Fix potential data inconsistency under heavy ddl operation (pingcap#5044

) (pingcap#5050)

close pingcap#5032

dbms/src/Common/FailPoint.cpp

@@ -49,7 +49,8 @@ std::unordered_map<String, std::shared_ptr<FailPointChannel>> FailPointHelper::f
     M(force_legacy_or_checkpoint_page_file_exists)                \
     M(exception_in_creating_set_input_stream)                     \
     M(exception_when_read_from_log)                               \
-    M(exception_mpp_hash_build)
+    M(exception_mpp_hash_build)                                   \
+    M(exception_between_schema_change_in_the_same_diff)
 
 #define APPLY_FOR_FAILPOINTS(M)                              \
     M(force_set_page_file_write_errno)                       \

dbms/src/Debug/dbgFuncSchema.cpp

@@ -16,6 +16,7 @@ namespace DB
 {
 namespace ErrorCodes
 {
+extern const int FAIL_POINT_ERROR;
 extern const int UNKNOWN_TABLE;
 } // namespace ErrorCodes
 
@@ -44,7 +45,22 @@ void dbgFuncRefreshSchemas(Context & context, const ASTs &, DBGInvoker::Printer
 {
     TMTContext & tmt = context.getTMTContext();
     auto schema_syncer = tmt.getSchemaSyncer();
-    schema_syncer->syncSchemas(context);
+    try
+    {
+        schema_syncer->syncSchemas(context);
+    }
+    catch (Exception & e)
+    {
+        if (e.code() == ErrorCodes::FAIL_POINT_ERROR)
+        {
+            output(e.message());
+            return;
+        }
+        else
+        {
+            throw;
+        }
+    }
 
     output("schemas refreshed");
 }
@@ -56,7 +72,7 @@ void dbgFuncGcSchemas(Context & context, const ASTs & args, DBGInvoker::Printer
 {
     auto & service = context.getSchemaSyncService();
     Timestamp gc_safe_point = 0;
-    if (args.size() == 0)
+    if (args.empty())
         gc_safe_point = PDClientHelper::getGCSafePointWithRetry(context.getTMTContext().getPDClient());
     else
         gc_safe_point = safeGet<Timestamp>(typeid_cast<const ASTLiteral &>(*args[0]).value);
@@ -76,7 +92,7 @@ void dbgFuncResetSchemas(Context & context, const ASTs &, DBGInvoker::Printer ou
 
 void dbgFuncIsTombstone(Context & context, const ASTs & args, DBGInvoker::Printer output)
 {
-    if (args.size() < 1 || args.size() > 2)
+    if (args.empty() || args.size() > 2)
         throw Exception("Args not matched, should be: database-name[, table-name]", ErrorCodes::BAD_ARGUMENTS);
 
     const String & database_name = typeid_cast<const ASTIdentifier &>(*args[0]).name;

dbms/src/Storages/IManageableStorage.h

@@ -146,15 +146,18 @@ class IManageableStorage : public IStorage
 
     virtual size_t getRowKeyColumnSize() const { return 1; }
 
-    // when `need_block` is true, it will try return a cached block corresponding to DecodingStorageSchemaSnapshotConstPtr,
-    //     and `releaseDecodingBlock` need to be called when the block is free
-    // when `need_block` is false, it will just return an nullptr
-    virtual std::pair<DB::DecodingStorageSchemaSnapshotConstPtr, BlockUPtr> getSchemaSnapshotAndBlockForDecoding(bool /* need_block */)
+    /// when `need_block` is true, it will try return a cached block corresponding to DecodingStorageSchemaSnapshotConstPtr,
+    ///     and `releaseDecodingBlock` need to be called when the block is free
+    /// when `need_block` is false, it will just return an nullptr
+    /// This method must be called under the protection of table structure lock
+    virtual std::pair<DB::DecodingStorageSchemaSnapshotConstPtr, BlockUPtr> getSchemaSnapshotAndBlockForDecoding(const TableStructureLockHolder & /* table_structure_lock */, bool /* need_block */)
     {
         throw Exception("Method getDecodingSchemaSnapshot is not supported by storage " + getName(), ErrorCodes::NOT_IMPLEMENTED);
     };
 
