/* Copyright (c) 2012, Monty Program Ab This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; version 2 of the License. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02111-1301 USA */ #ifdef USE_PRAGMA_IMPLEMENTATION #pragma implementation // gcc: Class implementation #endif #define MYSQL_SERVER 1 /* For use of 'PRIu64': */ #define __STDC_FORMAT_MACROS #include #include /* The C++ file's header */ #include "./ha_rocksdb.h" /* C++ standard header files */ #include #include #include #include #include #include #include #include /* MySQL includes */ #include "./debug_sync.h" #include "./my_bit.h" #include "./my_stacktrace.h" #include "./my_sys.h" #include "./sql_audit.h" #include "./sql_table.h" #include "./sql_hset.h" #include #ifdef MARIAROCKS_NOT_YET #include #endif #include // Both MySQL and RocksDB define the same constant. To avoid compilation errors // till we make the fix in RocksDB, we'll temporary undefine it here. #undef CACHE_LINE_SIZE /* RocksDB includes */ #include "monitoring/histogram.h" #include "rocksdb/compaction_filter.h" #include "rocksdb/env.h" #include "rocksdb/persistent_cache.h" #include "rocksdb/rate_limiter.h" #include "rocksdb/slice_transform.h" #include "rocksdb/thread_status.h" #include "rocksdb/utilities/checkpoint.h" #include "rocksdb/utilities/convenience.h" #include "rocksdb/utilities/memory_util.h" #include "rocksdb/utilities/sim_cache.h" #include "util/stop_watch.h" #include "./rdb_source_revision.h" /* MyRocks includes */ #include "./event_listener.h" #include "./ha_rocksdb_proto.h" #include "./logger.h" #include "./rdb_cf_manager.h" #include "./rdb_cf_options.h" #include "./rdb_datadic.h" #include "./rdb_i_s.h" #include "./rdb_index_merge.h" #include "./rdb_mutex_wrapper.h" #include "./rdb_psi.h" #include "./rdb_threads.h" #include "./rdb_mariadb_server_port.h" // Internal MySQL APIs not exposed in any header. extern "C" { /** Mark transaction to rollback and mark error as fatal to a sub-statement. @param thd Thread handle @param all TRUE <=> rollback main transaction. */ void thd_mark_transaction_to_rollback(MYSQL_THD thd, bool all); /** * Get the user thread's binary logging format * @param thd user thread * @return Value to be used as index into the binlog_format_names array */ int thd_binlog_format(const MYSQL_THD thd); /** * Check if binary logging is filtered for thread's current db. * @param thd Thread handle * @retval 1 the query is not filtered, 0 otherwise. */ bool thd_binlog_filter_ok(const MYSQL_THD thd); } MYSQL_PLUGIN_IMPORT bool my_disable_leak_check; // Needed in rocksdb_init_func void ignore_db_dirs_append(const char *dirname_arg); namespace myrocks { static st_global_stats global_stats; static st_export_stats export_stats; static st_memory_stats memory_stats; static st_io_stall_stats io_stall_stats; const std::string DEFAULT_CF_NAME("default"); const std::string DEFAULT_SYSTEM_CF_NAME("__system__"); const std::string PER_INDEX_CF_NAME("$per_index_cf"); class Rdb_explicit_snapshot; std::mutex explicit_snapshot_mutex; ulonglong explicit_snapshot_counter = 0; std::unordered_map> explicit_snapshots; static std::vector rdb_indexes_to_recalc; #ifdef MARIADB_NOT_YET class Rdb_explicit_snapshot : public explicit_snapshot { std::unique_ptr snapshot; public: static std::shared_ptr create(snapshot_info_st *ss_info, rocksdb::DB *db, const rocksdb::Snapshot *snapshot) { std::lock_guard lock(explicit_snapshot_mutex); auto s = std::unique_ptr( new rocksdb::ManagedSnapshot(db, snapshot)); if (!s) { return nullptr; } ss_info->snapshot_id = ++explicit_snapshot_counter; auto ret = std::make_shared(*ss_info, std::move(s)); if (!ret) { return nullptr; } explicit_snapshots[ss_info->snapshot_id] = ret; return ret; } static std::shared_ptr get(const ulonglong snapshot_id) { std::lock_guard lock(explicit_snapshot_mutex); auto elem = explicit_snapshots.find(snapshot_id); if (elem == explicit_snapshots.end()) { return nullptr; } return elem->second.lock(); } rocksdb::ManagedSnapshot *get_snapshot() { return snapshot.get(); } Rdb_explicit_snapshot(snapshot_info_st ss_info, std::unique_ptr snapshot) : explicit_snapshot(ss_info), snapshot(std::move(snapshot)) {} virtual ~Rdb_explicit_snapshot() { std::lock_guard lock(explicit_snapshot_mutex); explicit_snapshots.erase(ss_info.snapshot_id); } }; #endif /** Updates row counters based on the table type and operation type. */ void ha_rocksdb::update_row_stats(const operation_type &type) { DBUG_ASSERT(type < ROWS_MAX); // Find if we are modifying system databases. if (table->s && m_tbl_def->m_is_mysql_system_table) global_stats.system_rows[type].inc(); else global_stats.rows[type].inc(); } void dbug_dump_database(rocksdb::DB *db); static handler *rocksdb_create_handler(my_core::handlerton *hton, my_core::TABLE_SHARE *table_arg, my_core::MEM_ROOT *mem_root); static rocksdb::CompactRangeOptions getCompactRangeOptions(int concurrency = 0) { rocksdb::CompactRangeOptions compact_range_options; compact_range_options.bottommost_level_compaction = rocksdb::BottommostLevelCompaction::kForce; compact_range_options.exclusive_manual_compaction = false; if (concurrency > 0) { compact_range_options.max_subcompactions = concurrency; } return compact_range_options; } /////////////////////////////////////////////////////////// // Parameters and settings /////////////////////////////////////////////////////////// static char *rocksdb_default_cf_options = nullptr; static char *rocksdb_override_cf_options = nullptr; static char *rocksdb_update_cf_options = nullptr; /////////////////////////////////////////////////////////// // Globals /////////////////////////////////////////////////////////// handlerton *rocksdb_hton; rocksdb::TransactionDB *rdb = nullptr; rocksdb::HistogramImpl *commit_latency_stats = nullptr; static std::shared_ptr rocksdb_stats; static std::unique_ptr flashcache_aware_env; static std::shared_ptr properties_collector_factory; Rdb_dict_manager dict_manager; Rdb_cf_manager cf_manager; Rdb_ddl_manager ddl_manager; Rdb_binlog_manager binlog_manager; #if !defined(_WIN32) && !defined(__APPLE__) Rdb_io_watchdog *io_watchdog = nullptr; #endif /** MyRocks background thread control N.B. This is besides RocksDB's own background threads (@see rocksdb::CancelAllBackgroundWork()) */ static Rdb_background_thread rdb_bg_thread; static Rdb_manual_compaction_thread rdb_mc_thread; // List of table names (using regex) that are exceptions to the strict // collation check requirement. Regex_list_handler *rdb_collation_exceptions; static const char **rdb_get_error_messages(int nr); static void rocksdb_flush_all_memtables() { const Rdb_cf_manager &cf_manager = rdb_get_cf_manager(); for (const auto &cf_handle : cf_manager.get_all_cf()) { rdb->Flush(rocksdb::FlushOptions(), cf_handle); } } /////////////////////////////////////////////////////////// // Hash map: table name => open table handler /////////////////////////////////////////////////////////// namespace // anonymous namespace = not visible outside this source file { const ulong TABLE_HASH_SIZE = 32; typedef Hash_set Rdb_table_set; struct Rdb_open_tables_map { /* Hash table used to track the handlers of open tables */ Rdb_table_set m_hash; /* The mutex used to protect the hash table */ mutable mysql_mutex_t m_mutex; static uchar *get_hash_key(const Rdb_table_handler *const table_handler, size_t *const length, my_bool not_used MY_ATTRIBUTE((__unused__))); Rdb_table_handler *get_table_handler(const char *const table_name); void release_table_handler(Rdb_table_handler *const table_handler); Rdb_open_tables_map() : m_hash(get_hash_key, system_charset_info) { } void free_hash(void) { m_hash.~Rdb_table_set(); } std::vector get_table_names(void) const; }; } // anonymous namespace static Rdb_open_tables_map rdb_open_tables; static std::string rdb_normalize_dir(std::string dir) { while (dir.size() > 0 && dir.back() == '/') { dir.resize(dir.size() - 1); } return dir; } static int rocksdb_create_checkpoint( THD *const thd MY_ATTRIBUTE((__unused__)), struct st_mysql_sys_var *const var MY_ATTRIBUTE((__unused__)), void *const save MY_ATTRIBUTE((__unused__)), struct st_mysql_value *const value) { char buf[FN_REFLEN]; int len = sizeof(buf); const char *const checkpoint_dir_raw = value->val_str(value, buf, &len); if (checkpoint_dir_raw) { if (rdb != nullptr) { std::string checkpoint_dir = rdb_normalize_dir(checkpoint_dir_raw); // NO_LINT_DEBUG sql_print_information("RocksDB: creating checkpoint in directory : %s\n", checkpoint_dir.c_str()); rocksdb::Checkpoint *checkpoint; auto status = rocksdb::Checkpoint::Create(rdb, &checkpoint); // We can only return HA_EXIT_FAILURE/HA_EXIT_SUCCESS here which is why // the return code is ignored, but by calling into rdb_error_to_mysql, // it will call my_error for us, which will propogate up to the client. int rc __attribute__((__unused__)); if (status.ok()) { status = checkpoint->CreateCheckpoint(checkpoint_dir.c_str()); delete checkpoint; if (status.ok()) { sql_print_information( "RocksDB: created checkpoint in directory : %s\n", checkpoint_dir.c_str()); return HA_EXIT_SUCCESS; } else { rc = ha_rocksdb::rdb_error_to_mysql(status); } } else { rc = ha_rocksdb::rdb_error_to_mysql(status); } } } return HA_EXIT_FAILURE; } /* This method is needed to indicate that the ROCKSDB_CREATE_CHECKPOINT command is not read-only */ static void rocksdb_create_checkpoint_stub(THD *const thd, struct st_mysql_sys_var *const var, void *const var_ptr, const void *const save) {} static void rocksdb_force_flush_memtable_now_stub( THD *const thd, struct st_mysql_sys_var *const var, void *const var_ptr, const void *const save) {} static int rocksdb_force_flush_memtable_now( THD *const thd, struct st_mysql_sys_var *const var, void *const var_ptr, struct st_mysql_value *const value) { sql_print_information("RocksDB: Manual memtable flush."); rocksdb_flush_all_memtables(); return HA_EXIT_SUCCESS; } static void rocksdb_force_flush_memtable_and_lzero_now_stub( THD *const thd, struct st_mysql_sys_var *const var, void *const var_ptr, const void *const save) {} static int rocksdb_force_flush_memtable_and_lzero_now( THD *const thd, struct st_mysql_sys_var *const var, void *const var_ptr, struct st_mysql_value *const value) { sql_print_information("RocksDB: Manual memtable and L0 flush."); rocksdb_flush_all_memtables(); const Rdb_cf_manager &cf_manager = rdb_get_cf_manager(); rocksdb::CompactionOptions c_options = rocksdb::CompactionOptions(); rocksdb::ColumnFamilyMetaData metadata; rocksdb::ColumnFamilyDescriptor cf_descr; for (const auto &cf_handle : cf_manager.get_all_cf()) { rdb->GetColumnFamilyMetaData(cf_handle, &metadata); cf_handle->GetDescriptor(&cf_descr); c_options.output_file_size_limit = cf_descr.options.target_file_size_base; DBUG_ASSERT(metadata.levels[0].level == 0); std::vector file_names; for (auto &file : metadata.levels[0].files) { file_names.emplace_back(file.db_path + file.name); } if (!file_names.empty()) { rocksdb::Status s; s = rdb->CompactFiles(c_options, cf_handle, file_names, 1); if (!s.ok() && !s.IsAborted()) { rdb_handle_io_error(s, RDB_IO_ERROR_GENERAL); return HA_EXIT_FAILURE; } } } return HA_EXIT_SUCCESS; } static void rocksdb_drop_index_wakeup_thread( my_core::THD *const thd MY_ATTRIBUTE((__unused__)), struct st_mysql_sys_var *const var MY_ATTRIBUTE((__unused__)), void *const var_ptr MY_ATTRIBUTE((__unused__)), const void *const save); static my_bool rocksdb_pause_background_work = 0; static mysql_mutex_t rdb_sysvars_mutex; static mysql_mutex_t rdb_block_cache_resize_mutex; static void rocksdb_set_pause_background_work( my_core::THD *const, struct st_mysql_sys_var *const, void *const, const void *const save) { RDB_MUTEX_LOCK_CHECK(rdb_sysvars_mutex); const my_bool pause_requested = *static_cast(save); if (rocksdb_pause_background_work != pause_requested) { if (pause_requested) { rdb->PauseBackgroundWork(); } else { rdb->ContinueBackgroundWork(); } rocksdb_pause_background_work = pause_requested; } RDB_MUTEX_UNLOCK_CHECK(rdb_sysvars_mutex); } static void rocksdb_set_compaction_options(THD *thd, struct st_mysql_sys_var *var, void *var_ptr, const void *save); static void rocksdb_set_table_stats_sampling_pct(THD *thd, struct st_mysql_sys_var *var, void *var_ptr, const void *save); static void rocksdb_set_rate_limiter_bytes_per_sec(THD *thd, struct st_mysql_sys_var *var, void *var_ptr, const void *save); static void rocksdb_set_sst_mgr_rate_bytes_per_sec(THD *thd, struct st_mysql_sys_var *var, void *var_ptr, const void *save); static void rocksdb_set_delayed_write_rate(THD *thd, struct st_mysql_sys_var *var, void *var_ptr, const void *save); static void rocksdb_set_max_latest_deadlocks(THD *thd, struct st_mysql_sys_var *var, void *var_ptr, const void *save); static void rdb_set_collation_exception_list(const char *exception_list); static void rocksdb_set_collation_exception_list(THD *thd, struct st_mysql_sys_var *var, void *var_ptr, const void *save); static int rocksdb_validate_update_cf_options(THD *thd, struct st_mysql_sys_var *var, void *save, st_mysql_value *value); static void rocksdb_set_update_cf_options(THD *thd, struct st_mysql_sys_var *var, void *var_ptr, const void *save); static int rocksdb_check_bulk_load(THD *const thd, struct st_mysql_sys_var *var MY_ATTRIBUTE((__unused__)), void *save, struct st_mysql_value *value); static int rocksdb_check_bulk_load_allow_unsorted( THD *const thd, struct st_mysql_sys_var *var MY_ATTRIBUTE((__unused__)), void *save, struct st_mysql_value *value); static void rocksdb_set_max_background_jobs(THD *thd, struct st_mysql_sys_var *const var, void *const var_ptr, const void *const save); static void rocksdb_set_bytes_per_sync(THD *thd, struct st_mysql_sys_var *const var, void *const var_ptr, const void *const save); static void rocksdb_set_wal_bytes_per_sync(THD *thd, struct st_mysql_sys_var *const var, void *const var_ptr, const void *const save); static int rocksdb_validate_set_block_cache_size( THD *thd, struct st_mysql_sys_var *const var, void *var_ptr, struct st_mysql_value *value); ////////////////////////////////////////////////////////////////////////////// // Options definitions ////////////////////////////////////////////////////////////////////////////// static long long rocksdb_block_cache_size; static long long rocksdb_sim_cache_size; static my_bool rocksdb_use_clock_cache; /* Use unsigned long long instead of uint64_t because of MySQL compatibility */ static unsigned long long // NOLINT(runtime/int) rocksdb_rate_limiter_bytes_per_sec; static unsigned long long // NOLINT(runtime/int) rocksdb_sst_mgr_rate_bytes_per_sec; static unsigned long long rocksdb_delayed_write_rate; static uint32_t rocksdb_max_latest_deadlocks; static unsigned long // NOLINT(runtime/int) rocksdb_persistent_cache_size_mb; static ulong rocksdb_info_log_level; static char *rocksdb_wal_dir; static char *rocksdb_persistent_cache_path; static ulong rocksdb_index_type; static uint32_t rocksdb_flush_log_at_trx_commit; static uint32_t rocksdb_debug_optimizer_n_rows; static my_bool rocksdb_force_compute_memtable_stats; static uint32_t rocksdb_force_compute_memtable_stats_cachetime; static my_bool rocksdb_debug_optimizer_no_zero_cardinality; static uint32_t rocksdb_wal_recovery_mode; static uint32_t rocksdb_access_hint_on_compaction_start; static char *rocksdb_compact_cf_name; static char *rocksdb_checkpoint_name; static my_bool rocksdb_signal_drop_index_thread; static my_bool rocksdb_signal_remove_mariabackup_checkpoint; static my_bool rocksdb_strict_collation_check = 1; static my_bool rocksdb_ignore_unknown_options = 1; static my_bool rocksdb_enable_2pc = 0; static char *rocksdb_strict_collation_exceptions; static my_bool rocksdb_collect_sst_properties = 1; static my_bool rocksdb_force_flush_memtable_now_var = 0; static my_bool rocksdb_force_flush_memtable_and_lzero_now_var = 0; static my_bool rocksdb_enable_ttl = 1; static my_bool rocksdb_enable_ttl_read_filtering = 1; static int rocksdb_debug_ttl_rec_ts = 0; static int rocksdb_debug_ttl_snapshot_ts = 0; static int rocksdb_debug_ttl_read_filter_ts = 0; static my_bool rocksdb_debug_ttl_ignore_pk = 0; static my_bool rocksdb_reset_stats = 0; static uint32_t rocksdb_io_write_timeout_secs = 0; static uint32_t rocksdb_seconds_between_stat_computes = 3600; static long long rocksdb_compaction_sequential_deletes = 0l; static long long rocksdb_compaction_sequential_deletes_window = 0l; static long long rocksdb_compaction_sequential_deletes_file_size = 0l; static uint32_t rocksdb_validate_tables = 1; static char *rocksdb_datadir; static uint32_t rocksdb_table_stats_sampling_pct; static my_bool rocksdb_enable_bulk_load_api = 1; static my_bool rocksdb_print_snapshot_conflict_queries = 0; static my_bool rocksdb_large_prefix = 0; static my_bool rocksdb_allow_to_start_after_corruption = 0; static char* rocksdb_git_hash; char *compression_types_val= const_cast(get_rocksdb_supported_compression_types()); static unsigned long rocksdb_write_policy = rocksdb::TxnDBWritePolicy::WRITE_COMMITTED; static my_bool rocksdb_error_on_suboptimal_collation = 1; static uint32_t rocksdb_stats_recalc_rate = 0; static uint32_t rocksdb_debug_manual_compaction_delay = 0; static uint32_t rocksdb_max_manual_compactions = 0; std::atomic rocksdb_row_lock_deadlocks(0); std::atomic rocksdb_row_lock_wait_timeouts(0); std::atomic rocksdb_snapshot_conflict_errors(0); std::atomic rocksdb_wal_group_syncs(0); std::atomic rocksdb_manual_compactions_processed(0); std::atomic rocksdb_manual_compactions_running(0); /* Remove directory with files in it. Used to remove checkpoint created by mariabackup. */ #ifdef _WIN32 #include /* unlink*/ #ifndef F_OK #define F_OK 0 #endif #endif static int rmdir_force(const char *dir) { if (access(dir, F_OK)) return true; char path[FN_REFLEN]; char sep[] = {FN_LIBCHAR, 0}; int err = 0; MY_DIR *dir_info = my_dir(dir, MYF(MY_DONT_SORT | MY_WANT_STAT)); if (!dir_info) return 1; for (uint i = 0; i < dir_info->number_of_files; i++) { FILEINFO *file = dir_info->dir_entry + i; strxnmov(path, sizeof(path), dir, sep, file->name, NULL); err = my_delete(path, 0); if (err) { break; } } my_dirend(dir_info); if (!err) err = rmdir(dir); return (err == 0) ? HA_EXIT_SUCCESS : HA_EXIT_FAILURE; } static void rocksdb_remove_mariabackup_checkpoint( my_core::THD *const, struct st_mysql_sys_var *const , void *const var_ptr, const void *const) { std::string mariabackup_checkpoint_dir(rocksdb_datadir); mariabackup_checkpoint_dir.append("/mariabackup-checkpoint"); if (unlink(mariabackup_checkpoint_dir.c_str()) == 0) return; rmdir_force(mariabackup_checkpoint_dir.c_str()); } static std::unique_ptr rdb_init_rocksdb_db_options(void) { auto o = std::unique_ptr(new rocksdb::DBOptions()); o->create_if_missing = true; o->listeners.push_back(std::make_shared(&ddl_manager)); o->info_log_level = rocksdb::InfoLogLevel::INFO_LEVEL; o->max_subcompactions = DEFAULT_SUBCOMPACTIONS; o->max_open_files = -2; // auto-tune to 50% open_files_limit o->two_write_queues = true; o->manual_wal_flush = true; return o; } /* DBOptions contains Statistics and needs to be