putPage(1, 0, buff, buf_sz); + batch.putPage(2, 0, buff, buf_sz); page_writer_mix->write(std::move(batch), nullptr); } - { WriteBatch batch; - batch.delPage(1); + batch.delPage(2); page_writer_mix->write(std::move(batch), nullptr); } +<<<<<<< HEAD { ASSERT_EQ(reloadMixedStoragePool(), PageStorageRunMode::ONLY_V3); ASSERT_EQ(storage_pool_mix->newLogPageId(), 2); } +======= + // Mock restart + ASSERT_EQ(reloadMixedStoragePool(), PageStorageRunMode::ONLY_V3); + ASSERT_EQ(storage_pool_mix->newLogPageId(), 3); +>>>>>>> 961f9dded3 (Storage: Fix page_id being mis-reuse when upgrade from cluster < 6.5 (#9041)) } CATCH