-    virtual void releaseDecodingBlock(Int64 /* schema_version */, BlockUPtr /* block */)
+    /// The `block_decoding_schema_version` is just an internal version for `DecodingStorageSchemaSnapshot`,
+    /// And it has no relation with the table schema version.
+    virtual void releaseDecodingBlock(Int64 /* block_decoding_schema_version */, BlockUPtr /* block */)
     {
         throw Exception("Method getDecodingSchemaSnapshot is not supported by storage " + getName(), ErrorCodes::NOT_IMPLEMENTED);
     }

dbms/src/Storages/PrimaryKeyNotMatchException.cpp

@@ -15,26 +15,29 @@
 
 namespace DB
 {
-
 String fixCreateStatementWithPriKeyNotMatchException( //
-    Context & context, const String old_definition, const String & table_metadata_path, const PrimaryKeyNotMatchException & ex,
+    Context & context,
+    const String & old_definition,
+    const String & table_metadata_path,
+    const PrimaryKeyNotMatchException & ex,
     Poco::Logger * log)
 {
     LOG_WARNING(
-        log, "Try to fix statement in " + table_metadata_path + ", primary key [" + ex.pri_key + "] -> [" + ex.actual_pri_key + "]");
+        log,
+        "Try to fix statement in " + table_metadata_path + ", primary key [" + ex.pri_key + "] -> [" + ex.actual_pri_key + "]");
     // Try to fix the create statement.
     ParserCreateQuery parser;
     ASTPtr ast
         = parseQuery(parser, old_definition.data(), old_definition.data() + old_definition.size(), "in file " + table_metadata_path, 0);
     ASTCreateQuery & ast_create_query = typeid_cast<ASTCreateQuery &>(*ast);
     auto args = ast_create_query.storage->engine->arguments;
-    if (args->children.size() >= 1)
+    if (!args->children.empty())
     {
         ASTPtr pk_ast = std::make_shared<ASTExpressionList>();
         pk_ast->children.emplace_back(std::make_shared<ASTIdentifier>(ex.actual_pri_key));
         args->children[0] = pk_ast;
     }
-    const String statement = getTableDefinitionFromCreateQuery(ast);
+    String statement = getTableDefinitionFromCreateQuery(ast);
     const String table_metadata_tmp_path = table_metadata_path + ".tmp";
 
     {
@@ -47,9 +50,8 @@ String fixCreateStatementWithPriKeyNotMatchException( //
         EncryptionPath encryption_path
             = use_target_encrypt_info ? EncryptionPath(table_metadata_path, "") : EncryptionPath(table_metadata_tmp_path, "");
         {
-            bool create_new_encryption_info = !use_target_encrypt_info && statement.size();
-            WriteBufferFromFileProvider out(context.getFileProvider(), table_metadata_tmp_path, encryption_path, create_new_encryption_info,
-                nullptr, statement.size(), O_WRONLY | O_CREAT | O_EXCL);
+            bool create_new_encryption_info = !use_target_encrypt_info && !statement.empty();
+            WriteBufferFromFileProvider out(context.getFileProvider(), table_metadata_tmp_path, encryption_path, create_new_encryption_info, nullptr, statement.size(), O_WRONLY | O_CREAT | O_EXCL);
             writeString(statement, out);
             out.next();
             if (context.getSettingsRef().fsync_metadata)
@@ -60,8 +62,7 @@ String fixCreateStatementWithPriKeyNotMatchException( //
         try
         {
             /// rename atomically replaces the old file with the new one.
-            context.getFileProvider()->renameFile(table_metadata_tmp_path, encryption_path, table_metadata_path,
-                EncryptionPath(table_metadata_path, ""), !use_target_encrypt_info);
+            context.getFileProvider()->renameFile(table_metadata_tmp_path, encryption_path, table_metadata_path, EncryptionPath(table_metadata_path, ""), !use_target_encrypt_info);
         }
         catch (...)
         {

dbms/src/Storages/PrimaryKeyNotMatchException.h

@@ -9,22 +9,21 @@ class Logger;
 
 namespace DB
 {
-
 class Context;
 
-struct PrimaryKeyNotMatchException
+struct PrimaryKeyNotMatchException : public std::exception
 {
     // The primary key name in definition
     const String pri_key;
     // The actual primary key name in TiDB::TableInfo
     const String actual_pri_key;
     PrimaryKeyNotMatchException(const String & pri_key_, const String & actual_pri_key_)
-        : pri_key(pri_key_), actual_pri_key(actual_pri_key_)
+        : pri_key(pri_key_)
+        , actual_pri_key(actual_pri_key_)
     {}
 };
 