destructed last */ static std::unique_ptr rocksdb_tbl_options = std::unique_ptr( new rocksdb::BlockBasedTableOptions()); static std::unique_ptr rocksdb_db_options = rdb_init_rocksdb_db_options(); static std::shared_ptr rocksdb_rate_limiter; /* This enum needs to be kept up to date with rocksdb::TxnDBWritePolicy */ static const char *write_policy_names[] = {"write_committed", "write_prepared", "write_unprepared", NullS}; static TYPELIB write_policy_typelib = {array_elements(write_policy_names) - 1, "write_policy_typelib", write_policy_names, nullptr}; /* This enum needs to be kept up to date with rocksdb::InfoLogLevel */ static const char *info_log_level_names[] = {"debug_level", "info_level", "warn_level", "error_level", "fatal_level", NullS}; static TYPELIB info_log_level_typelib = { array_elements(info_log_level_names) - 1, "info_log_level_typelib", info_log_level_names, nullptr}; static void rocksdb_set_rocksdb_info_log_level( THD *const thd, struct st_mysql_sys_var *const var, void *const var_ptr, const void *const save) { DBUG_ASSERT(save != nullptr); RDB_MUTEX_LOCK_CHECK(rdb_sysvars_mutex); rocksdb_info_log_level = *static_cast(save); rocksdb_db_options->info_log->SetInfoLogLevel( static_cast(rocksdb_info_log_level)); RDB_MUTEX_UNLOCK_CHECK(rdb_sysvars_mutex); } static void rocksdb_set_reset_stats( my_core::THD *const /* unused */, my_core::st_mysql_sys_var *const var MY_ATTRIBUTE((__unused__)), void *const var_ptr, const void *const save) { DBUG_ASSERT(save != nullptr); DBUG_ASSERT(rdb != nullptr); DBUG_ASSERT(rocksdb_stats != nullptr); RDB_MUTEX_LOCK_CHECK(rdb_sysvars_mutex); *static_cast(var_ptr) = *static_cast(save); if (rocksdb_reset_stats) { rocksdb::Status s = rdb->ResetStats(); // RocksDB will always return success. Let's document this assumption here // as well so that we'll get immediately notified when contract changes. DBUG_ASSERT(s == rocksdb::Status::OK()); s = rocksdb_stats->Reset(); DBUG_ASSERT(s == rocksdb::Status::OK()); } RDB_MUTEX_UNLOCK_CHECK(rdb_sysvars_mutex); } static void rocksdb_set_io_write_timeout( my_core::THD *const thd MY_ATTRIBUTE((__unused__)), my_core::st_mysql_sys_var *const var MY_ATTRIBUTE((__unused__)), void *const var_ptr MY_ATTRIBUTE((__unused__)), const void *const save) { DBUG_ASSERT(save != nullptr); DBUG_ASSERT(rdb != nullptr); #if !defined(_WIN32) && !defined(__APPLE__) DBUG_ASSERT(io_watchdog != nullptr); #endif RDB_MUTEX_LOCK_CHECK(rdb_sysvars_mutex); const uint32_t new_val = *static_cast(save); rocksdb_io_write_timeout_secs = new_val; #if !defined(_WIN32) && !defined(__APPLE__) io_watchdog->reset_timeout(rocksdb_io_write_timeout_secs); #endif RDB_MUTEX_UNLOCK_CHECK(rdb_sysvars_mutex); } enum rocksdb_flush_log_at_trx_commit_type : unsigned int { FLUSH_LOG_NEVER = 0, FLUSH_LOG_SYNC, FLUSH_LOG_BACKGROUND, FLUSH_LOG_MAX /* must be last */ }; static int rocksdb_validate_flush_log_at_trx_commit( THD *const thd, struct st_mysql_sys_var *const var, /* in: pointer to system variable */ void *var_ptr, /* out: immediate result for update function */ struct st_mysql_value *const value /* in: incoming value */) { long long new_value; /* value is NULL */ if (value->val_int(value, &new_value)) { return HA_EXIT_FAILURE; } if (rocksdb_db_options->allow_mmap_writes && new_value != FLUSH_LOG_NEVER) { return HA_EXIT_FAILURE; } *static_cast(var_ptr) = static_cast(new_value); return HA_EXIT_SUCCESS; } static void rocksdb_compact_column_family_stub( THD *const thd, struct st_mysql_sys_var *const var, void *const var_ptr, const void *const save) {} static int rocksdb_compact_column_family(THD *const thd, struct st_mysql_sys_var *const var, void *const var_ptr, struct st_mysql_value *const value); static const char *index_type_names[] = {"kBinarySearch", "kHashSearch", NullS}; static TYPELIB index_type_typelib = {array_elements(index_type_names) - 1, "index_type_typelib", index_type_names, nullptr}; const ulong RDB_MAX_LOCK_WAIT_SECONDS = 1024 * 1024 * 1024; const ulong RDB_DEFAULT_MAX_ROW_LOCKS = 1024 * 1024; const ulong RDB_MAX_ROW_LOCKS = 1024 * 1024 * 1024; const ulong RDB_DEFAULT_BULK_LOAD_SIZE = 1000; const ulong RDB_MAX_BULK_LOAD_SIZE = 1024 * 1024 * 1024; const size_t RDB_DEFAULT_MERGE_BUF_SIZE = 64 * 1024 * 1024; const size_t RDB_MIN_MERGE_BUF_SIZE = 100; const size_t RDB_DEFAULT_MERGE_COMBINE_READ_SIZE = 1024 * 1024 * 1024; const size_t RDB_MIN_MERGE_COMBINE_READ_SIZE = 100; const size_t RDB_DEFAULT_MERGE_TMP_FILE_REMOVAL_DELAY = 0; const size_t RDB_MIN_MERGE_TMP_FILE_REMOVAL_DELAY = 0; const int64 RDB_DEFAULT_BLOCK_CACHE_SIZE = 512 * 1024 * 1024; const int64 RDB_MIN_BLOCK_CACHE_SIZE = 1024; const int RDB_MAX_CHECKSUMS_PCT = 100; const ulong RDB_DEADLOCK_DETECT_DEPTH = 50; // TODO: 0 means don't wait at all, and we don't support it yet? static MYSQL_THDVAR_ULONG(lock_wait_timeout, PLUGIN_VAR_RQCMDARG, "Number of seconds to wait for lock", nullptr, nullptr, /*default*/ 1, /*min*/ 1, /*max*/ RDB_MAX_LOCK_WAIT_SECONDS, 0); static MYSQL_THDVAR_BOOL(deadlock_detect, PLUGIN_VAR_RQCMDARG, "Enables deadlock detection", nullptr, nullptr, FALSE); static MYSQL_THDVAR_ULONG(deadlock_detect_depth, PLUGIN_VAR_RQCMDARG, "Number of transactions deadlock detection will " "traverse through before assuming deadlock", nullptr, nullptr, /*default*/ RDB_DEADLOCK_DETECT_DEPTH, /*min*/ 2, /*max*/ ULONG_MAX, 0); static MYSQL_THDVAR_BOOL( commit_time_batch_for_recovery, PLUGIN_VAR_RQCMDARG, "TransactionOptions::commit_time_batch_for_recovery for RocksDB", nullptr, nullptr, FALSE); static MYSQL_THDVAR_BOOL( trace_sst_api, PLUGIN_VAR_RQCMDARG, "Generate trace output in the log for each call to the SstFileWriter", nullptr, nullptr, FALSE); static MYSQL_THDVAR_BOOL( bulk_load, PLUGIN_VAR_RQCMDARG, "Use bulk-load mode for inserts. This disables " "unique_checks and enables rocksdb_commit_in_the_middle.", rocksdb_check_bulk_load, nullptr, FALSE); static MYSQL_THDVAR_BOOL(bulk_load_allow_sk, PLUGIN_VAR_RQCMDARG, "Allow bulk loading of sk keys during bulk-load. " "Can be changed only when bulk load is disabled.", /* Intentionally reuse unsorted's check function */ rocksdb_check_bulk_load_allow_unsorted, nullptr, FALSE); static MYSQL_THDVAR_BOOL(bulk_load_allow_unsorted, PLUGIN_VAR_RQCMDARG, "Allow unsorted input during bulk-load. " "Can be changed only when bulk load is disabled.", rocksdb_check_bulk_load_allow_unsorted, nullptr, FALSE); static MYSQL_SYSVAR_BOOL(enable_bulk_load_api, rocksdb_enable_bulk_load_api, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "Enables using SstFileWriter for bulk loading", nullptr, nullptr, rocksdb_enable_bulk_load_api); static MYSQL_SYSVAR_STR(git_hash, rocksdb_git_hash, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "Git revision of the RocksDB library used by MyRocks", nullptr, nullptr, ROCKSDB_GIT_HASH); static MYSQL_THDVAR_STR(tmpdir, PLUGIN_VAR_OPCMDARG | PLUGIN_VAR_MEMALLOC, "Directory for temporary files during DDL operations.", nullptr, nullptr, ""); static MYSQL_THDVAR_STR( skip_unique_check_tables, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_MEMALLOC, "Skip unique constraint checking for the specified tables", nullptr, nullptr, ".*"); static MYSQL_THDVAR_BOOL( commit_in_the_middle, PLUGIN_VAR_RQCMDARG, "Commit rows implicitly every rocksdb_bulk_load_size, on bulk load/insert, " "update and delete", nullptr, nullptr, FALSE); static MYSQL_THDVAR_BOOL( blind_delete_primary_key, PLUGIN_VAR_RQCMDARG, "Deleting rows by primary key lookup, without reading rows (Blind Deletes)." " Blind delete is disabled if the table has secondary key", nullptr, nullptr, FALSE); static MYSQL_THDVAR_STR( read_free_rpl_tables, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_MEMALLOC, "List of tables that will use read-free replication on the slave " "(i.e. not lookup a row during replication)", nullptr, nullptr, ""); static MYSQL_THDVAR_BOOL(skip_bloom_filter_on_read, PLUGIN_VAR_RQCMDARG, "Skip using bloom filter for reads", nullptr, nullptr, FALSE); static MYSQL_THDVAR_ULONG(max_row_locks, PLUGIN_VAR_RQCMDARG, "Maximum number of locks a transaction can have", nullptr, nullptr, /*default*/ RDB_DEFAULT_MAX_ROW_LOCKS, /*min*/ 1, /*max*/ RDB_MAX_ROW_LOCKS, 0); static MYSQL_THDVAR_ULONGLONG( write_batch_max_bytes, PLUGIN_VAR_RQCMDARG, "Maximum size of write batch in bytes. 0 means no limit.", nullptr, nullptr, /* default */ 0, /* min */ 0, /* max */ SIZE_T_MAX, 1); static MYSQL_THDVAR_BOOL( lock_scanned_rows, PLUGIN_VAR_RQCMDARG, "Take and hold locks on rows that are scanned but not updated", nullptr, nullptr, FALSE); static MYSQL_THDVAR_ULONG(bulk_load_size, PLUGIN_VAR_RQCMDARG, "Max #records in a batch for bulk-load mode", nullptr, nullptr, /*default*/ RDB_DEFAULT_BULK_LOAD_SIZE, /*min*/ 1, /*max*/ RDB_MAX_BULK_LOAD_SIZE, 0); static MYSQL_THDVAR_ULONGLONG( merge_buf_size, PLUGIN_VAR_RQCMDARG, "Size to allocate for merge sort buffers written out to disk " "during inplace index creation.", nullptr, nullptr, /* default (64MB) */ RDB_DEFAULT_MERGE_BUF_SIZE, /* min (100B) */ RDB_MIN_MERGE_BUF_SIZE, /* max */ SIZE_T_MAX, 1); static MYSQL_THDVAR_ULONGLONG( merge_combine_read_size, PLUGIN_VAR_RQCMDARG, "Size that we have to work with during combine (reading from disk) phase " "of " "external sort during fast index creation.", nullptr, nullptr, /* default (1GB) */ RDB_DEFAULT_MERGE_COMBINE_READ_SIZE, /* min (100B) */ RDB_MIN_MERGE_COMBINE_READ_SIZE, /* max */ SIZE_T_MAX, 1); static MYSQL_THDVAR_ULONGLONG( merge_tmp_file_removal_delay_ms, PLUGIN_VAR_RQCMDARG, "Fast index creation creates a large tmp file on disk during index " "creation. Removing this large file all at once when index creation is " "complete can cause trim stalls on Flash. This variable specifies a " "duration to sleep (in milliseconds) between calling chsize() to truncate " "the file in chunks. The chunk size is the same as merge_buf_size.", nullptr, nullptr, /* default (0ms) */ RDB_DEFAULT_MERGE_TMP_FILE_REMOVAL_DELAY, /* min (0ms) */ RDB_MIN_MERGE_TMP_FILE_REMOVAL_DELAY, /* max */ SIZE_T_MAX, 1); static MYSQL_THDVAR_INT( manual_compaction_threads, PLUGIN_VAR_RQCMDARG, "How many rocksdb threads to run for manual compactions", nullptr, nullptr, /* default rocksdb.dboption max_subcompactions */ 0, /* min */ 0, /* max */ 128, 0); static MYSQL_SYSVAR_BOOL( create_if_missing, *reinterpret_cast(&rocksdb_db_options->create_if_missing), PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::create_if_missing for RocksDB", nullptr, nullptr, rocksdb_db_options->create_if_missing); static MYSQL_SYSVAR_BOOL( two_write_queues, *reinterpret_cast(&rocksdb_db_options->two_write_queues), PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::two_write_queues for RocksDB", nullptr, nullptr, rocksdb_db_options->two_write_queues); static MYSQL_SYSVAR_BOOL( manual_wal_flush, *reinterpret_cast(&rocksdb_db_options->manual_wal_flush), PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::manual_wal_flush for RocksDB", nullptr, nullptr, rocksdb_db_options->manual_wal_flush); static MYSQL_SYSVAR_ENUM(write_policy, rocksdb_write_policy, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::write_policy for RocksDB", nullptr, nullptr, rocksdb::TxnDBWritePolicy::WRITE_COMMITTED, &write_policy_typelib); static MYSQL_SYSVAR_BOOL( create_missing_column_families, *reinterpret_cast( &rocksdb_db_options->create_missing_column_families), PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::create_missing_column_families for RocksDB", nullptr, nullptr, rocksdb_db_options->create_missing_column_families); static MYSQL_SYSVAR_BOOL( error_if_exists, *reinterpret_cast(&rocksdb_db_options->error_if_exists), PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::error_if_exists for RocksDB", nullptr, nullptr, rocksdb_db_options->error_if_exists); static MYSQL_SYSVAR_BOOL( paranoid_checks, *reinterpret_cast(&rocksdb_db_options->paranoid_checks), PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::paranoid_checks for RocksDB", nullptr, nullptr, rocksdb_db_options->paranoid_checks); static MYSQL_SYSVAR_ULONGLONG( rate_limiter_bytes_per_sec, rocksdb_rate_limiter_bytes_per_sec, PLUGIN_VAR_RQCMDARG, "DBOptions::rate_limiter bytes_per_sec for RocksDB", nullptr, rocksdb_set_rate_limiter_bytes_per_sec, /* default */ 0L, /* min */ 0L, /* max */ MAX_RATE_LIMITER_BYTES_PER_SEC, 0); static MYSQL_SYSVAR_ULONGLONG( sst_mgr_rate_bytes_per_sec, rocksdb_sst_mgr_rate_bytes_per_sec, PLUGIN_VAR_RQCMDARG, "DBOptions::sst_file_manager rate_bytes_per_sec for RocksDB", nullptr, rocksdb_set_sst_mgr_rate_bytes_per_sec, /* default */ DEFAULT_SST_MGR_RATE_BYTES_PER_SEC, /* min */ 0L, /* max */ UINT64_MAX, 0); static MYSQL_SYSVAR_ULONGLONG(delayed_write_rate, rocksdb_delayed_write_rate, PLUGIN_VAR_RQCMDARG, "DBOptions::delayed_write_rate", nullptr, rocksdb_set_delayed_write_rate, rocksdb_db_options->delayed_write_rate, 0, UINT64_MAX, 0); static MYSQL_SYSVAR_UINT(max_latest_deadlocks, rocksdb_max_latest_deadlocks, PLUGIN_VAR_RQCMDARG, "Maximum number of recent " "deadlocks to store", nullptr, rocksdb_set_max_latest_deadlocks, rocksdb::kInitialMaxDeadlocks, 0, UINT32_MAX, 0); static MYSQL_SYSVAR_ENUM( info_log_level, rocksdb_info_log_level, PLUGIN_VAR_RQCMDARG, "Filter level for info logs to be written mysqld error log. " "Valid values include 'debug_level', 'info_level', 'warn_level'" "'error_level' and 'fatal_level'.", nullptr, rocksdb_set_rocksdb_info_log_level, rocksdb::InfoLogLevel::ERROR_LEVEL, &info_log_level_typelib); static MYSQL_THDVAR_INT( perf_context_level, PLUGIN_VAR_RQCMDARG, "Perf Context Level for rocksdb internal timer stat collection", nullptr, nullptr, /* default */ rocksdb::PerfLevel::kUninitialized, /* min */ rocksdb::PerfLevel::kUninitialized, /* max */ rocksdb::PerfLevel::kOutOfBounds - 1, 0); static MYSQL_SYSVAR_UINT( wal_recovery_mode, rocksdb_wal_recovery_mode, PLUGIN_VAR_RQCMDARG, "DBOptions::wal_recovery_mode for RocksDB. Default is kAbsoluteConsistency", nullptr, nullptr, /* default */ (uint)rocksdb::WALRecoveryMode::kAbsoluteConsistency, /* min */ (uint)rocksdb::WALRecoveryMode::kTolerateCorruptedTailRecords, /* max */ (uint)rocksdb::WALRecoveryMode::kSkipAnyCorruptedRecords, 0); static MYSQL_SYSVAR_SIZE_T(compaction_readahead_size, rocksdb_db_options->compaction_readahead_size, PLUGIN_VAR_RQCMDARG, "DBOptions::compaction_readahead_size for RocksDB", nullptr, nullptr, rocksdb_db_options->compaction_readahead_size, /* min */ 0L, /* max */ SIZE_T_MAX, 0); static MYSQL_SYSVAR_BOOL( new_table_reader_for_compaction_inputs, *reinterpret_cast( &rocksdb_db_options->new_table_reader_for_compaction_inputs), PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::new_table_reader_for_compaction_inputs for RocksDB", nullptr, nullptr, rocksdb_db_options->new_table_reader_for_compaction_inputs); static MYSQL_SYSVAR_UINT( access_hint_on_compaction_start, rocksdb_access_hint_on_compaction_start, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::access_hint_on_compaction_start for RocksDB", nullptr, nullptr, /* default */ (uint)rocksdb::Options::AccessHint::NORMAL, /* min */ (uint)rocksdb::Options::AccessHint::NONE, /* max */ (uint)rocksdb::Options::AccessHint::WILLNEED, 0); static MYSQL_SYSVAR_BOOL( allow_concurrent_memtable_write, *reinterpret_cast( &rocksdb_db_options->allow_concurrent_memtable_write), PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::allow_concurrent_memtable_write for RocksDB", nullptr, nullptr, false); static MYSQL_SYSVAR_BOOL( enable_write_thread_adaptive_yield, *reinterpret_cast( &rocksdb_db_options->enable_write_thread_adaptive_yield), PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::enable_write_thread_adaptive_yield for RocksDB", nullptr, nullptr, false); static MYSQL_SYSVAR_INT(max_open_files, rocksdb_db_options->max_open_files, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::max_open_files for RocksDB", nullptr, nullptr, rocksdb_db_options->max_open_files, /* min */ -2, /* max */ INT_MAX, 0); static MYSQL_SYSVAR_UINT64_T(max_total_wal_size, rocksdb_db_options->max_total_wal_size, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::max_total_wal_size for RocksDB", nullptr, nullptr, rocksdb_db_options->max_total_wal_size, /* min */ 0, /* max */ LONGLONG_MAX, 0); static MYSQL_SYSVAR_BOOL( use_fsync, *reinterpret_cast(&rocksdb_db_options->use_fsync), PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::use_fsync for RocksDB", nullptr, nullptr, rocksdb_db_options->use_fsync); static MYSQL_SYSVAR_STR(wal_dir, rocksdb_wal_dir, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::wal_dir for RocksDB", nullptr, nullptr, rocksdb_db_options->wal_dir.c_str()); static MYSQL_SYSVAR_STR( persistent_cache_path, rocksdb_persistent_cache_path, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "Path for BlockBasedTableOptions::persistent_cache for RocksDB", nullptr, nullptr, ""); static MYSQL_SYSVAR_ULONG( persistent_cache_size_mb, rocksdb_persistent_cache_size_mb, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "Size of cache in MB for BlockBasedTableOptions::persistent_cache " "for RocksDB", nullptr, nullptr, rocksdb_persistent_cache_size_mb, /* min */ 0L, /* max */ ULONG_MAX, 0); static MYSQL_SYSVAR_UINT64_T( delete_obsolete_files_period_micros, rocksdb_db_options->delete_obsolete_files_period_micros, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::delete_obsolete_files_period_micros for RocksDB", nullptr, nullptr, rocksdb_db_options->delete_obsolete_files_period_micros, /* min */ 0, /* max */ LONGLONG_MAX, 0); static MYSQL_SYSVAR_INT(max_background_jobs, rocksdb_db_options->max_background_jobs, PLUGIN_VAR_RQCMDARG, "DBOptions::max_background_jobs for RocksDB", nullptr, rocksdb_set_max_background_jobs, rocksdb_db_options->max_background_jobs, /* min */ -1, /* max */ MAX_BACKGROUND_JOBS, 0); static MYSQL_SYSVAR_UINT(max_subcompactions, rocksdb_db_options->max_subcompactions, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::max_subcompactions for RocksDB", nullptr, nullptr, rocksdb_db_options->max_subcompactions, /* min */ 1, /* max */ MAX_SUBCOMPACTIONS, 0); static MYSQL_SYSVAR_SIZE_T(max_log_file_size, rocksdb_db_options->max_log_file_size, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::max_log_file_size for RocksDB", nullptr, nullptr, rocksdb_db_options->max_log_file_size, /* min */ 0L, /* max */ SIZE_T_MAX, 0); static MYSQL_SYSVAR_SIZE_T(log_file_time_to_roll, rocksdb_db_options->log_file_time_to_roll, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::log_file_time_to_roll for RocksDB", nullptr, nullptr, rocksdb_db_options->log_file_time_to_roll, /* min */ 0L, /* max */ SIZE_T_MAX, 0); static MYSQL_SYSVAR_SIZE_T(keep_log_file_num, rocksdb_db_options->keep_log_file_num, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::keep_log_file_num for RocksDB", nullptr, nullptr, rocksdb_db_options->keep_log_file_num, /* min */ 0L, /* max */ SIZE_T_MAX, 0); static MYSQL_SYSVAR_UINT64_T(max_manifest_file_size, rocksdb_db_options->max_manifest_file_size, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::max_manifest_file_size for RocksDB", nullptr, nullptr, rocksdb_db_options->max_manifest_file_size, /* min */ 0L, /* max */ ULONGLONG_MAX, 0); static MYSQL_SYSVAR_INT(table_cache_numshardbits, rocksdb_db_options->table_cache_numshardbits, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::table_cache_numshardbits for RocksDB", nullptr, nullptr, rocksdb_db_options->table_cache_numshardbits, // LRUCache limits this to 19 bits, anything greater // fails to create a cache and returns a nullptr /* min */ 0, /* max */ 19, 0); static MYSQL_SYSVAR_UINT64_T(wal_ttl_seconds, rocksdb_db_options->WAL_ttl_seconds, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::WAL_ttl_seconds for RocksDB", nullptr, nullptr, rocksdb_db_options->WAL_ttl_seconds, /* min */ 0L, /* max */ LONGLONG_MAX, 0); static MYSQL_SYSVAR_UINT64_T(wal_size_limit_mb, rocksdb_db_options->WAL_size_limit_MB, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::WAL_size_limit_MB for RocksDB", nullptr, nullptr, rocksdb_db_options->WAL_size_limit_MB, /* min */ 0L, /* max */ LONGLONG_MAX, 0); static MYSQL_SYSVAR_SIZE_T(manifest_preallocation_size, rocksdb_db_options->manifest_preallocation_size, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::manifest_preallocation_size for RocksDB", nullptr, nullptr, rocksdb_db_options->manifest_preallocation_size, /* min */ 0L, /* max */ SIZE_T_MAX, 0); static MYSQL_SYSVAR_BOOL( use_direct_reads, *reinterpret_cast(&rocksdb_db_options->use_direct_reads), PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::use_direct_reads for RocksDB", nullptr, nullptr, rocksdb_db_options->use_direct_reads); static MYSQL_SYSVAR_BOOL( use_direct_io_for_flush_and_compaction, *reinterpret_cast(&rocksdb_db_options->use_direct_io_for_flush_and_compaction), PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::use_direct_io_for_flush_and_compaction for RocksDB", nullptr, nullptr, rocksdb_db_options->use_direct_io_for_flush_and_compaction); static MYSQL_SYSVAR_BOOL( allow_mmap_reads, *reinterpret_cast(&rocksdb_db_options->allow_mmap_reads), PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::allow_mmap_reads for RocksDB", nullptr, nullptr, rocksdb_db_options->allow_mmap_reads); static MYSQL_SYSVAR_BOOL( allow_mmap_writes, *reinterpret_cast(&rocksdb_db_options->allow_mmap_writes), PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::allow_mmap_writes for RocksDB", nullptr, nullptr, rocksdb_db_options->allow_mmap_writes); static MYSQL_SYSVAR_BOOL( is_fd_close_on_exec, *reinterpret_cast(&rocksdb_db_options->is_fd_close_on_exec), PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::is_fd_close_on_exec for RocksDB", nullptr, nullptr, rocksdb_db_options->is_fd_close_on_exec); static MYSQL_SYSVAR_UINT(stats_dump_period_sec, rocksdb_db_options->stats_dump_period_sec, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::stats_dump_period_sec for RocksDB", nullptr, nullptr, rocksdb_db_options->stats_dump_period_sec, /* min */ 0, /* max */ INT_MAX, 0); static MYSQL_SYSVAR_BOOL( advise_random_on_open, *reinterpret_cast(&rocksdb_db_options->advise_random_on_open), PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::advise_random_on_open for RocksDB", nullptr, nullptr, rocksdb_db_options->advise_random_on_open); static MYSQL_SYSVAR_SIZE_T(db_write_buffer_size, rocksdb_db_options->db_write_buffer_size, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::db_write_buffer_size for RocksDB", nullptr, nullptr, rocksdb_db_options->db_write_buffer_size, /* min */ 0L, /* max */ SIZE_T_MAX, 0); static MYSQL_SYSVAR_BOOL( use_adaptive_mutex, *reinterpret_cast(&rocksdb_db_options->use_adaptive_mutex), PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::use_adaptive_mutex for RocksDB", nullptr, nullptr, rocksdb_db_options->use_adaptive_mutex); static MYSQL_SYSVAR_UINT64_T(bytes_per_sync, rocksdb_db_options->bytes_per_sync, PLUGIN_VAR_RQCMDARG, "DBOptions::bytes_per_sync for RocksDB", nullptr, rocksdb_set_bytes_per_sync, rocksdb_db_options->bytes_per_sync, /* min */ 0L, /* max */ ULONGLONG_MAX, 0); static MYSQL_SYSVAR_UINT64_T(wal_bytes_per_sync, rocksdb_db_options->wal_bytes_per_sync, PLUGIN_VAR_RQCMDARG, "DBOptions::wal_bytes_per_sync for RocksDB", nullptr, rocksdb_set_wal_bytes_per_sync, rocksdb_db_options->wal_bytes_per_sync, /* min */ 0L, /* max */ ULONGLONG_MAX, 0); static MYSQL_SYSVAR_BOOL( enable_thread_tracking, *reinterpret_cast(&rocksdb_db_options->enable_thread_tracking), PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "DBOptions::enable_thread_tracking for RocksDB", nullptr, nullptr, true); static MYSQL_SYSVAR_LONGLONG(block_cache_size, rocksdb_block_cache_size, PLUGIN_VAR_RQCMDARG, "block_cache size for RocksDB", rocksdb_validate_set_block_cache_size, nullptr, /* default */ RDB_DEFAULT_BLOCK_CACHE_SIZE, /* min */ RDB_MIN_BLOCK_CACHE_SIZE, /* max */ LONGLONG_MAX, /* Block size */ RDB_MIN_BLOCK_CACHE_SIZE); static MYSQL_SYSVAR_LONGLONG(sim_cache_size, rocksdb_sim_cache_size, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "Simulated cache size for RocksDB", nullptr, nullptr, /* default */ 0, /* min */ 0, /* max */ LONGLONG_MAX, /* Block size */ 0); static MYSQL_SYSVAR_BOOL( use_clock_cache, rocksdb_use_clock_cache, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "Use ClockCache instead of default LRUCache for RocksDB", nullptr, nullptr, false); static MYSQL_SYSVAR_BOOL( cache_index_and_filter_blocks, *reinterpret_cast( &rocksdb_tbl_options->cache_index_and_filter_blocks), PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "BlockBasedTableOptions::cache_index_and_filter_blocks for RocksDB", nullptr, nullptr, true); // When pin_l0_filter_and_index_blocks_in_cache is true, RocksDB will use the // LRU cache, but will always keep the filter & idndex block's handle checked // out (=won't call ShardedLRUCache::Release), plus the parsed out objects // the LRU cache will never push flush them out, hence they're pinned. // // This fixes the mutex contention between :ShardedLRUCache::Lookup and // ShardedLRUCache::Release which reduced the QPS ratio (QPS using secondary // index / QPS using PK). static MYSQL_SYSVAR_BOOL( pin_l0_filter_and_index_blocks_in_cache, *reinterpret_cast( &rocksdb_tbl_options->pin_l0_filter_and_index_blocks_in_cache), PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "pin_l0_filter_and_index_blocks_in_cache for RocksDB", nullptr, nullptr, true); static MYSQL_SYSVAR_ENUM(index_type, rocksdb_index_type, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "BlockBasedTableOptions::index_type for RocksDB", nullptr, nullptr, (ulong)rocksdb_tbl_options->index_type, &index_type_typelib); static MYSQL_SYSVAR_BOOL( hash_index_allow_collision, *reinterpret_cast( &rocksdb_tbl_options->hash_index_allow_collision), PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "BlockBasedTableOptions::hash_index_allow_collision for RocksDB", nullptr, nullptr, rocksdb_tbl_options->hash_index_allow_collision); static MYSQL_SYSVAR_BOOL( no_block_cache, *reinterpret_cast(&rocksdb_tbl_options->no_block_cache), PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "BlockBasedTableOptions::no_block_cache for RocksDB", nullptr, nullptr, rocksdb_tbl_options->no_block_cache); static MYSQL_SYSVAR_SIZE_T(block_size, rocksdb_tbl_options->block_size, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "BlockBasedTableOptions::block_size for RocksDB", nullptr, nullptr, rocksdb_tbl_options->block_size, /* min */ 1L, /* max */ SIZE_T_MAX, 0); static MYSQL_SYSVAR_INT( block_size_deviation, rocksdb_tbl_options->block_size_deviation, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "BlockBasedTableOptions::block_size_deviation for RocksDB", nullptr, nullptr, rocksdb_tbl_options->block_size_deviation, /* min */ 0, /* max */ INT_MAX, 0); static MYSQL_SYSVAR_INT( block_restart_interval, rocksdb_tbl_options->block_restart_interval, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "BlockBasedTableOptions::block_restart_interval for RocksDB", nullptr, nullptr, rocksdb_tbl_options->block_restart_interval, /* min */ 1, /* max */ INT_MAX, 0); static MYSQL_SYSVAR_BOOL( whole_key_filtering, *reinterpret_cast(&rocksdb_tbl_options->whole_key_filtering), PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "BlockBasedTableOptions::whole_key_filtering for RocksDB", nullptr, nullptr, rocksdb_tbl_options->whole_key_filtering); static MYSQL_SYSVAR_STR(default_cf_options, rocksdb_default_cf_options, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "default cf options for RocksDB", nullptr, nullptr, ""); static MYSQL_SYSVAR_STR(override_cf_options, rocksdb_override_cf_options, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "option overrides per cf for RocksDB", nullptr, nullptr, ""); static MYSQL_SYSVAR_STR(update_cf_options, rocksdb_update_cf_options, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_MEMALLOC /* psergey-merge: need this? : PLUGIN_VAR_ALLOCATED*/, "Option updates per column family for RocksDB", rocksdb_validate_update_cf_options, rocksdb_set_update_cf_options, nullptr); static MYSQL_SYSVAR_UINT(flush_log_at_trx_commit, rocksdb_flush_log_at_trx_commit, PLUGIN_VAR_RQCMDARG, "Sync on transaction commit. Similar to " "innodb_flush_log_at_trx_commit. 1: sync on commit, " "0,2: not sync on commit", rocksdb_validate_flush_log_at_trx_commit, nullptr, /* default */ FLUSH_LOG_SYNC, /* min */ FLUSH_LOG_NEVER, /* max */ FLUSH_LOG_BACKGROUND, 0); static MYSQL_THDVAR_BOOL(write_disable_wal, PLUGIN_VAR_RQCMDARG, "WriteOptions::disableWAL for RocksDB", nullptr, nullptr, rocksdb::WriteOptions().disableWAL); static MYSQL_THDVAR_BOOL( write_ignore_missing_column_families, PLUGIN_VAR_RQCMDARG, "WriteOptions::ignore_missing_column_families for RocksDB", nullptr, nullptr, rocksdb::WriteOptions().ignore_missing_column_families); static MYSQL_THDVAR_BOOL(skip_fill_cache, PLUGIN_VAR_RQCMDARG, "Skip filling block cache on read requests", nullptr, nullptr, FALSE); static MYSQL_THDVAR_BOOL( unsafe_for_binlog, PLUGIN_VAR_RQCMDARG, "Allowing statement based binary logging which may break consistency", nullptr, nullptr, FALSE); static MYSQL_THDVAR_UINT(records_in_range, PLUGIN_VAR_RQCMDARG, "Used to override the result of records_in_range(). " "Set to a positive number to override", nullptr, nullptr, 0, /* min */ 0, /* max */ INT_MAX, 0); static MYSQL_THDVAR_UINT(force_index_records_in_range, PLUGIN_VAR_RQCMDARG, "Used to override the result of records_in_range() " "when FORCE INDEX is used.", nullptr, nullptr, 0, /* min */ 0, /* max */ INT_MAX, 0); static MYSQL_SYSVAR_UINT( debug_optimizer_n_rows, rocksdb_debug_optimizer_n_rows, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY | PLUGIN_VAR_NOSYSVAR, "Test only to override rocksdb estimates of table size in a memtable", nullptr, nullptr, 0, /* min */ 0, /* max */ INT_MAX, 0); static MYSQL_SYSVAR_BOOL(force_compute_memtable_stats, rocksdb_force_compute_memtable_stats, PLUGIN_VAR_RQCMDARG, "Force to always compute memtable stats", nullptr, nullptr, TRUE); static MYSQL_SYSVAR_UINT(force_compute_memtable_stats_cachetime, rocksdb_force_compute_memtable_stats_cachetime, PLUGIN_VAR_RQCMDARG, "Time in usecs to cache memtable estimates", nullptr, nullptr, /* default */ 60 * 1000 * 1000, /* min */ 0, /* max */ INT_MAX, 0); static MYSQL_SYSVAR_BOOL( debug_optimizer_no_zero_cardinality, rocksdb_debug_optimizer_no_zero_cardinality, PLUGIN_VAR_RQCMDARG, "In case if cardinality is zero, overrides it with some value", nullptr, nullptr, TRUE); static MYSQL_SYSVAR_STR(compact_cf, rocksdb_compact_cf_name, PLUGIN_VAR_RQCMDARG, "Compact column family", rocksdb_compact_column_family, rocksdb_compact_column_family_stub, ""); static MYSQL_SYSVAR_STR(create_checkpoint, rocksdb_checkpoint_name, PLUGIN_VAR_RQCMDARG, "Checkpoint directory", rocksdb_create_checkpoint, rocksdb_create_checkpoint_stub, ""); static MYSQL_SYSVAR_BOOL(remove_mariabackup_checkpoint, rocksdb_signal_remove_mariabackup_checkpoint, PLUGIN_VAR_RQCMDARG, "Remove mariabackup checkpoint", nullptr, rocksdb_remove_mariabackup_checkpoint, FALSE); static MYSQL_SYSVAR_BOOL(signal_drop_index_thread, rocksdb_signal_drop_index_thread, PLUGIN_VAR_RQCMDARG, "Wake up drop index thread", nullptr, rocksdb_drop_index_wakeup_thread, FALSE); static MYSQL_SYSVAR_BOOL(pause_background_work, rocksdb_pause_background_work, PLUGIN_VAR_RQCMDARG, "Disable all rocksdb background operations", nullptr, rocksdb_set_pause_background_work, FALSE); static MYSQL_SYSVAR_BOOL( enable_ttl, rocksdb_enable_ttl, PLUGIN_VAR_RQCMDARG, "Enable expired TTL records to be dropped during compaction.", nullptr, nullptr, TRUE); static MYSQL_SYSVAR_BOOL( enable_ttl_read_filtering, rocksdb_enable_ttl_read_filtering, PLUGIN_VAR_RQCMDARG, "For tables with TTL, expired records are skipped/filtered out during " "processing and in query results. Disabling this will allow these records " "to be seen, but as a result rows may disappear in the middle of " "transactions as they are dropped during compaction. Use with caution.", nullptr, nullptr, TRUE); static MYSQL_SYSVAR_INT( debug_ttl_rec_ts, rocksdb_debug_ttl_rec_ts, PLUGIN_VAR_RQCMDARG, "For debugging purposes only. Overrides the TTL of records to " "now() + debug_ttl_rec_ts. The value can be +/- to simulate " "a record inserted in the past vs a record inserted in the 'future'. " "A value of 0 denotes that the variable is not set. This variable is a " "no-op in non-debug builds.", nullptr, nullptr, 0, /* min */ -3600, /* max */ 3600, 0); static MYSQL_SYSVAR_INT( debug_ttl_snapshot_ts, rocksdb_debug_ttl_snapshot_ts, PLUGIN_VAR_RQCMDARG, "For debugging purposes only. Sets the snapshot during compaction to " "now() + debug_set_ttl_snapshot_ts. The value can be +/- to simulate " "a snapshot in the past vs a snapshot created in the 'future'. " "A value of 0 denotes that the variable is not set. This variable is a " "no-op in non-debug builds.", nullptr, nullptr, 0, /* min */ -3600, /* max */ 3600, 0); static MYSQL_SYSVAR_INT( debug_ttl_read_filter_ts, rocksdb_debug_ttl_read_filter_ts, PLUGIN_VAR_RQCMDARG, "For debugging purposes only. Overrides the TTL read filtering time to " "time + debug_ttl_read_filter_ts. A value of 0 denotes that the variable " "is not set. This variable is a no-op in non-debug builds.", nullptr, nullptr, 0, /* min */ -3600, /* max */ 3600, 0); static MYSQL_SYSVAR_BOOL( debug_ttl_ignore_pk, rocksdb_debug_ttl_ignore_pk, PLUGIN_VAR_RQCMDARG, "For debugging purposes only. If true, compaction filtering will not occur " "on PK TTL data. This variable is a no-op in non-debug builds.", nullptr, nullptr, FALSE); static MYSQL_SYSVAR_UINT( max_manual_compactions, rocksdb_max_manual_compactions, PLUGIN_VAR_RQCMDARG, "Maximum number of pending + ongoing number of manual compactions.", nullptr, nullptr, /* default */ 10, /* min */ 0, /* max */ UINT_MAX, 0); static MYSQL_SYSVAR_UINT( debug_manual_compaction_delay, rocksdb_debug_manual_compaction_delay, PLUGIN_VAR_RQCMDARG, "For debugging purposes only. Sleeping specified seconds " "for simulating long running compactions.", nullptr, nullptr, 0, /* min */ 0, /* max */ UINT_MAX, 0); static MYSQL_SYSVAR_BOOL( reset_stats, rocksdb_reset_stats, PLUGIN_VAR_RQCMDARG, "Reset the RocksDB internal statistics without restarting the DB.", nullptr, rocksdb_set_reset_stats, FALSE); static MYSQL_SYSVAR_UINT(io_write_timeout, rocksdb_io_write_timeout_secs, PLUGIN_VAR_RQCMDARG, "Timeout for experimental I/O watchdog.", nullptr, rocksdb_set_io_write_timeout, /* default */ 0, /* min */ 0L, /* max */ UINT_MAX, 0); static MYSQL_SYSVAR_BOOL(enable_2pc, rocksdb_enable_2pc, PLUGIN_VAR_RQCMDARG, "Enable two phase commit for MyRocks", nullptr, nullptr, TRUE); static MYSQL_SYSVAR_BOOL(ignore_unknown_options, rocksdb_ignore_unknown_options, PLUGIN_VAR_OPCMDARG | PLUGIN_VAR_READONLY, "Enable ignoring unknown options passed to RocksDB", nullptr, nullptr, TRUE); static MYSQL_SYSVAR_BOOL(strict_collation_check, rocksdb_strict_collation_check, PLUGIN_VAR_RQCMDARG, "Enforce case sensitive collation for MyRocks indexes", nullptr, nullptr, TRUE); static MYSQL_SYSVAR_STR(strict_collation_exceptions, rocksdb_strict_collation_exceptions, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_MEMALLOC, "List of tables (using regex) that are excluded " "from the case sensitive collation enforcement", nullptr, rocksdb_set_collation_exception_list, ""); static MYSQL_SYSVAR_BOOL(collect_sst_properties, rocksdb_collect_sst_properties, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "Enables collecting SST file properties on each flush", nullptr, nullptr, rocksdb_collect_sst_properties); static MYSQL_SYSVAR_BOOL( force_flush_memtable_now, rocksdb_force_flush_memtable_now_var, PLUGIN_VAR_RQCMDARG, "Forces memstore flush which may block all write requests so be careful", rocksdb_force_flush_memtable_now, rocksdb_force_flush_memtable_now_stub, FALSE); static MYSQL_SYSVAR_BOOL( force_flush_memtable_and_lzero_now, rocksdb_force_flush_memtable_and_lzero_now_var, PLUGIN_VAR_RQCMDARG, "Acts similar to force_flush_memtable_now, but also compacts all L0 files.", rocksdb_force_flush_memtable_and_lzero_now, rocksdb_force_flush_memtable_and_lzero_now_stub, FALSE); static MYSQL_SYSVAR_UINT( seconds_between_stat_computes, rocksdb_seconds_between_stat_computes, PLUGIN_VAR_RQCMDARG, "Sets a number of seconds to wait between optimizer stats recomputation. " "Only changed indexes will be refreshed.", nullptr, nullptr, rocksdb_seconds_between_stat_computes, /* min */ 0L, /* max */ UINT_MAX, 0); static MYSQL_SYSVAR_LONGLONG(compaction_sequential_deletes, rocksdb_compaction_sequential_deletes, PLUGIN_VAR_RQCMDARG, "RocksDB will trigger compaction for the file if " "it has more than this number sequential deletes " "per window", nullptr, rocksdb_set_compaction_options, DEFAULT_COMPACTION_SEQUENTIAL_DELETES, /* min */ 0L, /* max */ MAX_COMPACTION_SEQUENTIAL_DELETES, 0); static MYSQL_SYSVAR_LONGLONG( compaction_sequential_deletes_window, rocksdb_compaction_sequential_deletes_window, PLUGIN_VAR_RQCMDARG, "Size of the window for counting rocksdb_compaction_sequential_deletes", nullptr, rocksdb_set_compaction_options, DEFAULT_COMPACTION_SEQUENTIAL_DELETES_WINDOW, /* min */ 0L, /* max */ MAX_COMPACTION_SEQUENTIAL_DELETES_WINDOW, 0); static MYSQL_SYSVAR_LONGLONG( compaction_sequential_deletes_file_size, rocksdb_compaction_sequential_deletes_file_size, PLUGIN_VAR_RQCMDARG, "Minimum file size required for compaction_sequential_deletes", nullptr, rocksdb_set_compaction_options, 0L, /* min */ -1L, /* max */ LONGLONG_MAX, 0); static MYSQL_SYSVAR_BOOL( compaction_sequential_deletes_count_sd, rocksdb_compaction_sequential_deletes_count_sd, PLUGIN_VAR_RQCMDARG, "Counting SingleDelete as rocksdb_compaction_sequential_deletes", nullptr, nullptr, rocksdb_compaction_sequential_deletes_count_sd); static MYSQL_SYSVAR_BOOL( print_snapshot_conflict_queries, rocksdb_print_snapshot_conflict_queries, PLUGIN_VAR_RQCMDARG, "Logging queries that got snapshot conflict errors into *.err log", nullptr, nullptr, rocksdb_print_snapshot_conflict_queries); static MYSQL_THDVAR_INT(checksums_pct, PLUGIN_VAR_RQCMDARG, "How many percentages of rows to be