 // This function will replace the primary key and update statement in `table_metadata_path`. The correct statement will be return.
-String fixCreateStatementWithPriKeyNotMatchException(Context & context, const String old_definition, const String & table_metadata_path,
-    const PrimaryKeyNotMatchException & ex, Poco::Logger * log);
+String fixCreateStatementWithPriKeyNotMatchException(Context & context, const String & old_definition, const String & table_metadata_path, const PrimaryKeyNotMatchException & ex, Poco::Logger * log);
 
 } // namespace DB

dbms/src/Storages/StorageDeltaMerge.cpp

@@ -424,7 +424,7 @@ class DMBlockOutputStream : public IBlockOutputStream
 
 BlockOutputStreamPtr StorageDeltaMerge::write(const ASTPtr & query, const Settings & settings)
 {
-    auto & insert_query = typeid_cast<const ASTInsertQuery &>(*query);
+    const auto & insert_query = typeid_cast<const ASTInsertQuery &>(*query);
     auto decorator = [&](const Block & block) { //
         return this->buildInsertBlock(insert_query.is_import, insert_query.is_delete, block);
     };
@@ -563,7 +563,7 @@ BlockInputStreams StorageDeltaMerge::read(
     // failed to parsed.
     ColumnDefines columns_to_read;
     auto header = store->getHeader();
-    for (auto & n : column_names)
+    for (const auto & n : column_names)
     {
         ColumnDefine col_define;
         if (n == EXTRA_HANDLE_COLUMN_NAME)
@@ -798,7 +798,7 @@ size_t getRows(DM::DeltaMergeStorePtr & store, const Context & context, const DM
 
 DM::RowKeyRange getRange(DM::DeltaMergeStorePtr & store, const Context & context, size_t total_rows, size_t delete_rows)
 {
-    auto start_index = rand() % (total_rows - delete_rows + 1);
+    auto start_index = rand() % (total_rows - delete_rows + 1); // NOLINT(cert-msc50-cpp)
 
     DM::RowKeyRange range = DM::RowKeyRange::newAll(store->isCommonHandle(), store->getRowKeyColumnSize());
     {
@@ -849,14 +849,16 @@ void StorageDeltaMerge::deleteRows(const Context & context, size_t delete_rows)
         LOG_FMT_ERROR(log, "Rows after delete range not match, expected: {}, got: {}", (total_rows - delete_rows), after_delete_rows);
 }
 
-std::pair<DB::DecodingStorageSchemaSnapshotConstPtr, BlockUPtr> StorageDeltaMerge::getSchemaSnapshotAndBlockForDecoding(bool need_block)
+std::pair<DB::DecodingStorageSchemaSnapshotConstPtr, BlockUPtr> StorageDeltaMerge::getSchemaSnapshotAndBlockForDecoding(const TableStructureLockHolder & table_structure_lock, bool need_block)
 {
+    (void)table_structure_lock;
     std::lock_guard lock{decode_schema_mutex};
-    if (!decoding_schema_snapshot || decoding_schema_snapshot->schema_version < tidb_table_info.schema_version)
+    if (!decoding_schema_snapshot || decoding_schema_changed)
     {
         auto & store = getAndMaybeInitStore();
-        decoding_schema_snapshot = std::make_shared<DecodingStorageSchemaSnapshot>(store->getStoreColumns(), tidb_table_info, store->getHandle());
+        decoding_schema_snapshot = std::make_shared<DecodingStorageSchemaSnapshot>(store->getStoreColumns(), tidb_table_info, store->getHandle(), decoding_schema_version++);
         cache_blocks.clear();
+        decoding_schema_changed = false;
     }
 
     if (need_block)
@@ -878,10 +880,10 @@ std::pair<DB::DecodingStorageSchemaSnapshotConstPtr, BlockUPtr> StorageDeltaMerg
     }
 }
 
-void StorageDeltaMerge::releaseDecodingBlock(Int64 schema_version, BlockUPtr block_ptr)
+void StorageDeltaMerge::releaseDecodingBlock(Int64 block_decoding_schema_version, BlockUPtr block_ptr)
 {
     std::lock_guard lock{decode_schema_mutex};
-    if (!decoding_schema_snapshot || schema_version < decoding_schema_snapshot->schema_version)
+    if (!decoding_schema_snapshot || block_decoding_schema_version < decoding_schema_snapshot->decoding_schema_version)
         return;
     if (cache_blocks.size() >= max_cached_blocks_num)
         return;
@@ -931,12 +933,12 @@ static void updateDeltaMergeTableCreateStatement(
     const SortDescription & pk_names,
     const ColumnsDescription & columns,
     const OrderedNameSet & hidden_columns,
-    const OptionTableInfoConstRef table_info,
+    OptionTableInfoConstRef table_info,
     Timestamp tombstone,
     const Context & context);
 
 inline OptionTableInfoConstRef getTableInfoForCreateStatement(
-    const OptionTableInfoConstRef table_info_from_tidb,
+    OptionTableInfoConstRef table_info_from_tidb,
     TiDB::TableInfo & table_info_from_store,
     const ColumnDefines & store_table_columns,
     const OrderedNameSet & hidden_columns)
@@ -1060,6 +1062,7 @@ try
             updateTableColumnInfo();
         }
     }
+    decoding_schema_changed = true;
 
     SortDescription pk_desc = getPrimarySortDescription();
     ColumnDefines store_columns = getStoreColumnDefines();
@@ -1416,7 +1419,7 @@ void StorageDeltaMerge::startup()
     tmt.getStorages().put(std::static_pointer_cast<StorageDeltaMerge>(shared_from_this()));
 }
 
-void StorageDeltaMerge::shutdown()
+void StorageDeltaMerge::shutdownImpl()
 {
     bool v = false;
     if (!shutdown_called.compare_exchange_strong(v, true))
@@ -1427,6 +1430,11 @@ void StorageDeltaMerge::shutdown()
     }
 }
 
+void StorageDeltaMerge::shutdown()
+{
+    shutdownImpl();
+}
+
 void StorageDeltaMerge::removeFromTMTContext()
 {
     // remove this table from TMTContext
@@ -1437,7 +1445,7 @@ void StorageDeltaMerge::removeFromTMTContext()
 
 StorageDeltaMerge::~StorageDeltaMerge()
 {
-    shutdown();
+    shutdownImpl();
 }
 
 DataTypePtr StorageDeltaMerge::getPKTypeImpl() const

dbms/src/Storages/StorageDeltaMerge.h

@@ -87,7 +87,7 @@ class StorageDeltaMerge
     // Apply AlterCommands synced from TiDB should use `alterFromTiDB` instead of `alter(...)`
     void alterFromTiDB(
         const TableLockHolder &,
-        const AlterCommands & commands,
+        const AlterCommands & params,
         const String & database_name,
         const TiDB::TableInfo & table_info,
         const SchemaNameMapper & name_mapper,
@@ -123,9 +123,9 @@ class StorageDeltaMerge
 
     size_t getRowKeyColumnSize() const override { return rowkey_column_size; }
 
-    std::pair<DB::DecodingStorageSchemaSnapshotConstPtr, BlockUPtr> getSchemaSnapshotAndBlockForDecoding(bool /* need_block */) override;
+    std::pair<DB::DecodingStorageSchemaSnapshotConstPtr, BlockUPtr> getSchemaSnapshotAndBlockForDecoding(const TableStructureLockHolder & table_structure_lock, bool /* need_block */) override;
 
-    void releaseDecodingBlock(Int64 schema_version, BlockUPtr block) override;
+    void releaseDecodingBlock(Int64 block_decoding_schema_version, BlockUPtr block) override;
 
     bool initStoreIfDataDirExist() override;
 
@@ -168,6 +168,7 @@ class StorageDeltaMerge
     void updateTableColumnInfo();
     DM::ColumnDefines getStoreColumnDefines() const;
     bool dataDirExist();
+    void shutdownImpl();
 
 #ifndef DBMS_PUBLIC_GTEST
 private:
@@ -206,6 +207,11 @@ class StorageDeltaMerge
 
     mutable std::mutex decode_schema_mutex;
     DecodingStorageSchemaSnapshotPtr decoding_schema_snapshot;
+    // The following two members must be used under the protection of table structure lock
+    bool decoding_schema_changed = false;
+    // internal version for `decoding_schema_snapshot`
+    Int64 decoding_schema_version = 1;
+
     // avoid creating block every time when decoding row
     std::vector<BlockUPtr> cache_blocks;
     // avoid creating too many cached blocks(the typical num should be less and equal than raft apply thread)

dbms/src/Storages/Transaction/DecodingStorageSchemaSnapshot.h

@@ -52,13 +52,14 @@ struct DecodingStorageSchemaSnapshot
     bool pk_is_handle;
     bool is_common_handle;
     TMTPKType pk_type = TMTPKType::UNSPECIFIED;
-    Int64 schema_version = DEFAULT_UNSPECIFIED_SCHEMA_VERSION;
+    // an internal increasing version for `DecodingStorageSchemaSnapshot`, has no relation with the table schema version
+    Int64 decoding_schema_version;
 