checksummed", nullptr, nullptr, RDB_MAX_CHECKSUMS_PCT, /* min */ 0, /* max */ RDB_MAX_CHECKSUMS_PCT, 0); static MYSQL_THDVAR_BOOL(store_row_debug_checksums, PLUGIN_VAR_RQCMDARG, "Include checksums when writing index/table records", nullptr, nullptr, false /* default value */); static MYSQL_THDVAR_BOOL(verify_row_debug_checksums, PLUGIN_VAR_RQCMDARG, "Verify checksums when reading index/table records", nullptr, nullptr, false /* default value */); static MYSQL_THDVAR_BOOL(master_skip_tx_api, PLUGIN_VAR_RQCMDARG, "Skipping holding any lock on row access. " "Not effective on slave.", nullptr, nullptr, false); static MYSQL_SYSVAR_UINT( validate_tables, rocksdb_validate_tables, PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY, "Verify all .frm files match all RocksDB tables (0 means no verification, " "1 means verify and fail on error, and 2 means verify but continue", nullptr, nullptr, 1 /* default value */, 0 /* min value */, 2 /* max value */, 0); static MYSQL_SYSVAR_STR(datadir, rocksdb_datadir, PLUGIN_VAR_OPCMDARG | PLUGIN_VAR_READONLY, "RocksDB data directory", nullptr, nullptr, "./#rocksdb"); static MYSQL_SYSVAR_STR(supported_compression_types, compression_types_val, PLUGIN_VAR_NOCMDOPT | PLUGIN_VAR_READONLY, "Compression algorithms supported by RocksDB", nullptr, nullptr, compression_types_val); static MYSQL_SYSVAR_UINT( table_stats_sampling_pct, rocksdb_table_stats_sampling_pct, PLUGIN_VAR_RQCMDARG, "Percentage of entries to sample when collecting statistics about table " "properties. Specify either 0 to sample everything or percentage " "[" STRINGIFY_ARG(RDB_TBL_STATS_SAMPLE_PCT_MIN) ".." STRINGIFY_ARG( RDB_TBL_STATS_SAMPLE_PCT_MAX) "]. " "By default " STRINGIFY_ARG( RDB_DEFAULT_TBL_STATS_SAMPLE_PCT) "% " "of" " e" "nt" "ri" "es" " a" "re" " " "sa" "mp" "le" "d" ".", nullptr, rocksdb_set_table_stats_sampling_pct, /* default */ RDB_DEFAULT_TBL_STATS_SAMPLE_PCT, /* everything */ 0, /* max */ RDB_TBL_STATS_SAMPLE_PCT_MAX, 0); static MYSQL_SYSVAR_UINT( stats_recalc_rate, rocksdb_stats_recalc_rate, PLUGIN_VAR_RQCMDARG, "The number of indexes per second to recalculate statistics for. 0 to " "disable background recalculation.", nullptr, nullptr, 0 /* default value */, 0 /* min value */, UINT_MAX /* max value */, 0); static MYSQL_SYSVAR_BOOL( large_prefix, rocksdb_large_prefix, PLUGIN_VAR_RQCMDARG, "Support large index prefix length of 3072 bytes. If off, the maximum " "index prefix length is 767.", nullptr, nullptr, FALSE); static MYSQL_SYSVAR_BOOL( allow_to_start_after_corruption, rocksdb_allow_to_start_after_corruption, PLUGIN_VAR_OPCMDARG | PLUGIN_VAR_READONLY, "Allow server still to start successfully even if RocksDB corruption is " "detected.", nullptr, nullptr, FALSE); static MYSQL_SYSVAR_BOOL(error_on_suboptimal_collation, rocksdb_error_on_suboptimal_collation, PLUGIN_VAR_OPCMDARG | PLUGIN_VAR_READONLY, "Raise an error instead of warning if a sub-optimal " "collation is used", nullptr, nullptr, TRUE); static const int ROCKSDB_ASSUMED_KEY_VALUE_DISK_SIZE = 100; static struct st_mysql_sys_var *rocksdb_system_variables[] = { MYSQL_SYSVAR(lock_wait_timeout), MYSQL_SYSVAR(deadlock_detect), MYSQL_SYSVAR(deadlock_detect_depth), MYSQL_SYSVAR(commit_time_batch_for_recovery), MYSQL_SYSVAR(max_row_locks), MYSQL_SYSVAR(write_batch_max_bytes), MYSQL_SYSVAR(lock_scanned_rows), MYSQL_SYSVAR(bulk_load), MYSQL_SYSVAR(bulk_load_allow_sk), MYSQL_SYSVAR(bulk_load_allow_unsorted), MYSQL_SYSVAR(skip_unique_check_tables), MYSQL_SYSVAR(trace_sst_api), MYSQL_SYSVAR(commit_in_the_middle), MYSQL_SYSVAR(blind_delete_primary_key), MYSQL_SYSVAR(read_free_rpl_tables), MYSQL_SYSVAR(bulk_load_size), MYSQL_SYSVAR(merge_buf_size), MYSQL_SYSVAR(enable_bulk_load_api), MYSQL_SYSVAR(tmpdir), MYSQL_SYSVAR(merge_combine_read_size), MYSQL_SYSVAR(merge_tmp_file_removal_delay_ms), MYSQL_SYSVAR(skip_bloom_filter_on_read), MYSQL_SYSVAR(create_if_missing), MYSQL_SYSVAR(two_write_queues), MYSQL_SYSVAR(manual_wal_flush), MYSQL_SYSVAR(write_policy), MYSQL_SYSVAR(create_missing_column_families), MYSQL_SYSVAR(error_if_exists), MYSQL_SYSVAR(paranoid_checks), MYSQL_SYSVAR(rate_limiter_bytes_per_sec), MYSQL_SYSVAR(sst_mgr_rate_bytes_per_sec), MYSQL_SYSVAR(delayed_write_rate), MYSQL_SYSVAR(max_latest_deadlocks), MYSQL_SYSVAR(info_log_level), MYSQL_SYSVAR(max_open_files), MYSQL_SYSVAR(max_total_wal_size), MYSQL_SYSVAR(use_fsync), MYSQL_SYSVAR(wal_dir), MYSQL_SYSVAR(persistent_cache_path), MYSQL_SYSVAR(persistent_cache_size_mb), MYSQL_SYSVAR(delete_obsolete_files_period_micros), MYSQL_SYSVAR(max_background_jobs), MYSQL_SYSVAR(max_log_file_size), MYSQL_SYSVAR(max_subcompactions), MYSQL_SYSVAR(log_file_time_to_roll), MYSQL_SYSVAR(keep_log_file_num), MYSQL_SYSVAR(max_manifest_file_size), MYSQL_SYSVAR(table_cache_numshardbits), MYSQL_SYSVAR(wal_ttl_seconds), MYSQL_SYSVAR(wal_size_limit_mb), MYSQL_SYSVAR(manifest_preallocation_size), MYSQL_SYSVAR(use_direct_reads), MYSQL_SYSVAR(use_direct_io_for_flush_and_compaction), MYSQL_SYSVAR(allow_mmap_reads), MYSQL_SYSVAR(allow_mmap_writes), MYSQL_SYSVAR(is_fd_close_on_exec), MYSQL_SYSVAR(stats_dump_period_sec), MYSQL_SYSVAR(advise_random_on_open), MYSQL_SYSVAR(db_write_buffer_size), MYSQL_SYSVAR(use_adaptive_mutex), MYSQL_SYSVAR(bytes_per_sync), MYSQL_SYSVAR(wal_bytes_per_sync), MYSQL_SYSVAR(enable_thread_tracking), MYSQL_SYSVAR(perf_context_level), MYSQL_SYSVAR(wal_recovery_mode), MYSQL_SYSVAR(access_hint_on_compaction_start), MYSQL_SYSVAR(new_table_reader_for_compaction_inputs), MYSQL_SYSVAR(compaction_readahead_size), MYSQL_SYSVAR(allow_concurrent_memtable_write), MYSQL_SYSVAR(enable_write_thread_adaptive_yield), MYSQL_SYSVAR(block_cache_size), MYSQL_SYSVAR(sim_cache_size), MYSQL_SYSVAR(use_clock_cache), MYSQL_SYSVAR(cache_index_and_filter_blocks), MYSQL_SYSVAR(pin_l0_filter_and_index_blocks_in_cache), MYSQL_SYSVAR(index_type), MYSQL_SYSVAR(hash_index_allow_collision), MYSQL_SYSVAR(no_block_cache), MYSQL_SYSVAR(block_size), MYSQL_SYSVAR(block_size_deviation), MYSQL_SYSVAR(block_restart_interval), MYSQL_SYSVAR(whole_key_filtering), MYSQL_SYSVAR(default_cf_options), MYSQL_SYSVAR(override_cf_options), MYSQL_SYSVAR(update_cf_options), MYSQL_SYSVAR(flush_log_at_trx_commit), MYSQL_SYSVAR(write_disable_wal), MYSQL_SYSVAR(write_ignore_missing_column_families), MYSQL_SYSVAR(skip_fill_cache), MYSQL_SYSVAR(unsafe_for_binlog), MYSQL_SYSVAR(records_in_range), MYSQL_SYSVAR(force_index_records_in_range), MYSQL_SYSVAR(debug_optimizer_n_rows), MYSQL_SYSVAR(force_compute_memtable_stats), MYSQL_SYSVAR(force_compute_memtable_stats_cachetime), MYSQL_SYSVAR(debug_optimizer_no_zero_cardinality), MYSQL_SYSVAR(compact_cf), MYSQL_SYSVAR(signal_drop_index_thread), MYSQL_SYSVAR(pause_background_work), MYSQL_SYSVAR(enable_2pc), MYSQL_SYSVAR(ignore_unknown_options), MYSQL_SYSVAR(strict_collation_check), MYSQL_SYSVAR(strict_collation_exceptions), MYSQL_SYSVAR(collect_sst_properties), MYSQL_SYSVAR(force_flush_memtable_now), MYSQL_SYSVAR(force_flush_memtable_and_lzero_now), MYSQL_SYSVAR(enable_ttl), MYSQL_SYSVAR(enable_ttl_read_filtering), MYSQL_SYSVAR(debug_ttl_rec_ts), MYSQL_SYSVAR(debug_ttl_snapshot_ts), MYSQL_SYSVAR(debug_ttl_read_filter_ts), MYSQL_SYSVAR(debug_ttl_ignore_pk), MYSQL_SYSVAR(reset_stats), MYSQL_SYSVAR(io_write_timeout), MYSQL_SYSVAR(seconds_between_stat_computes), MYSQL_SYSVAR(compaction_sequential_deletes), MYSQL_SYSVAR(compaction_sequential_deletes_window), MYSQL_SYSVAR(compaction_sequential_deletes_file_size), MYSQL_SYSVAR(compaction_sequential_deletes_count_sd), MYSQL_SYSVAR(print_snapshot_conflict_queries), MYSQL_SYSVAR(datadir), MYSQL_SYSVAR(supported_compression_types), MYSQL_SYSVAR(create_checkpoint), MYSQL_SYSVAR(remove_mariabackup_checkpoint), MYSQL_SYSVAR(checksums_pct), MYSQL_SYSVAR(store_row_debug_checksums), MYSQL_SYSVAR(verify_row_debug_checksums), MYSQL_SYSVAR(master_skip_tx_api), MYSQL_SYSVAR(validate_tables), MYSQL_SYSVAR(table_stats_sampling_pct), MYSQL_SYSVAR(large_prefix), MYSQL_SYSVAR(allow_to_start_after_corruption), MYSQL_SYSVAR(git_hash), MYSQL_SYSVAR(error_on_suboptimal_collation), MYSQL_SYSVAR(stats_recalc_rate), MYSQL_SYSVAR(debug_manual_compaction_delay), MYSQL_SYSVAR(max_manual_compactions), MYSQL_SYSVAR(manual_compaction_threads), nullptr}; static rocksdb::WriteOptions rdb_get_rocksdb_write_options(my_core::THD *const thd) { rocksdb::WriteOptions opt; opt.sync = (rocksdb_flush_log_at_trx_commit == FLUSH_LOG_SYNC); opt.disableWAL = THDVAR(thd, write_disable_wal); opt.ignore_missing_column_families = THDVAR(thd, write_ignore_missing_column_families); return opt; } static int rocksdb_compact_column_family(THD *const thd, struct st_mysql_sys_var *const var, void *const var_ptr, struct st_mysql_value *const value) { char buff[STRING_BUFFER_USUAL_SIZE]; int len = sizeof(buff); DBUG_ASSERT(value != nullptr); if (const char *const cf = value->val_str(value, buff, &len)) { auto cfh = cf_manager.get_cf(cf); if (cfh != nullptr && rdb != nullptr) { int mc_id = rdb_mc_thread.request_manual_compaction( cfh, nullptr, nullptr, THDVAR(thd, manual_compaction_threads)); if (mc_id == -1) { my_error(ER_INTERNAL_ERROR, MYF(0), "Can't schedule more manual compactions. " "Increase rocksdb_max_manual_compactions or stop issuing " "more manual compactions."); return HA_EXIT_FAILURE; } else if (mc_id < 0) { return HA_EXIT_FAILURE; } // NO_LINT_DEBUG sql_print_information("RocksDB: Manual compaction of column family: %s\n", cf); // Checking thd state every short cycle (100ms). This is for allowing to // exiting this function without waiting for CompactRange to finish. do { my_sleep(100000); } while (!thd->killed && !rdb_mc_thread.is_manual_compaction_finished(mc_id)); if (thd->killed) { // This cancels if requested compaction state is INITED. // TODO(yoshinorim): Cancel running compaction as well once // it is supported in RocksDB. rdb_mc_thread.clear_manual_compaction_request(mc_id, true); } } } return HA_EXIT_SUCCESS; } /////////////////////////////////////////////////////////////////////////////////////////// /** @brief Function we use in the creation of our hash to get key. */ uchar * Rdb_open_tables_map::get_hash_key(const Rdb_table_handler *const table_handler, size_t *const length, my_bool not_used MY_ATTRIBUTE((__unused__))) { *length = table_handler->m_table_name_length; return reinterpret_cast(table_handler->m_table_name); } /* Drop index thread's control */ static Rdb_drop_index_thread rdb_drop_idx_thread; static void rocksdb_drop_index_wakeup_thread( my_core::THD *const thd MY_ATTRIBUTE((__unused__)), struct st_mysql_sys_var *const var MY_ATTRIBUTE((__unused__)), void *const var_ptr MY_ATTRIBUTE((__unused__)), const void *const save) { if (*static_cast(save)) { rdb_drop_idx_thread.signal(); } } static inline uint32_t rocksdb_perf_context_level(THD *const thd) { DBUG_ASSERT(thd != nullptr); const int session_perf_context_level = THDVAR(thd, perf_context_level); if (session_perf_context_level > rocksdb::PerfLevel::kUninitialized) { return session_perf_context_level; } /* Fallback to global thdvar, if session specific one was not set to a valid value. */ const int global_perf_context_level = THDVAR(nullptr, perf_context_level); if (global_perf_context_level > rocksdb::PerfLevel::kUninitialized) { return global_perf_context_level; } return rocksdb::PerfLevel::kDisable; } /* Very short (functor-like) interface to be passed to Rdb_transaction::walk_tx_list() */ interface Rdb_tx_list_walker { virtual ~Rdb_tx_list_walker() {} virtual void process_tran(const Rdb_transaction *const) = 0; }; /* This is a helper class that is passed to RocksDB to get notifications when a snapshot gets created. */ class Rdb_snapshot_notifier : public rocksdb::TransactionNotifier { Rdb_transaction *m_owning_tx; void SnapshotCreated(const rocksdb::Snapshot *snapshot) override; public: Rdb_snapshot_notifier(const Rdb_snapshot_notifier &) = delete; Rdb_snapshot_notifier &operator=(const Rdb_snapshot_notifier &) = delete; explicit Rdb_snapshot_notifier(Rdb_transaction *const owning_tx) : m_owning_tx(owning_tx) {} // If the owning Rdb_transaction gets destructed we need to not reference // it anymore. void detach() { m_owning_tx = nullptr; } }; #ifdef MARIAROCKS_NOT_YET // ER_LOCK_WAIT_TIMEOUT error also has a reason in facebook/mysql-5.6 #endif String timeout_message(const char *command, const char *name1, const char *name2) { String msg; msg.append("Timeout on "); msg.append(command); msg.append(": "); msg.append(name1); if (name2 && name2[0]) { msg.append("."); msg.append(name2); } return msg; } /* This is the base class for transactions when interacting with rocksdb. */ class Rdb_transaction { protected: ulonglong m_write_count = 0; ulonglong m_insert_count = 0; ulonglong m_update_count = 0; ulonglong m_delete_count = 0; ulonglong m_lock_count = 0; std::unordered_map m_auto_incr_map; bool m_is_delayed_snapshot = false; bool m_is_two_phase = false; private: /* Number of write operations this transaction had when we took the last savepoint (the idea is not to take another savepoint if we haven't made any changes) */ ulonglong m_writes_at_last_savepoint; protected: protected: THD *m_thd = nullptr; static std::multiset s_tx_list; static mysql_mutex_t s_tx_list_mutex; Rdb_io_perf *m_tbl_io_perf; bool m_tx_read_only = false; int m_timeout_sec; /* Cached value of @@rocksdb_lock_wait_timeout */ /* Maximum number of locks the transaction can have */ ulonglong m_max_row_locks; bool m_is_tx_failed = false; bool m_rollback_only = false; std::shared_ptr m_notifier; // This should be used only when updating binlog information. virtual rocksdb::WriteBatchBase *get_write_batch() = 0; virtual bool commit_no_binlog() = 0; virtual rocksdb::Iterator * get_iterator(const rocksdb::ReadOptions &options, rocksdb::ColumnFamilyHandle *column_family) = 0; protected: /* The following two are helper functions to be overloaded by child classes. They should provide RocksDB's savepoint semantics. */ virtual void do_set_savepoint() = 0; virtual void do_rollback_to_savepoint() = 0; /* @detail This function takes in the WriteBatch of the transaction to add all the AUTO_INCREMENT merges. It does so by iterating through m_auto_incr_map and then constructing key/value pairs to call merge upon. @param wb */ rocksdb::Status merge_auto_incr_map(rocksdb::WriteBatchBase *const wb) { DBUG_EXECUTE_IF("myrocks_autoinc_upgrade", return rocksdb::Status::OK();); // Iterate through the merge map merging all keys into data dictionary. rocksdb::Status s; for (auto &it : m_auto_incr_map) { s = dict_manager.put_auto_incr_val(wb, it.first, it.second); if (!s.ok()) { return s; } } m_auto_incr_map.clear(); return s; } public: rocksdb::ReadOptions m_read_opts; const char *m_mysql_log_file_name; my_off_t m_mysql_log_offset; #ifdef MARIAROCKS_NOT_YET // TODO: MariaDB probably doesn't need these at all: const char *m_mysql_gtid; const char *m_mysql_max_gtid; #endif String m_detailed_error; int64_t m_snapshot_timestamp = 0; bool m_ddl_transaction; std::shared_ptr m_explicit_snapshot; /* Tracks the number of tables in use through external_lock. This should not be reset during start_tx(). */ int64_t m_n_mysql_tables_in_use = 0; /* MariaDB's group commit: */ bool commit_ordered_done; bool commit_ordered_res; /* for distinction between rdb_transaction_impl and rdb_writebatch_impl when using walk tx list */ virtual bool is_writebatch_trx() const = 0; static void init_mutex() { mysql_mutex_init(key_mutex_tx_list, &s_tx_list_mutex, MY_MUTEX_INIT_FAST); } static void term_mutex() { DBUG_ASSERT(s_tx_list.size() == 0); mysql_mutex_destroy(&s_tx_list_mutex); } static void walk_tx_list(Rdb_tx_list_walker *walker) { DBUG_ASSERT(walker != nullptr); RDB_MUTEX_LOCK_CHECK(s_tx_list_mutex); for (auto it : s_tx_list) walker->process_tran(it); RDB_MUTEX_UNLOCK_CHECK(s_tx_list_mutex); } int set_status_error(THD *const thd, const rocksdb::Status &s, const Rdb_key_def &kd, Rdb_tbl_def *const tbl_def, Rdb_table_handler *const table_handler) { DBUG_ASSERT(!s.ok()); DBUG_ASSERT(tbl_def != nullptr); if (s.IsTimedOut()) { /* SQL layer has weird expectations. If we return an error when doing a read in DELETE IGNORE, it will ignore the error ("because it's an IGNORE command!) but then will fail an assert, because "error code was returned, but no error happened". Do what InnoDB's convert_error_code_to_mysql() does: force a statement rollback before returning HA_ERR_LOCK_WAIT_TIMEOUT: */ my_core::thd_mark_transaction_to_rollback(thd, false /*just statement*/); m_detailed_error.copy(timeout_message( "index", tbl_def->full_tablename().c_str(), kd.get_name().c_str())); table_handler->m_lock_wait_timeout_counter.inc(); rocksdb_row_lock_wait_timeouts++; return HA_ERR_LOCK_WAIT_TIMEOUT; } if (s.IsDeadlock()) { my_core::thd_mark_transaction_to_rollback(thd, true /* whole transaction */); m_detailed_error = String(); table_handler->m_deadlock_counter.inc(); rocksdb_row_lock_deadlocks++; return HA_ERR_LOCK_DEADLOCK; } else if (s.IsBusy()) { rocksdb_snapshot_conflict_errors++; if (rocksdb_print_snapshot_conflict_queries) { char user_host_buff[MAX_USER_HOST_SIZE + 1]; make_user_name(thd, user_host_buff); // NO_LINT_DEBUG sql_print_warning("Got snapshot conflict errors: User: %s " "Query: %s", user_host_buff, thd->query()); } m_detailed_error = String(" (snapshot conflict)", system_charset_info); table_handler->m_deadlock_counter.inc(); return HA_ERR_ROCKSDB_STATUS_BUSY; } if (s.IsIOError() || s.IsCorruption()) { rdb_handle_io_error(s, RDB_IO_ERROR_GENERAL); } return ha_rocksdb::rdb_error_to_mysql(s); } THD *get_thd() const { return m_thd; } /* Used for tracking io_perf counters */ void io_perf_start(Rdb_io_perf *const io_perf) { /* Since perf_context is tracked per thread, it is difficult and expensive to maintain perf_context on a per table basis. Therefore, roll all perf_context data into the first table used in a query. This works well for single table queries and is probably good enough for queries that hit multiple tables. perf_context stats gathering is started when the table lock is acquired or when ha_rocksdb::start_stmt is called in case of LOCK TABLES. They are recorded when the table lock is released, or when commit/rollback is called on the transaction, whichever comes first. Table lock release and commit/rollback can happen in different orders. In the case where the lock is released before commit/rollback is called, an extra step to gather stats during commit/rollback is needed. */ if (m_tbl_io_perf == nullptr && io_perf->start(rocksdb_perf_context_level(m_thd))) { m_tbl_io_perf = io_perf; } } void io_perf_end_and_record(void) { if (m_tbl_io_perf != nullptr) { m_tbl_io_perf->end_and_record(rocksdb_perf_context_level(m_thd)); m_tbl_io_perf = nullptr; } } void io_perf_end_and_record(Rdb_io_perf *const io_perf) { if (m_tbl_io_perf == io_perf) { io_perf_end_and_record(); } } void update_bytes_written(ulonglong bytes_written) { if (m_tbl_io_perf != nullptr) { m_tbl_io_perf->update_bytes_written(rocksdb_perf_context_level(m_thd), bytes_written); } } void set_params(int timeout_sec_arg, int max_row_locks_arg) { m_timeout_sec = timeout_sec_arg; m_max_row_locks = max_row_locks_arg; set_lock_timeout(timeout_sec_arg); } virtual void set_lock_timeout(int timeout_sec_arg) = 0; ulonglong get_write_count() const { return m_write_count; } ulonglong get_insert_count() const { return m_insert_count; } ulonglong get_update_count() const { return m_update_count; } ulonglong get_delete_count() const { return m_delete_count; } void incr_insert_count() { ++m_insert_count; } void incr_update_count() { ++m_update_count; } void incr_delete_count() { ++m_delete_count; } int get_timeout_sec() const { return m_timeout_sec; } ulonglong get_lock_count() const { return m_lock_count; } virtual void set_sync(bool sync) = 0; virtual void release_lock(rocksdb::ColumnFamilyHandle *const column_family, const std::string &rowkey) = 0; virtual bool prepare(const rocksdb::TransactionName &name) = 0; bool commit_or_rollback() { bool res; if (m_is_tx_failed) { rollback(); res = false; } else res = commit(); return res; } bool commit() { if (get_write_count() == 0) { rollback(); return false; } else if (m_rollback_only) { /* Transactions marked as rollback_only are expected to be rolled back at prepare(). But there are some exceptions like below that prepare() is never called and commit() is called instead. 