-    DecodingStorageSchemaSnapshot(DM::ColumnDefinesPtr column_defines_, const TiDB::TableInfo & table_info_, const DM::ColumnDefine & original_handle_)
+    DecodingStorageSchemaSnapshot(DM::ColumnDefinesPtr column_defines_, const TiDB::TableInfo & table_info_, const DM::ColumnDefine & original_handle_, Int64 decoding_schema_version_)
         : column_defines{std::move(column_defines_)}
         , pk_is_handle{table_info_.pk_is_handle}
         , is_common_handle{table_info_.is_common_handle}
-        , schema_version{table_info_.schema_version}
+        , decoding_schema_version{decoding_schema_version_}
     {
         std::unordered_map<ColumnID, size_t> column_lut;
         for (size_t i = 0; i < table_info_.columns.size(); i++)

dbms/src/Storages/Transaction/PartitionStreams.cpp

@@ -100,14 +100,14 @@ static void writeRegionDataToStorage(
         /// Read region data as block.
         Stopwatch watch;
 
-        Int64 block_schema_version = DEFAULT_UNSPECIFIED_SCHEMA_VERSION;
+        Int64 block_decoding_schema_version = -1;
         BlockUPtr block_ptr = nullptr;
         if (need_decode)
         {
             LOG_FMT_TRACE(log, "{} begin to decode table {}, region {}", FUNCTION_NAME, table_id, region->id());
             DecodingStorageSchemaSnapshotConstPtr decoding_schema_snapshot;
-            std::tie(decoding_schema_snapshot, block_ptr) = storage->getSchemaSnapshotAndBlockForDecoding(true);
-            block_schema_version = decoding_schema_snapshot->schema_version;
+            std::tie(decoding_schema_snapshot, block_ptr) = storage->getSchemaSnapshotAndBlockForDecoding(lock, true);
+            block_decoding_schema_version = decoding_schema_snapshot->decoding_schema_version;
 
             auto reader = RegionBlockReader(decoding_schema_snapshot);
             if (!reader.read(*block_ptr, data_list_read, force_decode))
@@ -139,7 +139,7 @@ static void writeRegionDataToStorage(
         write_part_cost = watch.elapsedMilliseconds();
         GET_METRIC(tiflash_raft_write_data_to_storage_duration_seconds, type_write).Observe(write_part_cost / 1000.0);
         if (need_decode)
-            storage->releaseDecodingBlock(block_schema_version, std::move(block_ptr));
+            storage->releaseDecodingBlock(block_decoding_schema_version, std::move(block_ptr));
 
         LOG_TRACE(log,
                   FUNCTION_NAME << ": table " << table_id << ", region " << region->id() << ", cost [region decode " << region_decode_cost
@@ -496,7 +496,7 @@ RegionPtrWithBlock::CachePtr GenRegionPreDecodeBlockData(const RegionPtr & regio
         }
 
         DecodingStorageSchemaSnapshotConstPtr decoding_schema_snapshot;
-        std::tie(decoding_schema_snapshot, std::ignore) = storage->getSchemaSnapshotAndBlockForDecoding(false);
+        std::tie(decoding_schema_snapshot, std::ignore) = storage->getSchemaSnapshotAndBlockForDecoding(lock, false);
         res_block = createBlockSortByColumnID(decoding_schema_snapshot);
         auto reader = RegionBlockReader(decoding_schema_snapshot);
         if (!reader.read(res_block, *data_list_read, force_decode))
@@ -548,7 +548,7 @@ AtomicGetStorageSchema(const RegionPtr & region, TMTContext & tmt)
         auto table_lock = storage->lockStructureForShare(getThreadName());
         dm_storage = std::dynamic_pointer_cast<StorageDeltaMerge>(storage);
         // only dt storage engine support `getSchemaSnapshotAndBlockForDecoding`, other engine will throw exception
-        std::tie(schema_snapshot, std::ignore) = storage->getSchemaSnapshotAndBlockForDecoding(false);
+        std::tie(schema_snapshot, std::ignore) = storage->getSchemaSnapshotAndBlockForDecoding(table_lock, false);
         std::tie(std::ignore, drop_lock) = std::move(table_lock).release();
         return true;
     };