1. Binlog is disabled 2. No modification exists in binlog cache for the transaction (#195) In both cases, rolling back transaction is safe. Nothing is written to binlog. */ my_error(ER_ROLLBACK_ONLY, MYF(0)); rollback(); return true; } else { #ifdef MARIAROCKS_NOT_YET /* Storing binlog position inside MyRocks is needed only for restoring MyRocks from backups. This feature is not supported yet. */ mysql_bin_log_commit_pos(m_thd, &m_mysql_log_offset, &m_mysql_log_file_name); binlog_manager.update(m_mysql_log_file_name, m_mysql_log_offset, get_write_batch()); #endif return commit_no_binlog(); } } virtual void rollback() = 0; void snapshot_created(const rocksdb::Snapshot *const snapshot) { DBUG_ASSERT(snapshot != nullptr); m_read_opts.snapshot = snapshot; rdb->GetEnv()->GetCurrentTime(&m_snapshot_timestamp); m_is_delayed_snapshot = false; } virtual void acquire_snapshot(bool acquire_now) = 0; virtual void release_snapshot() = 0; bool has_snapshot() const { return m_read_opts.snapshot != nullptr; } private: // The Rdb_sst_info structures we are currently loading. In a partitioned // table this can have more than one entry std::vector> m_curr_bulk_load; std::string m_curr_bulk_load_tablename; /* External merge sorts for bulk load: key ID -> merge sort instance */ std::unordered_map m_key_merge; public: int get_key_merge(GL_INDEX_ID kd_gl_id, rocksdb::ColumnFamilyHandle *cf, Rdb_index_merge **key_merge) { int res; auto it = m_key_merge.find(kd_gl_id); if (it == m_key_merge.end()) { m_key_merge.emplace( std::piecewise_construct, std::make_tuple(kd_gl_id), std::make_tuple( get_rocksdb_tmpdir(), THDVAR(get_thd(), merge_buf_size), THDVAR(get_thd(), merge_combine_read_size), THDVAR(get_thd(), merge_tmp_file_removal_delay_ms), cf)); it = m_key_merge.find(kd_gl_id); if ((res = it->second.init()) != 0) { return res; } } *key_merge = &it->second; return HA_EXIT_SUCCESS; } int finish_bulk_load(int print_client_error = true) { int rc = 0, rc2; std::vector>::iterator it; for (it = m_curr_bulk_load.begin(); it != m_curr_bulk_load.end(); it++) { rc2 = (*it)->commit(print_client_error); if (rc2 != 0 && rc == 0) { rc = rc2; } } m_curr_bulk_load.clear(); m_curr_bulk_load_tablename.clear(); DBUG_ASSERT(m_curr_bulk_load.size() == 0); // Flush the index_merge sort buffers if (!m_key_merge.empty()) { rocksdb::Slice merge_key; rocksdb::Slice merge_val; for (auto it = m_key_merge.begin(); it != m_key_merge.end(); it++) { GL_INDEX_ID index_id = it->first; std::shared_ptr keydef = ddl_manager.safe_find(index_id); std::string table_name = ddl_manager.safe_get_table_name(index_id); // Unable to find key definition or table name since the // table could have been dropped. // TODO(herman): there is a race here between dropping the table // and detecting a drop here. If the table is dropped while bulk // loading is finishing, these keys being added here may // be missed by the compaction filter and not be marked for // removal. It is unclear how to lock the sql table from the storage // engine to prevent modifications to it while bulk load is occurring. if (keydef == nullptr || table_name.empty()) { rc2 = HA_ERR_ROCKSDB_BULK_LOAD; break; } const std::string &index_name = keydef->get_name(); Rdb_index_merge &rdb_merge = it->second; // Rdb_sst_info expects a denormalized table name in the form of // "./database/table" std::replace(table_name.begin(), table_name.end(), '.', '/'); table_name = "./" + table_name; Rdb_sst_info sst_info(rdb, table_name, index_name, rdb_merge.get_cf(), *rocksdb_db_options, THDVAR(get_thd(), trace_sst_api)); while ((rc2 = rdb_merge.next(&merge_key, &merge_val)) == 0) { if ((rc2 = sst_info.put(merge_key, merge_val)) != 0) { break; } } // rc2 == -1 => finished ok; rc2 > 0 => error if (rc2 > 0 || (rc2 = sst_info.commit(print_client_error)) != 0) { if (rc == 0) { rc = rc2; } break; } } m_key_merge.clear(); /* Explicitly tell jemalloc to clean up any unused dirty pages at this point. See https://reviews.facebook.net/D63723 for more details. */ purge_all_jemalloc_arenas(); } return rc; } int start_bulk_load(ha_rocksdb *const bulk_load, std::shared_ptr sst_info) { /* If we already have an open bulk load of a table and the name doesn't match the current one, close out the currently running one. This allows multiple bulk loads to occur on a partitioned table, but then closes them all out when we switch to another table. */ DBUG_ASSERT(bulk_load != nullptr); if (!m_curr_bulk_load.empty() && bulk_load->get_table_basename() != m_curr_bulk_load_tablename) { const auto res = finish_bulk_load(); if (res != HA_EXIT_SUCCESS) { m_curr_bulk_load.clear(); m_curr_bulk_load_tablename.clear(); return res; } } /* This used to track ha_rocksdb handler objects, but those can be freed by the table cache while this was referencing them. Instead of tracking ha_rocksdb handler objects, this now tracks the Rdb_sst_info allocated, and both the ha_rocksdb handler and the Rdb_transaction both have shared pointers to them. On transaction complete, it will commit each Rdb_sst_info structure found. If the ha_rocksdb object is freed, etc., it will also commit the Rdb_sst_info. The Rdb_sst_info commit path needs to be idempotent. */ m_curr_bulk_load.push_back(sst_info); m_curr_bulk_load_tablename = bulk_load->get_table_basename(); return HA_EXIT_SUCCESS; } int num_ongoing_bulk_load() const { return m_curr_bulk_load.size(); } const char *get_rocksdb_tmpdir() const { const char *tmp_dir = THDVAR(get_thd(), tmpdir); /* We want to treat an empty string as nullptr, in these cases DDL operations will use the default --tmpdir passed to mysql instead. */ if (tmp_dir != nullptr && *tmp_dir == '\0') { tmp_dir = nullptr; } return (tmp_dir); } /* Flush the data accumulated so far. This assumes we're doing a bulk insert. @detail This should work like transaction commit, except that we don't synchronize with the binlog (there is no API that would allow to have binlog flush the changes accumulated so far and return its current position) @todo Add test coverage for what happens when somebody attempts to do bulk inserts while inside a multi-statement transaction. */ bool flush_batch() { if (get_write_count() == 0) return false; /* Commit the current transaction */ if (commit_no_binlog()) return true; /* Start another one */ start_tx(); return false; } void set_auto_incr(const GL_INDEX_ID &gl_index_id, ulonglong curr_id) { m_auto_incr_map[gl_index_id] = std::max(m_auto_incr_map[gl_index_id], curr_id); } #ifndef NDEBUG ulonglong get_auto_incr(const GL_INDEX_ID &gl_index_id) { if (m_auto_incr_map.count(gl_index_id) > 0) { return m_auto_incr_map[gl_index_id]; } return 0; } #endif virtual rocksdb::Status put(rocksdb::ColumnFamilyHandle *const column_family, const rocksdb::Slice &key, const rocksdb::Slice &value) = 0; virtual rocksdb::Status delete_key(rocksdb::ColumnFamilyHandle *const column_family, const rocksdb::Slice &key) = 0; virtual rocksdb::Status single_delete(rocksdb::ColumnFamilyHandle *const column_family, const rocksdb::Slice &key) = 0; virtual bool has_modifications() const = 0; virtual rocksdb::WriteBatchBase *get_indexed_write_batch() = 0; /* Return a WriteBatch that one can write to. The writes will skip any transaction locking. The writes will NOT be visible to the transaction. */ rocksdb::WriteBatchBase *get_blind_write_batch() { return get_indexed_write_batch()->GetWriteBatch(); } virtual rocksdb::Status get(rocksdb::ColumnFamilyHandle *const column_family, const rocksdb::Slice &key, rocksdb::PinnableSlice *const value) const = 0; virtual rocksdb::Status get_for_update(rocksdb::ColumnFamilyHandle *const column_family, const rocksdb::Slice &key, rocksdb::PinnableSlice *const value, bool exclusive) = 0; rocksdb::Iterator * get_iterator(rocksdb::ColumnFamilyHandle *const column_family, bool skip_bloom_filter, bool fill_cache, const rocksdb::Slice &eq_cond_lower_bound, const rocksdb::Slice &eq_cond_upper_bound, bool read_current = false, bool create_snapshot = true) { // Make sure we are not doing both read_current (which implies we don't // want a snapshot) and create_snapshot which makes sure we create // a snapshot DBUG_ASSERT(column_family != nullptr); DBUG_ASSERT(!read_current || !create_snapshot); if (create_snapshot) acquire_snapshot(true); rocksdb::ReadOptions options = m_read_opts; if (skip_bloom_filter) { options.total_order_seek = true; options.iterate_lower_bound = &eq_cond_lower_bound; options.iterate_upper_bound = &eq_cond_upper_bound; } else { // With this option, Iterator::Valid() returns false if key // is outside of the prefix bloom filter range set at Seek(). // Must not be set to true if not using bloom filter. options.prefix_same_as_start = true; } options.fill_cache = fill_cache; if (read_current) { options.snapshot = nullptr; } return get_iterator(options, column_family); } virtual bool is_tx_started() const = 0; virtual void start_tx() = 0; virtual void start_stmt() = 0; void set_initial_savepoint() { /* Set the initial savepoint. If the first statement in the transaction fails, we need something to roll back to, without rolling back the entire transaction. */ do_set_savepoint(); m_writes_at_last_savepoint= m_write_count; } /* Called when a "top-level" statement inside a transaction completes successfully and its changes become part of the transaction's changes. */ void make_stmt_savepoint_permanent() { // Take another RocksDB savepoint only if we had changes since the last // one. This is very important for long transactions doing lots of // SELECTs. if (m_writes_at_last_savepoint != m_write_count) { do_set_savepoint(); m_writes_at_last_savepoint= m_write_count; } } /* Rollback to the savepoint we've set before the last statement */ void rollback_to_stmt_savepoint() { if (m_writes_at_last_savepoint != m_write_count) { do_rollback_to_savepoint(); /* RollbackToSavePoint "removes the most recent SetSavePoint()", so we need to set it again so that next statement can roll back to this stage. It's ok to do it here at statement end (instead of doing it at next statement start) because setting a savepoint is cheap. */ do_set_savepoint(); m_writes_at_last_savepoint= m_write_count; } } virtual void rollback_stmt() = 0; void set_tx_failed(bool failed_arg) { m_is_tx_failed = failed_arg; } bool can_prepare() const { if (m_rollback_only) { my_error(ER_ROLLBACK_ONLY, MYF(0)); return false; } return true; } int rollback_to_savepoint(void *const savepoint) { if (has_modifications()) { my_error(ER_ROLLBACK_TO_SAVEPOINT, MYF(0)); m_rollback_only = true; return HA_EXIT_FAILURE; } return HA_EXIT_SUCCESS; } /* This is used by transactions started with "START TRANSACTION WITH " "CONSISTENT [ROCKSDB] SNAPSHOT". When tx_read_only is turned on, snapshot has to be created via DB::GetSnapshot(), not via Transaction API. */ bool is_tx_read_only() const { return m_tx_read_only; } bool is_two_phase() const { return m_is_two_phase; } void set_tx_read_only(bool val) { m_tx_read_only = val; } explicit Rdb_transaction(THD *const thd) : m_thd(thd), m_tbl_io_perf(nullptr) { RDB_MUTEX_LOCK_CHECK(s_tx_list_mutex); s_tx_list.insert(this); RDB_MUTEX_UNLOCK_CHECK(s_tx_list_mutex); } virtual ~Rdb_transaction() { RDB_MUTEX_LOCK_CHECK(s_tx_list_mutex); s_tx_list.erase(this); RDB_MUTEX_UNLOCK_CHECK(s_tx_list_mutex); } }; /* This is a rocksdb transaction. Its members represent the current transaction, which consists of: - the snapshot - the changes we've made but are not seeing yet. The changes are made to individual tables, which store them here and then this object commits them on commit. */ class Rdb_transaction_impl : public Rdb_transaction { rocksdb::Transaction *m_rocksdb_tx = nullptr; rocksdb::Transaction *m_rocksdb_reuse_tx = nullptr; public: void set_lock_timeout(int timeout_sec_arg) override { if (m_rocksdb_tx) m_rocksdb_tx->SetLockTimeout(rdb_convert_sec_to_ms(m_timeout_sec)); } void set_sync(bool sync) override { if (m_rocksdb_tx) m_rocksdb_tx->GetWriteOptions()->sync = sync; } void release_lock(rocksdb::ColumnFamilyHandle *const column_family, const std::string &rowkey) override { if (!THDVAR(m_thd, lock_scanned_rows)) { m_rocksdb_tx->UndoGetForUpdate(column_family, rocksdb::Slice(rowkey)); } } virtual bool is_writebatch_trx() const override { return false; } private: void release_tx(void) { // We are done with the current active transaction object. Preserve it // for later reuse. DBUG_ASSERT(m_rocksdb_reuse_tx == nullptr); m_rocksdb_reuse_tx = m_rocksdb_tx; m_rocksdb_tx = nullptr; } bool prepare(const rocksdb::TransactionName &name) override { rocksdb::Status s; s = m_rocksdb_tx->SetName(name); if (!s.ok()) { rdb_handle_io_error(s, RDB_IO_ERROR_TX_COMMIT); return false; } s = merge_auto_incr_map(m_rocksdb_tx->GetWriteBatch()->GetWriteBatch()); if (!s.ok()) { rdb_handle_io_error(s, RDB_IO_ERROR_TX_COMMIT); return false; } s = m_rocksdb_tx->Prepare(); if (!s.ok()) { rdb_handle_io_error(s, RDB_IO_ERROR_TX_COMMIT); return false; } return true; } bool commit_no_binlog() override { bool res = false; rocksdb::Status s; s = merge_auto_incr_map(m_rocksdb_tx->GetWriteBatch()->GetWriteBatch()); if (!s.ok()) { rdb_handle_io_error(s, RDB_IO_ERROR_TX_COMMIT); res = true; goto error; } release_snapshot(); s = m_rocksdb_tx->Commit(); if (!s.ok()) { rdb_handle_io_error(s, RDB_IO_ERROR_TX_COMMIT); res = true; goto error; } error: /* Save the transaction object to be reused */ release_tx(); m_write_count = 0; m_insert_count = 0; m_update_count = 0; m_delete_count = 0; m_lock_count = 0; set_tx_read_only(false); m_rollback_only = false; return res; } public: void rollback() override { m_write_count = 0; m_insert_count = 0; m_update_count = 0; m_delete_count = 0; m_lock_count = 0; m_auto_incr_map.clear(); m_ddl_transaction = false; if (m_rocksdb_tx) { release_snapshot(); /* This will also release all of the locks: */ m_rocksdb_tx->Rollback(); /* Save the transaction object to be reused */ release_tx(); set_tx_read_only(false); m_rollback_only = false; } } void acquire_snapshot(bool acquire_now) override { if (m_read_opts.snapshot == nullptr) { #ifdef MARIAROCKS_NOT_YET const auto thd_ss = std::static_pointer_cast( m_thd->get_explicit_snapshot()); if (thd_ss) { m_explicit_snapshot = thd_ss; } if (m_explicit_snapshot) { auto snapshot = m_explicit_snapshot->get_snapshot()->snapshot(); snapshot_created(snapshot); } else #endif if (is_tx_read_only()) { snapshot_created(rdb->GetSnapshot()); } else if (acquire_now) { m_rocksdb_tx->SetSnapshot(); snapshot_created(m_rocksdb_tx->GetSnapshot()); } else if (!m_is_delayed_snapshot) { m_rocksdb_tx->SetSnapshotOnNextOperation(m_notifier); m_is_delayed_snapshot = true; } } } void release_snapshot() override { bool need_clear = m_is_delayed_snapshot; if (m_read_opts.snapshot != nullptr) { m_snapshot_timestamp = 0; #ifdef MARIAROCKS_NOT_YET if (m_explicit_snapshot) { m_explicit_snapshot.reset(); need_clear = false; } else #endif if (is_tx_read_only()) { rdb->ReleaseSnapshot(m_read_opts.snapshot); need_clear = false; } else { need_clear = true; } m_read_opts.snapshot = nullptr; } if (need_clear && m_rocksdb_tx != nullptr) m_rocksdb_tx->ClearSnapshot(); } bool has_snapshot() { return m_read_opts.snapshot != nullptr; } rocksdb::Status put(rocksdb::ColumnFamilyHandle *const column_family, const rocksdb::Slice &key, const rocksdb::Slice &value) override { ++m_write_count; ++m_lock_count; if (m_write_count > m_max_row_locks || m_lock_count > m_max_row_locks) return rocksdb::Status::Aborted(rocksdb::Status::kLockLimit); return m_rocksdb_tx->Put(column_family, key, value); } rocksdb::Status delete_key(rocksdb::ColumnFamilyHandle *const column_family, const rocksdb::Slice &key) override { ++m_write_count; ++m_lock_count; if (m_write_count > m_max_row_locks || m_lock_count > m_max_row_locks) return rocksdb::Status::Aborted(rocksdb::Status::kLockLimit); return m_rocksdb_tx->Delete(column_family, key); } rocksdb::Status single_delete(rocksdb::ColumnFamilyHandle *const column_family, const rocksdb::Slice &key) override { ++m_write_count; ++m_lock_count; if (m_write_count > m_max_row_locks || m_lock_count > m_max_row_locks) return rocksdb::Status::Aborted(rocksdb::Status::kLockLimit); return m_rocksdb_tx->SingleDelete(column_family, key); } bool has_modifications() const override { return m_rocksdb_tx->GetWriteBatch() && m_rocksdb_tx->GetWriteBatch()->GetWriteBatch() && m_rocksdb_tx->GetWriteBatch()->GetWriteBatch()->Count() > 0; } rocksdb::WriteBatchBase *get_write_batch() override { if (is_two_phase()) { return m_rocksdb_tx->GetCommitTimeWriteBatch(); } return m_rocksdb_tx->GetWriteBatch()->GetWriteBatch(); } /* Return a WriteBatch that one can write to. The writes will skip any transaction locking. The writes WILL be visible to the transaction. */ rocksdb::WriteBatchBase *get_indexed_write_batch() override { ++m_write_count; return m_rocksdb_tx->GetWriteBatch(); } rocksdb::Status get(rocksdb::ColumnFamilyHandle *const column_family, const rocksdb::Slice &key, rocksdb::PinnableSlice *const value) const override { // clean PinnableSlice right begfore Get() for multiple gets per statement // the resources after the last Get in a statement are cleared in // handler::reset call value->Reset(); global_stats.queries[QUERIES_POINT].inc(); return m_rocksdb_tx->Get(m_read_opts, column_family, key, value); } rocksdb::Status get_for_update(rocksdb::ColumnFamilyHandle *const column_family, const rocksdb::Slice &key, rocksdb::PinnableSlice *const value, bool exclusive) override { if (++m_lock_count > m_max_row_locks) return rocksdb::Status::Aborted(rocksdb::Status::kLockLimit); if (value != nullptr) { value->Reset(); } return m_rocksdb_tx->GetForUpdate(m_read_opts, column_family, key, value, exclusive); } rocksdb::Iterator * get_iterator(const rocksdb::ReadOptions &options, rocksdb::ColumnFamilyHandle *const column_family) override { global_stats.queries[QUERIES_RANGE].inc(); return m_rocksdb_tx->GetIterator(options, column_family); } const rocksdb::Transaction *get_rdb_trx() const { return m_rocksdb_tx; } bool is_tx_started() const override { return (m_rocksdb_tx != nullptr); } void start_tx() override { rocksdb::TransactionOptions tx_opts; rocksdb::WriteOptions write_opts; tx_opts.set_snapshot = false; tx_opts.lock_timeout = rdb_convert_sec_to_ms(m_timeout_sec); tx_opts.deadlock_detect = THDVAR(m_thd, deadlock_detect); tx_opts.deadlock_detect_depth = THDVAR(m_thd, deadlock_detect_depth); // If this variable is set, this will write commit time write batch // information on recovery or memtable flush. tx_opts.use_only_the_last_commit_time_batch_for_recovery = THDVAR(m_thd, commit_time_batch_for_recovery); tx_opts.max_write_batch_size = THDVAR(m_thd, write_batch_max_bytes); write_opts.sync = (rocksdb_flush_log_at_trx_commit == FLUSH_LOG_SYNC); write_opts.disableWAL = THDVAR(m_thd, write_disable_wal); write_opts.ignore_missing_column_families = THDVAR(m_thd, write_ignore_missing_column_families); m_is_two_phase = rocksdb_enable_2pc; commit_ordered_done= false; /* If m_rocksdb_reuse_tx is null this will create a new transaction object. Otherwise it will reuse the existing one. */ m_rocksdb_tx = rdb->BeginTransaction(write_opts, tx_opts, m_rocksdb_reuse_tx); m_rocksdb_reuse_tx = nullptr; m_read_opts = rocksdb::ReadOptions(); set_initial_savepoint(); m_ddl_transaction = false; } /* Implementations of do_*savepoint based on rocksdB::Transaction savepoints */ void do_set_savepoint() override { m_rocksdb_tx->SetSavePoint(); } void do_rollback_to_savepoint() override { m_rocksdb_tx->RollbackToSavePoint(); } /* Start a statement inside a multi-statement transaction. @todo: are we sure this is called once (and not several times) per statement start? For hooking to start of statement that is its own transaction, see ha_rocksdb::external_lock(). */ void start_stmt() override { // Set the snapshot to delayed acquisition (SetSnapshotOnNextOperation) acquire_snapshot(false); } /* This must be called when last statement is rolled back, but the transaction continues */ void rollback_stmt() override { /* TODO: here we must release the locks taken since the start_stmt() call */ if (m_rocksdb_tx) { const rocksdb::Snapshot *const org_snapshot = m_rocksdb_tx->GetSnapshot(); rollback_to_stmt_savepoint(); const rocksdb::Snapshot *const cur_snapshot = m_rocksdb_tx->GetSnapshot(); if (org_snapshot != cur_snapshot) { if (org_snapshot != nullptr) m_snapshot_timestamp = 0; m_read_opts.snapshot = cur_snapshot; if (cur_snapshot != nullptr) rdb->GetEnv()->GetCurrentTime(&m_snapshot_timestamp); else m_is_delayed_snapshot = true; } } } explicit Rdb_transaction_impl(THD *const thd) : Rdb_transaction(thd), m_rocksdb_tx(nullptr) { // Create a notifier that can be called when a snapshot gets generated. m_notifier = std::make_shared(this); } virtual ~Rdb_transaction_impl() { rollback(); // Theoretically the notifier could outlive the Rdb_transaction_impl // (because of the shared_ptr), so let it know it can't reference // the transaction anymore. m_notifier->detach(); // Free any transaction memory that is still hanging around. delete m_rocksdb_reuse_tx; DBUG_ASSERT(m_rocksdb_tx == nullptr); } }; /* This is a rocksdb write batch. This class doesn't hold or wait on any transaction locks (skips rocksdb transaction API) thus giving better performance. Currently this is only used for replication threads which are guaranteed to be non-conflicting. Any further usage of this class should completely be thought thoroughly. */ class Rdb_writebatch_impl : public Rdb_transaction { rocksdb::WriteBatchWithIndex *m_batch; rocksdb::WriteOptions write_opts; // Called after commit/rollback. void reset() { m_batch->Clear(); m_read_opts = rocksdb::ReadOptions(); m_ddl_transaction = false; } private: bool prepare(const rocksdb::TransactionName &name) override { return true; } bool commit_no_binlog() override { bool res = false; rocksdb::Status s; rocksdb::TransactionDBWriteOptimizations optimize; optimize.skip_concurrency_control = true; s = merge_auto_incr_map(m_batch->GetWriteBatch()); if (!s.ok()) { rdb_handle_io_error(s, RDB_IO_ERROR_TX_COMMIT); res = true; goto error; } release_snapshot(); s = rdb->Write(write_opts, optimize, m_batch->GetWriteBatch()); if (!s.ok()) { rdb_handle_io_error(s, RDB_IO_ERROR_TX_COMMIT); res = true; goto error; } error: reset(); m_write_count = 0; m_insert_count = 0; m_update_count = 0; m_delete_count = 0; set_tx_read_only(false); m_rollback_only = false; return res; } /* Implementations of do_*savepoint based on rocksdB::WriteBatch savepoints */ void do_set_savepoint() override { m_batch->SetSavePoint(); } void do_rollback_to_savepoint() override { m_batch->RollbackToSavePoint(); } public: bool is_writebatch_trx() const override { return true; } void set_lock_timeout(int timeout_sec_arg) override { // Nothing to do here. } void set_sync(bool sync) override { write_opts.sync = sync; } void release_lock(rocksdb::ColumnFamilyHandle *const column_family, const std::string &rowkey) override { // Nothing to do here since we don't hold any row locks. } void rollback() override { m_write_count = 0; m_insert_count = 0; m_update_count = 0; m_delete_count = 0; m_lock_count = 0; release_snapshot(); reset(); set_tx_read_only(false); m_rollback_only = false; } void acquire_snapshot(bool acquire_now) override { if (m_read_opts.snapshot == nullptr) snapshot_created(rdb->GetSnapshot()); } void release_snapshot() override { if (m_read_opts.snapshot != nullptr) { rdb->ReleaseSnapshot(m_read_opts.snapshot); m_read_opts.snapshot = nullptr; } } rocksdb::Status put(rocksdb::ColumnFamilyHandle *const column_family, const rocksdb::Slice &key, const rocksdb::Slice &value) override { ++m_write_count; m_batch->Put(column_family, key, value); // Note Put/Delete in write batch doesn't return any error code. We simply // return OK here. return rocksdb::Status::OK(); } rocksdb::Status delete_key(rocksdb::ColumnFamilyHandle *const column_family, const rocksdb::Slice &key) override { ++m_write_count; m_batch->Delete(column_family, key); return rocksdb::Status::OK(); } rocksdb::Status single_delete(rocksdb::ColumnFamilyHandle *const column_family, const rocksdb::Slice &key) override { ++m_write_count; m_batch->SingleDelete(column_family, key); return rocksdb::Status::OK(); } bool has_modifications() const override { return m_batch->GetWriteBatch()->Count() > 0; } rocksdb::WriteBatchBase *get_write_batch() override { return m_batch; } rocksdb::WriteBatchBase *get_indexed_write_batch() override { ++m_write_count; return m_batch; } rocksdb::Status get(rocksdb::ColumnFamilyHandle *const column_family, const rocksdb::Slice &key, rocksdb::PinnableSlice *const value) const override { value->Reset(); return m_batch->GetFromBatchAndDB(rdb, m_read_opts, column_family, key, value); } rocksdb::Status get_for_update(rocksdb::ColumnFamilyHandle *const column_family, const rocksdb::Slice &key, rocksdb::PinnableSlice *const value, bool exclusive) override { if (value == nullptr) { rocksdb::PinnableSlice pin_val; rocksdb::Status s = get(column_family, key, &pin_val); pin_val.Reset(); return s; } return get(column_family, key, value); } rocksdb::Iterator * get_iterator(const rocksdb::ReadOptions &options, rocksdb::ColumnFamilyHandle *const column_family) override { const auto it = rdb->NewIterator(options); return m_batch->NewIteratorWithBase(it); } bool is_tx_started() const override { return (m_batch != nullptr); } void start_tx() override { commit_ordered_done= false; // Do we need this here? reset(); write_opts.sync = (rocksdb_flush_log_at_trx_commit == FLUSH_LOG_SYNC); write_opts.disableWAL = THDVAR(m_thd, write_disable_wal); write_opts.ignore_missing_column_families = THDVAR(m_thd, write_ignore_missing_column_families); set_initial_savepoint(); } void start_stmt() override {} void rollback_stmt() override { if (m_batch) rollback_to_stmt_savepoint(); } explicit Rdb_writebatch_impl(THD *const thd) : Rdb_transaction(thd), m_batch(nullptr) { m_batch = new rocksdb::WriteBatchWithIndex(rocksdb::BytewiseComparator(), 0, true); } virtual ~Rdb_writebatch_impl() { rollback(); delete m_batch; } }; void Rdb_snapshot_notifier::SnapshotCreated( const rocksdb::Snapshot *const snapshot) { if (m_owning_tx != nullptr) { m_owning_tx->snapshot_created(snapshot); } } std::multiset Rdb_transaction::s_tx_list; mysql_mutex_t Rdb_transaction::s_tx_list_mutex; static Rdb_transaction *&get_tx_from_thd(THD *const thd) { return *reinterpret_cast( my_core::thd_ha_data(thd, rocksdb_hton)); } namespace { class Rdb_perf_context_guard { Rdb_io_perf m_io_perf; Rdb_io_perf *m_io_perf_ptr; Rdb_transaction *m_tx; uint m_level; public: Rdb_perf_context_guard(const Rdb_perf_context_guard &) = delete; Rdb_perf_context_guard &operator=(const Rdb_perf_context_guard &) = delete; explicit Rdb_perf_context_guard(Rdb_io_perf *io_perf, uint level) : m_io_perf_ptr(io_perf), m_tx(nullptr), m_level(level) { m_io_perf_ptr->start(m_level); } explicit Rdb_perf_context_guard(Rdb_transaction *tx, uint level) : m_io_perf_ptr(nullptr), m_tx(tx), m_level(level) { /* if perf_context information is already being recorded, this becomes a no-op */ if (tx != nullptr) { tx->io_perf_start(&m_io_perf); } } ~Rdb_perf_context_guard() { if (m_tx != nullptr) { m_tx->io_perf_end_and_record(); } else if (m_io_perf_ptr != nullptr) { m_io_perf_ptr->end_and_record(m_level); } } }; } // anonymous namespace /* TODO: maybe, call this in external_lock() and store in ha_rocksdb.. */ static Rdb_transaction *get_or_create_tx(THD *const thd) { Rdb_transaction *&tx = get_tx_from_thd(thd); // TODO: this is called too many times.. O(#rows) if (tx == nullptr) { bool rpl_skip_tx_api= false; // MARIAROCKS_NOT_YET. if ((rpl_skip_tx_api && thd->rgi_slave) || false /* MARIAROCKS_NOT_YET: THDVAR(thd, master_skip_tx_api) && !thd->rgi_slave)*/) { tx = new Rdb_writebatch_impl(thd); } else { tx = new Rdb_transaction_impl(thd); } tx->set_params(THDVAR(thd, lock_wait_timeout), THDVAR(thd, max_row_locks)); tx->start_tx(); } else { tx->set_params(THDVAR(thd, lock_wait_timeout), THDVAR(thd, max_row_locks)); if (!tx->is_tx_started()) { tx->start_tx(); } } return tx; } static int rocksdb_close_connection(handlerton *const hton, THD *const thd) { Rdb_transaction *&tx = get_tx_from_thd(thd); if (tx != nullptr) { int rc = tx->finish_bulk_load(false); if (rc != 0) { // NO_LINT_DEBUG sql_print_error("RocksDB: Error %d finalizing last SST file while " "disconnecting", rc); } delete tx; tx = nullptr; } return HA_EXIT_SUCCESS; } /* * Serializes an xid to a string so that it can * be used as a rocksdb transaction name */ static std::string rdb_xid_to_string(const XID &src) { DBUG_ASSERT(src.gtrid_length >= 0 && src.gtrid_length <= MAXGTRIDSIZE); DBUG_ASSERT(src.bqual_length >= 0 && src.bqual_length <= MAXBQUALSIZE); std::string buf; buf.reserve(RDB_XIDHDR_LEN + src.gtrid_length + src.bqual_length); /* * expand formatID to fill 8 bytes if it doesn't already * then reinterpret bit pattern as unsigned and store in network order */ uchar fidbuf[RDB_FORMATID_SZ]; int64 signed_fid8 = src.formatID; const uint64 raw_fid8 = *reinterpret_cast(&signed_fid8); rdb_netbuf_store_uint64(fidbuf, raw_fid8); buf.append(reinterpret_cast(fidbuf), RDB_FORMATID_SZ); buf.push_back(src.gtrid_length); buf.push_back(src.bqual_length); buf.append(src.data, (src.gtrid_length) + (src.bqual_length)); return buf; } #if 0 // MARIAROCKS: MariaDB doesn't have flush_wal method /** Called by hton->flush_logs after MySQL group commit prepares a set of transactions. */ static bool rocksdb_flush_wal(handlerton* hton __attribute__((__unused__))) DBUG_ASSERT(rdb != nullptr); rocksdb::Status s; /* target_lsn is set to 0 when MySQL wants to sync the wal files */ if ((target_lsn == 0 && !rocksdb_db_options->allow_mmap_writes) || rocksdb_flush_log_at_trx_commit != FLUSH_LOG_NEVER) { rocksdb_wal_group_syncs++; s = rdb->FlushWAL(target_lsn == 0 || rocksdb_flush_log_at_trx_commit == FLUSH_LOG_SYNC); } if (!s.ok()) { rdb_log_status_error(s); return HA_EXIT_FAILURE; } return HA_EXIT_SUCCESS; } #endif /** For a slave, prepare() updates the slave_gtid_info table which tracks the replication progress. */ static int rocksdb_prepare(handlerton* hton, THD* thd, bool prepare_tx) { bool async=false; // This is "ASYNC_COMMIT" feature which is only present in webscalesql Rdb_transaction *&tx = get_tx_from_thd(thd); if (!tx->can_prepare()) { return HA_EXIT_FAILURE; } if (prepare_tx || (!my_core::thd_test_options(thd, OPTION_NOT_AUTOCOMMIT | OPTION_BEGIN))) { /* We were instructed to prepare the whole transaction, or this is an SQL statement end and autocommit is on */ #ifdef MARIAROCKS_NOT_YET /* Storing binlog position inside MyRocks is needed only for restoring MyRocks from backups. This feature is not supported yet. */ std::vector slave_gtid_info; my_core::thd_slave_gtid_info(thd, &slave_gtid_info); for (const auto &it : slave_gtid_info) { rocksdb::WriteBatchBase *const write_batch = tx->get_blind_write_batch(); binlog_manager.update_slave_gtid_info(it.id, it.db, it.gtid, write_batch); } #endif if (tx->is_two_phase()) { /* MariaDB: the following branch is never taken. We always flush at Prepare and rely on RocksDB's internal Group Commit to do some grouping. */ if (thd->durability_property == HA_IGNORE_DURABILITY || async) { tx->set_sync(false); } /* MariaDB: do not flush logs if we are running in a non-crash-safe mode. */ if (!rocksdb_flush_log_at_trx_commit) tx->set_sync(false); XID xid; thd_get_xid(thd, reinterpret_cast(&xid)); if (!tx->prepare(rdb_xid_to_string(xid))) { return HA_EXIT_FAILURE; } /* MariaDB: our Group Commit implementation does not use the hton->flush_logs call (at least currently) so the following is not needed (TODO: will we need this for binlog rotation?) */ #ifdef MARIAROCKS_NOT_YET if (thd->durability_property == HA_IGNORE_DURABILITY ) (rocksdb_flush_log_at_trx_commit != FLUSH_LOG_NEVER)) && THDVAR(thd, flush_log_at_trx_commit)) #endif #ifdef MARIAROCKS_NOT_YET { // MariaRocks: disable the // "write/sync redo log before flushing binlog cache to file" // feature. See a869c56d361bb44f46c0efeb11a8f03561676247 /** we set the log sequence as '1' just to trigger hton->flush_logs */ thd_store_lsn(thd, 1, DB_TYPE_ROCKSDB); } #endif } DEBUG_SYNC(thd, "rocksdb.prepared"); } else tx->make_stmt_savepoint_permanent(); return HA_EXIT_SUCCESS; } /** do nothing for prepare/commit by xid this is needed to avoid crashes in XA scenarios */ static int rocksdb_commit_by_xid(handlerton *const hton, XID *const xid) { DBUG_ENTER_FUNC(); DBUG_ASSERT(hton != nullptr); DBUG_ASSERT(xid != nullptr); DBUG_ASSERT(commit_latency_stats != nullptr); rocksdb::StopWatchNano timer(rocksdb::Env::Default(), true); const auto name = rdb_xid_to_string(*xid); DBUG_ASSERT(!name.empty()); rocksdb::Transaction *const trx = rdb->GetTransactionByName(name); if (trx == nullptr) { DBUG_RETURN(HA_EXIT_FAILURE); } const rocksdb::Status s = trx->Commit(); if (!s.ok()) { rdb_log_status_error(s); DBUG_RETURN(HA_EXIT_FAILURE); } delete trx; // `Add()` is implemented in a thread-safe manner. commit_latency_stats->Add(timer.ElapsedNanos() / 1000); DBUG_RETURN(HA_EXIT_SUCCESS); } static int rocksdb_rollback_by_xid(handlerton *const hton MY_ATTRIBUTE((__unused__)), XID *const xid) { DBUG_ENTER_FUNC(); DBUG_ASSERT(hton != nullptr); DBUG_ASSERT(xid != nullptr); DBUG_ASSERT(rdb != nullptr); const auto name = rdb_xid_to_string(*xid); rocksdb::Transaction *const trx = rdb->GetTransactionByName(name); if (trx == nullptr) { DBUG_RETURN(HA_EXIT_FAILURE); } const rocksdb::Status s = trx->Rollback(); if (!s.ok()) { rdb_log_status_error(s); DBUG_RETURN(HA_EXIT_FAILURE); } delete trx; DBUG_RETURN(HA_EXIT_SUCCESS); } /** Rebuilds an XID from a serialized version stored in a string. */ static void rdb_xid_from_string(const std::string &src, XID *const dst) { DBUG_ASSERT(dst != nullptr); uint offset = 0; uint64 raw_fid8 = rdb_netbuf_to_uint64(reinterpret_cast(src.data())); const int64 signed_fid8 = *reinterpret_cast(&raw_fid8); dst->formatID = signed_fid8; offset += RDB_FORMATID_SZ; dst->gtrid_length = src.at(offset); offset += RDB_GTRID_SZ; dst->bqual_length = src.at(offset); offset += RDB_BQUAL_SZ; DBUG_ASSERT(dst->gtrid_length >= 0 && dst->gtrid_length <= MAXGTRIDSIZE); DBUG_ASSERT(dst->bqual_length >= 0 && dst->bqual_length <= MAXBQUALSIZE); src.copy(dst->data, (dst->gtrid_length) + (dst->bqual_length), RDB_XIDHDR_LEN); } /** Reading last committed binary log info from RocksDB system row. The info is needed for crash safe slave/master to work. */ static int rocksdb_recover(handlerton* hton, XID* xid_list, uint len) #ifdef MARIAROCKS_NOT_YET char* const binlog_file, my_off_t *const binlog_pos, Gtid *const binlog_max_gtid) { #endif { #ifdef MARIAROCKS_NOT_YET if (binlog_file && binlog_pos) { char file_buf[FN_REFLEN + 1] = {0}; my_off_t pos; char gtid_buf[FN_REFLEN + 1] = {0}; if (binlog_manager.read(file_buf, &pos, gtid_buf)) { if (is_binlog_advanced(binlog_file, *binlog_pos, file_buf, pos)) { memcpy(binlog_file, file_buf, FN_REFLEN + 1); *binlog_pos = pos; fprintf(stderr, "RocksDB: Last binlog file position %llu," " file name %s\n", pos, file_buf); if (*gtid_buf) { global_sid_lock->rdlock(); binlog_max_gtid->parse(global_sid_map, gtid_buf); global_sid_lock->unlock(); fprintf(stderr, "RocksDB: Last MySQL Gtid %s\n", gtid_buf); } } } } #endif if (len == 0 || xid_list == nullptr) { return HA_EXIT_SUCCESS; } std::vector trans_list; rdb->GetAllPreparedTransactions(&trans_list); uint count = 0; for (auto &trans : trans_list) { if (count >= len) { break; } auto name = trans->GetName(); rdb_xid_from_string(name, &xid_list[count]); count++; } return count; } /* Handle a commit checkpoint request from server layer. InnoDB does this: We put the request in a queue, so that we can notify upper layer about checkpoint complete when we have flushed the redo log. If we have already flushed all relevant redo log, we notify immediately. MariaRocks just flushes everything right away ATM */ static void rocksdb_checkpoint_request(handlerton *hton, void *cookie) { const rocksdb::Status s= rdb->SyncWAL(); //TODO: what to do on error? if (s.ok()) { rocksdb_wal_group_syncs++; commit_checkpoint_notify_ha(hton, cookie); } } /* @param all: TRUE - commit the transaction FALSE - SQL statement ended */ static void rocksdb_commit_ordered(handlerton *hton, THD* thd, bool all) { // Same assert as InnoDB has DBUG_ASSERT(all || (!thd_test_options(thd, OPTION_NOT_AUTOCOMMIT | OPTION_BEGIN))); Rdb_transaction *&tx = get_tx_from_thd(thd); if (!tx->is_two_phase()) { /* ordered_commit is supposedly slower as it is done sequentially in order to preserve commit order. if we are not required do 2-phase commit with the binlog, do not do anything here. */ return; } tx->set_sync(false); /* This will note the master position also */ tx->commit_ordered_res= tx->commit(); tx->commit_ordered_done= true; } static int rocksdb_commit(handlerton* hton, THD* thd, bool commit_tx) { DBUG_ENTER_FUNC(); DBUG_ASSERT(hton != nullptr); DBUG_ASSERT(thd != nullptr); DBUG_ASSERT(commit_latency_stats != nullptr); rocksdb::StopWatchNano timer(rocksdb::Env::Default(), true); /* note: h->external_lock(F_UNLCK) is called after this function is called) */ Rdb_transaction *&tx = get_tx_from_thd(thd); /* this will trigger saving of perf_context information */ Rdb_perf_context_guard guard(tx, rocksdb_perf_context_level(thd)); if (tx != nullptr) { if (commit_tx || (!my_core::thd_test_options(thd, OPTION_NOT_AUTOCOMMIT | OPTION_BEGIN))) { /* This will not add anything to commit_latency_stats, and this is correct right? */ if (tx->commit_ordered_done) { thd_wakeup_subsequent_commits(thd, 0); DBUG_RETURN((tx->commit_ordered_res? HA_ERR_INTERNAL_ERROR: 0)); } /* We get here - For a COMMIT statement that finishes a multi-statement transaction - For a statement that has its own transaction */ if (thd->slave_thread) { // An attempt to make parallel slave performant (not fully successful, // see MDEV-15372): // First, commit without syncing. This establishes the commit order tx->set_sync(false); bool tx_had_writes = tx->get_write_count()? true : false ; if (tx->commit()) { DBUG_RETURN(HA_ERR_ROCKSDB_COMMIT_FAILED); } thd_wakeup_subsequent_commits(thd, 0); if (tx_had_writes && rocksdb_flush_log_at_trx_commit == FLUSH_LOG_SYNC) { rocksdb::Status s= rdb->FlushWAL(true); if (!s.ok()) DBUG_RETURN(HA_ERR_INTERNAL_ERROR); } } else { /* Not a slave thread */ if (tx->commit()) { DBUG_RETURN(HA_ERR_ROCKSDB_COMMIT_FAILED); } } } else { /* We get here when committing a statement within a transaction. */ tx->make_stmt_savepoint_permanent(); } if (my_core::thd_tx_isolation(thd) <= ISO_READ_COMMITTED) { // For READ_COMMITTED, we release any existing snapshot so that we will // see any changes that occurred since the last statement. tx->release_snapshot(); } } // `Add()` is implemented in a thread-safe manner. commit_latency_stats->Add(timer.ElapsedNanos() / 1000); DBUG_RETURN(HA_EXIT_SUCCESS); } static int rocksdb_rollback(handlerton *const hton, THD *const thd, bool rollback_tx) { Rdb_transaction *&tx = get_tx_from_thd(thd); Rdb_perf_context_guard guard(tx, rocksdb_perf_context_level(thd)); if (tx != nullptr) { if (rollback_tx) { /* We get here, when - ROLLBACK statement is issued. Discard the changes made by the transaction */ tx->rollback(); } else { /* We get here when - a statement with AUTOCOMMIT=1 is being rolled back (because of some error) - a statement inside a transaction is rolled back */ tx->rollback_stmt(); tx->set_tx_failed(true); } if (my_core::thd_tx_isolation(thd) <= ISO_READ_COMMITTED) { // For READ_COMMITTED, we release any existing snapshot so that we will // see any changes that occurred since the last statement. tx->release_snapshot(); } } return HA_EXIT_SUCCESS; } static bool print_stats(THD *const thd, std::string const &type, std::string const &name, std::string const &status, stat_print_fn *stat_print) { return stat_print(thd, type.c_str(), type.size(), name.c_str(), name.size(), status.c_str(), status.size()); } static std::string format_string(const char *const format, ...) { std::string res; va_list args; va_list args_copy; char static_buff[256]; DBUG_ASSERT(format != nullptr); va_start(args, format); va_copy(args_copy, args); // Calculate how much space we will need int len = vsnprintf(nullptr, 0, format, args); va_end(args); if (len < 0) { res = std::string(""); } else if (len == 0) { // Shortcut for an empty string res = std::string(""); } else { // For short enough output use a static buffer char *buff = static_buff; std::unique_ptr dynamic_buff = nullptr; len++; // Add one for null terminator // for longer output use an allocated buffer if (static_cast(len) > sizeof(static_buff)) { dynamic_buff.reset(new char[len]); buff = dynamic_buff.get(); } // Now re-do the vsnprintf with the buffer which is now large enough (void)vsnprintf(buff, len, format, args_copy); // Convert to a std::string. Note we could have created a std::string // large enough and then converted the buffer to a 'char*' and created // the output in place. This would probably work but feels like a hack. // Since this isn't code that needs to be super-performant we are going // with this 'safer' method. res = std::string(buff); } va_end(args_copy); return res; } class Rdb_snapshot_status : public Rdb_tx_list_walker { private: std::string m_data; static std::string current_timestamp(void) { static const char *const format = "%d-%02d-%02d %02d:%02d:%02d"; time_t currtime; struct tm currtm; time(&currtime); localtime_r(&currtime, &currtm); return format_string(format, currtm.tm_year + 1900, currtm.tm_mon + 1, currtm.tm_mday, currtm.tm_hour, currtm.tm_min, currtm.tm_sec); } static std::string get_header(void) { return "\n============================================================\n" + current_timestamp() + " ROCKSDB TRANSACTION MONITOR OUTPUT\n" "============================================================\n" "---------\n" "SNAPSHOTS\n" "---------\n" "LIST OF SNAPSHOTS FOR EACH SESSION:\n"; } static std::string get_footer(void) { return "-----------------------------------------\n" "END OF ROCKSDB TRANSACTION MONITOR OUTPUT\n" "=========================================\n"; } static Rdb_deadlock_info::Rdb_dl_trx_info get_dl_txn_info(const rocksdb::DeadlockInfo &txn, const GL_INDEX_ID &gl_index_id) { Rdb_deadlock_info::Rdb_dl_trx_info txn_data; txn_data.trx_id = txn.m_txn_id; txn_data.table_name = ddl_manager.safe_get_table_name(gl_index_id); if (txn_data.table_name.empty()) { txn_data.table_name = "NOT FOUND; INDEX_ID: " + std::to_string(gl_index_id.index_id); } auto kd = ddl_manager.safe_find(gl_index_id); txn_data.index_name = (kd) ? kd->get_name() : "NOT FOUND; INDEX_ID: " + std::to_string(gl_index_id.index_id); rocksdb::ColumnFamilyHandle *cfh = cf_manager.get_cf(txn.m_cf_id); txn_data.cf_name = cfh->GetName(); txn_data.waiting_key = rdb_hexdump(txn.m_waiting_key.c_str(), txn.m_waiting_key.length()); txn_data.exclusive_lock = txn.m_exclusive; return txn_data; } static Rdb_deadlock_info get_dl_path_trx_info(const rocksdb::DeadlockPath &path_entry) { Rdb_deadlock_info deadlock_info; for (auto it = path_entry.path.begin(); it != path_entry.path.end(); it++) { auto txn = *it; const GL_INDEX_ID gl_index_id = { txn.m_cf_id, rdb_netbuf_to_uint32(reinterpret_cast( txn.m_waiting_key.c_str()))}; deadlock_info.path.push_back(get_dl_txn_info(txn, gl_index_id)); } DBUG_ASSERT_IFF(path_entry.limit_exceeded, path_entry.path.empty()); /* print the first txn in the path to display the full deadlock cycle */ if (!path_entry.path.empty() && !path_entry.limit_exceeded) { auto deadlocking_txn = *(path_entry.path.end() - 1); deadlock_info.victim_trx_id = deadlocking_txn.m_txn_id; deadlock_info.deadlock_time = path_entry.deadlock_time; } return deadlock_info; } public: Rdb_snapshot_status() : m_data(get_header()) {} std::string getResult() { return m_data + get_footer(); } /* Implement Rdb_transaction interface */ /* Create one row in the snapshot status table */ void process_tran(const Rdb_transaction *const tx) override { DBUG_ASSERT(tx != nullptr); /* Calculate the duration the snapshot has existed */ int64_t snapshot_timestamp = tx->m_snapshot_timestamp; if (snapshot_timestamp != 0) { int64_t curr_time; rdb->GetEnv()->GetCurrentTime(&curr_time); char buffer[1024]; #ifdef MARIAROCKS_NOT_YET thd_security_context(tx->get_thd(), buffer, sizeof buffer, 0); #endif m_data += format_string( "---SNAPSHOT, ACTIVE %lld sec\n" "%s\n" "lock count %llu, write count %llu\n" "insert count %llu, update count %llu, delete count %llu\n", (longlong)(curr_time - snapshot_timestamp), buffer, tx->get_lock_count(), tx->get_write_count(), tx->get_insert_count(), tx->get_update_count(), tx->get_delete_count()); } } void populate_deadlock_buffer() { auto dlock_buffer = rdb->GetDeadlockInfoBuffer(); m_data += "----------LATEST DETECTED DEADLOCKS----------\n"; for (auto path_entry : dlock_buffer) { std::string path_data; if (path_entry.limit_exceeded) { path_data += "\n-------DEADLOCK EXCEEDED MAX DEPTH-------\n"; } else { path_data += "\n*** DEADLOCK PATH\n" "=========================================\n"; const auto dl_info = get_dl_path_trx_info(path_entry); const auto deadlock_time = dl_info.deadlock_time; for (auto it = dl_info.path.begin(); it != dl_info.path.end(); it++) { const auto trx_info = *it; path_data += format_string( "TIMESTAMP: %" PRId64 "\n" "TRANSACTION ID: %u\n" "COLUMN FAMILY NAME: %s\n" "WAITING KEY: %s\n" "LOCK TYPE: %s\n" "INDEX NAME: %s\n" "TABLE NAME: %s\n", deadlock_time, trx_info.trx_id, trx_info.cf_name.c_str(), trx_info.waiting_key.c_str(), trx_info.exclusive_lock ? "EXCLUSIVE" : "SHARED", trx_info.index_name.c_str(), trx_info.table_name.c_str()); if (it != dl_info.path.end() - 1) { path_data += "---------------WAITING FOR---------------\n"; } } path_data += format_string("\n--------TRANSACTION ID: %u GOT DEADLOCK---------\n", dl_info.victim_trx_id); } m_data += path_data; } } std::vector get_deadlock_info() { std::vector deadlock_info; auto dlock_buffer = rdb->GetDeadlockInfoBuffer(); for (auto path_entry : dlock_buffer) { if (!path_entry.limit_exceeded) { deadlock_info.push_back(get_dl_path_trx_info(path_entry)); } } return deadlock_info; } }; /** * @brief * walks through all non-replication transactions and copies * out relevant information for information_schema.rocksdb_trx */ class Rdb_trx_info_aggregator : public Rdb_tx_list_walker { private: std::vector *m_trx_info; public: explicit Rdb_trx_info_aggregator(std::vector *const trx_info) : m_trx_info(trx_info) {} void process_tran(const Rdb_transaction *const tx) override { static const std::map state_map = { {rocksdb::Transaction::STARTED, "STARTED"}, {rocksdb::Transaction::AWAITING_PREPARE, "AWAITING_PREPARE"}, {rocksdb::Transaction::PREPARED, "PREPARED"}, {rocksdb::Transaction::AWAITING_COMMIT, "AWAITING_COMMIT"}, {rocksdb::Transaction::COMMITED, "COMMITED"}, {rocksdb::Transaction::AWAITING_ROLLBACK, "AWAITING_ROLLBACK"}, {rocksdb::Transaction::ROLLEDBACK, "ROLLEDBACK"}, }; DBUG_ASSERT(tx != nullptr); THD *const thd = tx->get_thd(); ulong thread_id = thd_get_thread_id(thd); if (tx->is_writebatch_trx()) { const auto wb_impl = static_cast(tx); DBUG_ASSERT(wb_impl); m_trx_info->push_back( {"", /* name */ 0, /* trx_id */ wb_impl->get_write_count(), 0, /* lock_count */ 0, /* timeout_sec */ "", /* state */ "", /* waiting_key */ 0, /* waiting_cf_id */ 1, /*is_replication */ 1, /* skip_trx_api */ wb_impl->is_tx_read_only(), 0, /* deadlock detection */ wb_impl->num_ongoing_bulk_load(), thread_id, "" /* query string */}); } else { const auto tx_impl = static_cast(tx); DBUG_ASSERT(tx_impl); const rocksdb::Transaction *rdb_trx = tx_impl->get_rdb_trx(); if (rdb_trx == nullptr) { return; } char query_buf[NAME_LEN+1]; thd_query_safe(thd, query_buf, sizeof(query_buf)); std::string query_str(query_buf); const auto state_it = state_map.find(rdb_trx->GetState()); DBUG_ASSERT(state_it != state_map.end()); const int is_replication = (thd->rgi_slave != nullptr); uint32_t waiting_cf_id; std::string waiting_key; rdb_trx->GetWaitingTxns(&waiting_cf_id, &waiting_key), m_trx_info->push_back( {rdb_trx->GetName(), rdb_trx->GetID(), tx_impl->get_write_count(), tx_impl->get_lock_count(), tx_impl->get_timeout_sec(), state_it->second, waiting_key, waiting_cf_id, is_replication, 0, /* skip_trx_api */ tx_impl->is_tx_read_only(), rdb_trx->IsDeadlockDetect(), tx_impl->num_ongoing_bulk_load(), thread_id, query_str}); } } }; /* returns a vector of info for all non-replication threads for use by information_schema.rocksdb_trx */ std::vector rdb_get_all_trx_info() { std::vector trx_info; Rdb_trx_info_aggregator trx_info_agg(&trx_info); Rdb_transaction::walk_tx_list(&trx_info_agg); return trx_info; } /* returns a vector of info of recent deadlocks for use by information_schema.rocksdb_deadlock */ std::vector rdb_get_deadlock_info() { Rdb_snapshot_status showStatus; Rdb_transaction::walk_tx_list(&showStatus); return showStatus.get_deadlock_info(); } #ifdef MARIAROCKS_NOT_YET /* Generate the snapshot status table */ static bool rocksdb_show_snapshot_status(handlerton *const hton, THD *const thd, stat_print_fn *const stat_print) { Rdb_snapshot_status showStatus; Rdb_transaction::walk_tx_list(&showStatus); showStatus.populate_deadlock_buffer(); /* Send the result data back to MySQL */ return print_stats(thd, "rocksdb", "", showStatus.getResult(), stat_print); } #endif /* This is called for SHOW ENGINE ROCKSDB STATUS | LOGS | etc. For now, produce info about live files (which gives an imprecise idea about what column families are there). */ static bool rocksdb_show_status(handlerton *const hton, THD *const thd, stat_print_fn *const stat_print, enum ha_stat_type stat_type) { DBUG_ASSERT(hton != nullptr); DBUG_ASSERT(thd != nullptr); DBUG_ASSERT(stat_print != nullptr); bool res = false; char buf[100] = {'\0'}; if (stat_type == HA_ENGINE_STATUS) { DBUG_ASSERT(rdb != nullptr); std::string str; /* Global DB Statistics */ if (rocksdb_stats) { str = rocksdb_stats->ToString(); // Use the same format as internal RocksDB statistics entries to make // sure that output will look unified. DBUG_ASSERT(commit_latency_stats != nullptr); snprintf(buf, sizeof(buf), "rocksdb.commit_latency statistics " "Percentiles :=> 50 : %.2f 95 : %.2f " "99 : %.2f 100 : %.2f\n", commit_latency_stats->Percentile(50), commit_latency_stats->Percentile(95), commit_latency_stats->Percentile(99), commit_latency_stats->Percentile(100)); str.append(buf); uint64_t v = 0; // Retrieve additional stalling related numbers from RocksDB and append // them to the buffer meant for displaying detailed statistics. The intent // here is to avoid adding another row to the query output because of // just two numbers. // // NB! We're replacing hyphens with underscores in output to better match // the existing naming convention. if (rdb->GetIntProperty("rocksdb.is-write-stopped", &v)) { snprintf(buf, sizeof(buf), "rocksdb.is_write_stopped COUNT : %llu\n", (ulonglong)v); str.append(buf); } if (rdb->GetIntProperty("rocksdb.actual-delayed-write-rate", &v)) { snprintf(buf, sizeof(buf), "rocksdb.actual_delayed_write_rate " "COUNT : %llu\n", (ulonglong)v); str.append(buf); } res |= print_stats(thd, "STATISTICS", "rocksdb", str, stat_print); } /* Per DB stats */ if (rdb->GetProperty("rocksdb.dbstats", &str)) { res |= print_stats(thd, "DBSTATS", "rocksdb", str, stat_print); } /* Per column family stats */ for (const auto &cf_name : cf_manager.get_cf_names()) { rocksdb::ColumnFamilyHandle *cfh = cf_manager.get_cf(cf_name); if (cfh == nullptr) { continue; } if (!rdb->GetProperty(cfh, "rocksdb.cfstats", &str)) { continue; } res |= print_stats(thd, "CF_COMPACTION", cf_name, str, stat_print); } /* Memory Statistics */ std::vector dbs; std::unordered_set cache_set; size_t internal_cache_count = 0; size_t kDefaultInternalCacheSize = 8 * 1024 * 1024; dbs.push_back(rdb); cache_set.insert(rocksdb_tbl_options->block_cache.get()); for (const auto &cf_handle : cf_manager.get_all_cf()) { rocksdb::ColumnFamilyDescriptor cf_desc; cf_handle->GetDescriptor(&cf_desc); auto *const table_factory = cf_desc.options.table_factory.get(); if (table_factory != nullptr) { std::string tf_name = table_factory->Name(); if (tf_name.find("BlockBasedTable") != std::string::npos) { const rocksdb::BlockBasedTableOptions *const bbt_opt = reinterpret_cast( table_factory->GetOptions()); if (bbt_opt != nullptr) { if (bbt_opt->block_cache.get() != nullptr) { cache_set.insert(bbt_opt->block_cache.get()); } else { internal_cache_count++; } cache_set.insert(bbt_opt->block_cache_compressed.get()); } } } } std::map temp_usage_by_type; str.clear(); rocksdb::MemoryUtil::GetApproximateMemoryUsageByType(dbs, cache_set, &temp_usage_by_type); snprintf(buf, sizeof(buf), "\nMemTable Total: %llu", (ulonglong)temp_usage_by_type[rocksdb::MemoryUtil::kMemTableTotal]); str.append(buf); snprintf(buf, sizeof(buf), "\nMemTable Unflushed: %llu", (ulonglong)temp_usage_by_type[rocksdb::MemoryUtil::kMemTableUnFlushed]); str.append(buf); snprintf(buf, sizeof(buf), "\nTable Readers Total: %llu", (ulonglong)temp_usage_by_type[rocksdb::MemoryUtil::kTableReadersTotal]); str.append(buf); snprintf(buf, sizeof(buf), "\nCache Total: %llu", (ulonglong)temp_usage_by_type[rocksdb::MemoryUtil::kCacheTotal]); str.append(buf); snprintf(buf, sizeof(buf), "\nDefault Cache Capacity: %llu", (ulonglong)internal_cache_count * kDefaultInternalCacheSize); str.append(buf); res |= print_stats(thd, "MEMORY_STATS", "rocksdb", str, stat_print); /* Show the background thread status */ std::vector thread_list; rocksdb::Status s = rdb->GetEnv()->GetThreadList(&thread_list); if (!s.ok()) { sql_print_error("RocksDB: Returned error (%s) from GetThreadList.\n", s.ToString().c_str()); res |= true; } else { /* For each background thread retrieved, print out its information */ for (auto &it : thread_list) { /* Only look at background threads. Ignore user threads, if any. */ if (it.thread_type > rocksdb::ThreadStatus::LOW_PRIORITY) { continue; } str = "\nthread_type: " + it.GetThreadTypeName(it.thread_type) + "\ncf_name: " + it.cf_name + "\noperation_type: " + it.GetOperationName(it.operation_type) + "\noperation_stage: " + it.GetOperationStageName(it.operation_stage) + "\nelapsed_time_ms: " + it.MicrosToString(it.op_elapsed_micros); for (auto &it_props : it.InterpretOperationProperties(it.operation_type, it.op_properties)) { str += "\n" + it_props.first + ": " + std::to_string(it_props.second); } str += "\nstate_type: " + it.GetStateName(it.state_type); res |= print_stats(thd, "BG_THREADS", std::to_string(it.thread_id), str, stat_print); } } #ifdef MARIAROCKS_NOT_YET /* Explicit snapshot information */ str.clear(); { std::lock_guard lock(explicit_snapshot_mutex); for (const auto &elem : explicit_snapshots) { const auto &ss = elem.second.lock(); DBUG_ASSERT(ss != nullptr); const auto &info = ss->ss_info; str += "\nSnapshot ID: " + std::to_string(info.snapshot_id) + "\nBinlog File: " + info.binlog_file + "\nBinlog Pos: " + std::to_string(info.binlog_pos) + "\nGtid Executed: " + info.gtid_executed + "\n"; } } #endif if (!str.empty()) { res |= print_stats(thd, "EXPLICIT_SNAPSHOTS", "rocksdb", str, stat_print); } #ifdef MARIAROCKS_NOT_YET } else if (stat_type == HA_ENGINE_TRX) { /* Handle the SHOW ENGINE ROCKSDB TRANSACTION STATUS command */ res |= rocksdb_show_snapshot_status(hton, thd, stat_print); #endif } return res; } static inline void rocksdb_register_tx(handlerton *const hton, THD *const thd, Rdb_transaction *const tx) { DBUG_ASSERT(tx != nullptr); trans_register_ha(thd, FALSE, rocksdb_hton); if (my_core::thd_test_options(thd, OPTION_NOT_AUTOCOMMIT | OPTION_BEGIN)) { tx->start_stmt(); trans_register_ha(thd, TRUE, rocksdb_hton); } } static const char *ha_rocksdb_exts[] = {NullS}; #ifdef MARIAROCKS_NOT_YET static bool rocksdb_explicit_snapshot( handlerton *const /* hton */, /*!< in: RocksDB handlerton */ THD *const thd, /*!< in: MySQL thread handle */ snapshot_info_st *ss_info) /*!< out: Snapshot information */ { switch (ss_info->op) { case snapshot_operation::SNAPSHOT_CREATE: { if (mysql_bin_log_is_open()) { mysql_bin_log_lock_commits(ss_info); } auto s = Rdb_explicit_snapshot::create(ss_info, rdb, rdb->GetSnapshot()); if (mysql_bin_log_is_open()) { mysql_bin_log_unlock_commits(ss_info); } thd->set_explicit_snapshot(s); return s == nullptr; } case snapshot_operation::SNAPSHOT_ATTACH: { auto s = Rdb_explicit_snapshot::get(ss_info->snapshot_id); if (!s) { return true; } *ss_info = s->ss_info; thd->set_explicit_snapshot(s); return false; } case snapshot_operation::SNAPSHOT_RELEASE: { if (!thd->get_explicit_snapshot()) { return true; } *ss_info = thd->get_explicit_snapshot()->ss_info; thd->set_explicit_snapshot(nullptr); return false; } default: DBUG_ASSERT(false); return true; } return true; } #endif /* Supporting START TRANSACTION WITH CONSISTENT [ROCKSDB] SNAPSHOT Features: 1. Supporting START TRANSACTION WITH CONSISTENT SNAPSHOT 2. Getting current binlog position in addition to #1. The second feature is done by START TRANSACTION WITH CONSISTENT ROCKSDB SNAPSHOT. This is Facebook's extension, and it works like existing START TRANSACTION WITH CONSISTENT INNODB SNAPSHOT. - When not setting engine, START TRANSACTION WITH CONSISTENT SNAPSHOT takes both InnoDB and RocksDB snapshots, and both InnoDB and RocksDB participate in transaction. When executing COMMIT, both InnoDB and RocksDB modifications are committed. Remember that XA is not supported yet, so mixing engines is not recommended anyway. - When setting engine, START TRANSACTION WITH CONSISTENT.. takes snapshot for the specified engine only. But it starts both InnoDB and RocksDB transactions. */ static int rocksdb_start_tx_and_assign_read_view( handlerton *const hton, /*!< in: RocksDB handlerton */ THD *const thd /*!< in: MySQL thread handle of the user for whom the transaction should be committed */ ) #ifdef MARIAROCKS_NOT_YET snapshot_info_st *ss_info) /*!< in/out: Snapshot info like binlog file, pos, gtid executed and snapshot ID */ #endif { ulong const tx_isolation = my_core::thd_tx_isolation(thd); if (tx_isolation != ISO_REPEATABLE_READ) { my_error(ER_ISOLATION_LEVEL_WITH_CONSISTENT_SNAPSHOT, MYF(0)); return HA_EXIT_FAILURE; } #ifdef MARIADB_NOT_YET if (ss_info) { if (mysql_bin_log_is_open()) { mysql_bin_log_lock_commits(ss_info); } else { return HA_EXIT_FAILURE; } #endif /* MariaDB: there is no need to call mysql_bin_log_lock_commits and then unlock back. SQL layer calls start_consistent_snapshot() for all engines, including the binlog under LOCK_commit_ordered mutex. The mutex prevents binlog commits from happening (right?) while the storage engine(s) allocate read snapshots. That way, each storage engine is synchronized with current binlog position. */ mysql_mutex_assert_owner(&LOCK_commit_ordered); Rdb_transaction *const tx = get_or_create_tx(thd); Rdb_perf_context_guard guard(tx, rocksdb_perf_context_level(thd)); DBUG_ASSERT(!tx->has_snapshot()); tx->set_tx_read_only(true); rocksdb_register_tx(hton, thd, tx); tx->acquire_snapshot(true); #ifdef MARIADB_NOT_YET if (ss_info) { mysql_bin_log_unlock_commits(ss_info); } #endif return HA_EXIT_SUCCESS; } #ifdef MARIADB_NOT_YET static int rocksdb_start_tx_with_shared_read_view( handlerton *const hton, /*!< in: RocksDB handlerton */ THD *const thd) /*!< in: MySQL thread handle of the user for whom the transaction should be committed */ #ifdef MARIADB_NOT_YET snapshot_info_st *ss_info) /*!< out: Snapshot info like binlog file, pos, gtid executed and snapshot ID */ #endif { DBUG_ASSERT(thd != nullptr); int error = HA_EXIT_SUCCESS; ulong const tx_isolation = my_core::thd_tx_isolation(thd); if (tx_isolation != ISO_REPEATABLE_READ) { my_error(ER_ISOLATION_LEVEL_WITH_CONSISTENT_SNAPSHOT, MYF(0)); return HA_EXIT_FAILURE; } Rdb_transaction *tx = nullptr; #ifdef MARIADB_NOT_YET std::shared_ptr explicit_snapshot; const auto op = ss_info->op; DBUG_ASSERT(op == snapshot_operation::SNAPSHOT_CREATE || op == snapshot_operation::SNAPSHOT_ATTACH); // case: if binlogs are available get binlog file/pos and gtid info if (op == snapshot_operation::SNAPSHOT_CREATE && mysql_bin_log_is_open()) { mysql_bin_log_lock_commits(ss_info); } if (op == snapshot_operation::SNAPSHOT_ATTACH) { explicit_snapshot = Rdb_explicit_snapshot::get(ss_info->snapshot_id); if (!explicit_snapshot) { my_printf_error(ER_UNKNOWN_ERROR, "Snapshot %llu does not exist", MYF(0), ss_info->snapshot_id); error = HA_EXIT_FAILURE; } } #endif // case: all good till now if (error == HA_EXIT_SUCCESS) { tx = get_or_create_tx(thd); Rdb_perf_context_guard guard(tx, rocksdb_perf_context_level(thd)); #ifdef MARIADB_NOT_YET if (explicit_snapshot) { tx->m_explicit_snapshot = explicit_snapshot; } #endif DBUG_ASSERT(!tx->has_snapshot()); tx->set_tx_read_only(true); rocksdb_register_tx(hton, thd, tx); tx->acquire_snapshot(true); #ifdef MARIADB_NOT_YET // case: an explicit snapshot was not assigned to this transaction if (!tx->m_explicit_snapshot) { tx->m_explicit_snapshot = Rdb_explicit_snapshot::create(ss_info, rdb, tx->m_read_opts.snapshot); if (!tx->m_explicit_snapshot) { my_printf_error(ER_UNKNOWN_ERROR, "Could not create snapshot", MYF(0)); error = HA_EXIT_FAILURE; } } #endif } #ifdef MARIADB_NOT_YET // case: unlock the binlog if (op == snapshot_operation::SNAPSHOT_CREATE && mysql_bin_log_is_open()) { mysql_bin_log_unlock_commits(ss_info); } DBUG_ASSERT(error == HA_EXIT_FAILURE || tx->m_explicit_snapshot); // copy over the snapshot details to pass to the upper layers if (tx->m_explicit_snapshot) { *ss_info = tx->m_explicit_snapshot->ss_info; ss_info->op = op; } #endif return error; } #endif /* Dummy SAVEPOINT support. This is needed for long running transactions * like mysqldump (https://bugs.mysql.com/bug.php?id=71017). * Current SAVEPOINT does not correctly handle ROLLBACK and does not return * errors. This needs to be addressed in future versions (Issue#96). */ static int rocksdb_savepoint(handlerton *const hton, THD *const thd, void *const savepoint) { return HA_EXIT_SUCCESS; } static int rocksdb_rollback_to_savepoint(handlerton *const hton, THD *const thd, void *const savepoint) { Rdb_transaction *&tx = get_tx_from_thd(thd); return tx->rollback_to_savepoint(savepoint); } static bool rocksdb_rollback_to_savepoint_can_release_mdl(handlerton *const hton, THD *const thd) { return true; } #ifdef MARIAROCKS_NOT_YET /* This is called for INFORMATION_SCHEMA */ static void rocksdb_update_table_stats( /* per-table stats callback */ void (*cb)(const char *db, const char *tbl, bool is_partition, my_io_perf_t *r, my_io_perf_t *w, my_io_perf_t *r_blob, my_io_perf_t *r_primary, my_io_perf_t *r_secondary, page_stats_t *page_stats, comp_stats_t *comp_stats, int n_lock_wait, int n_lock_wait_timeout, int n_lock_deadlock, const char *engine)) { my_io_perf_t io_perf_read; my_io_perf_t io_perf_write; my_io_perf_t io_perf; page_stats_t page_stats; comp_stats_t comp_stats; uint lock_wait_timeout_stats; uint deadlock_stats; uint lock_wait_stats; std::vector tablenames; /* Most of these are for innodb, so setting them to 0. TODO: possibly separate out primary vs. secondary index reads */ memset(&io_perf, 0, sizeof(io_perf)); memset(&page_stats, 0, sizeof(page_stats)); memset(&comp_stats, 0, sizeof(comp_stats)); memset(&io_perf_write, 0, sizeof(io_perf_write)); tablenames = rdb_open_tables.get_table_names(); for (const auto &it : tablenames) { Rdb_table_handler *table_handler; std::string str, dbname, tablename, partname; char dbname_sys[NAME_LEN + 1]; char tablename_sys[NAME_LEN + 1]; bool is_partition; if (rdb_normalize_tablename(it, &str) != HA_EXIT_SUCCESS) { /* Function needs to return void because of the interface and we've * detected an error which shouldn't happen. There's no way to let * caller know that something failed. */ SHIP_ASSERT(false); return; } if (rdb_split_normalized_tablename(str, &dbname, &tablename, &partname)) { continue; } is_partition = (partname.size() != 0); table_handler = rdb_open_tables.get_table_handler(it.c_str()); if (table_handler == nullptr) { continue; } io_perf_read.bytes = table_handler->m_io_perf_read.bytes.load(); io_perf_read.requests = table_handler->m_io_perf_read.requests.load(); io_perf_write.bytes = table_handler->m_io_perf_write.bytes.load(); io_perf_write.requests = table_handler->m_io_perf_write.requests.load(); lock_wait_timeout_stats = table_handler->m_lock_wait_timeout_counter.load(); deadlock_stats = table_handler->m_deadlock_counter.load(); lock_wait_stats = table_handler->m_table_perf_context.m_value[PC_KEY_LOCK_WAIT_COUNT] .load(); /* Convert from rocksdb timer to mysql timer. RocksDB values are in nanoseconds, but table statistics expect the value to be in my_timer format. */ io_perf_read.svc_time = my_core::microseconds_to_my_timer( table_handler->m_io_perf_read.svc_time.load() / 1000); io_perf_read.svc_time_max = my_core::microseconds_to_my_timer( table_handler->m_io_perf_read.svc_time_max.load() / 1000); io_perf_read.wait_time = my_core::microseconds_to_my_timer( table_handler->m_io_perf_read.wait_time.load() / 1000); io_perf_read.wait_time_max = my_core::microseconds_to_my_timer( table_handler->m_io_perf_read.wait_time_max.load() / 1000); io_perf_read.slow_ios = table_handler->m_io_perf_read.slow_ios.load(); rdb_open_tables.release_table_handler(table_handler); /* Table stats expects our database and table name to be in system encoding, not filename format. Convert before calling callback. */ my_core::filename_to_tablename(dbname.c_str(), dbname_sys, sizeof(dbname_sys)); my_core::filename_to_tablename(tablename.c_str(), tablename_sys, sizeof(tablename_sys)); (*cb)(dbname_sys, tablename_sys, is_partition, &io_perf_read, &io_perf_write, &io_perf, &io_perf, &io_perf, &page_stats, &comp_stats, lock_wait_stats, lock_wait_timeout_stats, deadlock_stats, rocksdb_hton_name); } } #endif static rocksdb::Status check_rocksdb_options_compatibility( const char *const dbpath, const rocksdb::Options &main_opts, const std::vector &cf_descr) { DBUG_ASSERT(rocksdb_datadir != nullptr); rocksdb::DBOptions loaded_db_opt; std::vector loaded_cf_descs; rocksdb::Status status = LoadLatestOptions(dbpath, rocksdb::Env::Default(), &loaded_db_opt, &loaded_cf_descs, rocksdb_ignore_unknown_options); // If we're starting from scratch and there are no options saved yet then this // is a valid case. Therefore we can't compare the current set of options to // anything. if (status.IsNotFound()) { return rocksdb::Status::OK(); } if (!status.ok()) { return status; } if (loaded_cf_descs.size() != cf_descr.size()) { return rocksdb::Status::NotSupported("Mismatched size of column family " "descriptors."); } // Please see RocksDB documentation for more context about why we need to set // user-defined functions and pointer-typed options manually. for (size_t i = 0; i < loaded_cf_descs.size(); i++) { loaded_cf_descs[i].options.compaction_filter = cf_descr[i].options.compaction_filter; loaded_cf_descs[i].options.compaction_filter_factory = cf_descr[i].options.compaction_filter_factory; loaded_cf_descs[i].options.comparator = cf_descr[i].options.comparator; loaded_cf_descs[i].options.memtable_factory = cf_descr[i].options.memtable_factory; loaded_cf_descs[i].options.merge_operator = cf_descr[i].options.merge_operator; loaded_cf_descs[i].options.prefix_extractor = cf_descr[i].options.prefix_extractor; loaded_cf_descs[i].options.table_factory = cf_descr[i].options.table_factory; } // This is the essence of the function - determine if it's safe to open the // database or not. status = CheckOptionsCompatibility(dbpath, rocksdb::Env::Default(), main_opts, loaded_cf_descs, rocksdb_ignore_unknown_options); return status; } bool prevent_myrocks_loading= false; /* Storage Engine initialization function, invoked when plugin is loaded. */ static int rocksdb_init_func(void *const p) { DBUG_ENTER_FUNC(); if (prevent_myrocks_loading) { my_error(ER_INTERNAL_ERROR, MYF(0), "Loading MyRocks plugin after it has been unloaded is not " "supported. Please restart mysqld"); DBUG_RETURN(1); } if (rdb_check_rocksdb_corruption()) { sql_print_error("RocksDB: There was a corruption detected in RockDB files. " "Check error log emitted earlier for more details."); if (rocksdb_allow_to_start_after_corruption) { sql_print_information( "RocksDB: Remove rocksdb_allow_to_start_after_corruption to prevent " "server operating if RocksDB corruption is detected."); } else { sql_print_error("RocksDB: The server will exit normally and stop restart " "attempts. Remove %s file from data directory and " "start mysqld manually.", rdb_corruption_marker_file_name().c_str()); exit(0); } } // Validate the assumption about the size of ROCKSDB_SIZEOF_HIDDEN_PK_COLUMN. static_assert(sizeof(longlong) == 8, "Assuming that longlong is 8 bytes."); init_rocksdb_psi_keys(); rocksdb_hton = (handlerton *)p; mysql_mutex_init(rdb_psi_open_tbls_mutex_key, &rdb_open_tables.m_mutex, MY_MUTEX_INIT_FAST); #ifdef HAVE_PSI_INTERFACE rdb_bg_thread.init(rdb_signal_bg_psi_mutex_key, rdb_signal_bg_psi_cond_key); rdb_drop_idx_thread.init(rdb_signal_drop_idx_psi_mutex_key, rdb_signal_drop_idx_psi_cond_key); rdb_mc_thread.init(rdb_signal_mc_psi_mutex_key, rdb_signal_mc_psi_cond_key); #else rdb_bg_thread.init(); rdb_drop_idx_thread.init(); rdb_mc_thread.init(); #endif mysql_mutex_init(rdb_collation_data_mutex_key, &rdb_collation_data_mutex, MY_MUTEX_INIT_FAST); mysql_mutex_init(rdb_mem_cmp_space_mutex_key, &rdb_mem_cmp_space_mutex, MY_MUTEX_INIT_FAST); const char* initial_rocksdb_datadir_for_ignore_dirs= rocksdb_datadir; if (!strncmp(rocksdb_datadir, "./", 2)) initial_rocksdb_datadir_for_ignore_dirs += 2; ignore_db_dirs_append(initial_rocksdb_datadir_for_ignore_dirs); #if defined(HAVE_PSI_INTERFACE) rdb_collation_exceptions = new Regex_list_handler(key_rwlock_collation_exception_list); #else rdb_collation_exceptions = new Regex_list_handler(); #endif mysql_mutex_init(rdb_sysvars_psi_mutex_key, &rdb_sysvars_mutex, MY_MUTEX_INIT_FAST); mysql_mutex_init(rdb_block_cache_resize_mutex_key, &rdb_block_cache_resize_mutex, MY_MUTEX_INIT_FAST); Rdb_transaction::init_mutex(); rocksdb_hton->state = SHOW_OPTION_YES; rocksdb_hton->create = rocksdb_create_handler; rocksdb_hton->close_connection = rocksdb_close_connection; rocksdb_hton->prepare = rocksdb_prepare; rocksdb_hton->prepare_ordered = NULL; // Do not need it rocksdb_hton->commit_by_xid = rocksdb_commit_by_xid; rocksdb_hton->rollback_by_xid = rocksdb_rollback_by_xid; rocksdb_hton->recover = rocksdb_recover; rocksdb_hton->commit_ordered= rocksdb_commit_ordered; rocksdb_hton->commit = rocksdb_commit; rocksdb_hton->commit_checkpoint_request= rocksdb_checkpoint_request; rocksdb_hton->rollback = rocksdb_rollback; rocksdb_hton->show_status = rocksdb_show_status; #ifdef MARIADB_NOT_YET rocksdb_hton->explicit_snapshot = rocksdb_explicit_snapshot; #endif rocksdb_hton->start_consistent_snapshot = rocksdb_start_tx_and_assign_read_view; #ifdef MARIADB_NOT_YET rocksdb_hton->start_shared_snapshot = rocksdb_start_tx_with_shared_read_view; #endif rocksdb_hton->savepoint_set = rocksdb_savepoint; rocksdb_hton->savepoint_rollback = rocksdb_rollback_to_savepoint; rocksdb_hton->savepoint_rollback_can_release_mdl = rocksdb_rollback_to_savepoint_can_release_mdl; #ifdef MARIAROCKS_NOT_YET rocksdb_hton->update_table_stats = rocksdb_update_table_stats; #endif // MARIAROCKS_NOT_YET /* Not needed in MariaDB: rocksdb_hton->flush_logs = rocksdb_flush_wal; */ rocksdb_hton->flags = HTON_TEMPORARY_NOT_SUPPORTED | HTON_SUPPORTS_EXTENDED_KEYS | HTON_CAN_RECREATE; rocksdb_hton->tablefile_extensions= ha_rocksdb_exts; DBUG_ASSERT(!mysqld_embedded); if (rocksdb_db_options->max_open_files > (long)open_files_limit) { sql_print_information("RocksDB: rocksdb_max_open_files should not be " "greater than the open_files_limit, effective value " "of rocksdb_max_open_files is being set to " "open_files_limit / 2."); rocksdb_db_options->max_open_files = open_files_limit / 2; } else if (rocksdb_db_options->max_open_files == -2) { rocksdb_db_options->max_open_files = open_files_limit / 2; } rocksdb_stats = rocksdb::CreateDBStatistics(); rocksdb_db_options->statistics = rocksdb_stats; if (rocksdb_rate_limiter_bytes_per_sec != 0) { rocksdb_rate_limiter.reset( rocksdb::NewGenericRateLimiter(rocksdb_rate_limiter_bytes_per_sec)); rocksdb_db_options->rate_limiter = rocksdb_rate_limiter; } rocksdb_db_options->delayed_write_rate = rocksdb_delayed_write_rate; std::shared_ptr myrocks_logger = std::make_shared(); rocksdb::Status s = rocksdb::CreateLoggerFromOptions( rocksdb_datadir, *rocksdb_db_options, &rocksdb_db_options->info_log); if (s.ok()) { myrocks_logger->SetRocksDBLogger(rocksdb_db_options->info_log); } rocksdb_db_options->info_log = myrocks_logger; myrocks_logger->SetInfoLogLevel( static_cast(rocksdb_info_log_level)); rocksdb_db_options->wal_dir = rocksdb_wal_dir; rocksdb_db_options->wal_recovery_mode = static_cast(rocksdb_wal_recovery_mode); rocksdb_db_options->access_hint_on_compaction_start = static_cast( rocksdb_access_hint_on_compaction_start); if (rocksdb_db_options->allow_mmap_reads && rocksdb_db_options->use_direct_reads) { // allow_mmap_reads implies !use_direct_reads and RocksDB will not open if // mmap_reads and direct_reads are both on. (NO_LINT_DEBUG) sql_print_error("RocksDB: Can't enable both use_direct_reads " "and allow_mmap_reads\n"); DBUG_RETURN(HA_EXIT_FAILURE); } // Check whether the filesystem backing rocksdb_datadir allows O_DIRECT if (rocksdb_db_options->use_direct_reads) { rocksdb::EnvOptions soptions; rocksdb::Status check_status; rocksdb::Env *const env = rocksdb_db_options->env; std::string fname = format_string("%s/DIRECT_CHECK", rocksdb_datadir); if (env->FileExists(fname).ok()) { std::unique_ptr file; soptions.use_direct_reads = true; check_status = env->NewSequentialFile(fname, &file, soptions); } else { std::unique_ptr file; soptions.use_direct_writes = true; check_status = env->ReopenWritableFile(fname, &file, soptions); if (file != nullptr) { file->Close(); } env->DeleteFile(fname); } if (!check_status.ok()) { sql_print_error("RocksDB: Unable to use direct io in rocksdb-datadir:" "(%s)", check_status.getState()); rdb_open_tables.free_hash(); DBUG_RETURN(HA_EXIT_FAILURE); } } if (rocksdb_db_options->allow_mmap_writes && rocksdb_db_options->use_direct_io_for_flush_and_compaction) { // See above comment for allow_mmap_reads. (NO_LINT_DEBUG) sql_print_error("RocksDB: Can't enable both " "use_direct_io_for_flush_and_compaction and " "allow_mmap_writes\n"); DBUG_RETURN(HA_EXIT_FAILURE); } if (rocksdb_db_options->allow_mmap_writes && rocksdb_flush_log_at_trx_commit != FLUSH_LOG_NEVER) { // NO_LINT_DEBUG sql_print_error("RocksDB: rocksdb_flush_log_at_trx_commit needs to be 0 " "to use allow_mmap_writes"); DBUG_RETURN(HA_EXIT_FAILURE); } // sst_file_manager will move deleted rocksdb sst files to trash_dir // to be deleted in a background thread. std::string trash_dir = std::string(rocksdb_datadir) + "/trash"; rocksdb_db_options->sst_file_manager.reset(NewSstFileManager( rocksdb_db_options->env, myrocks_logger, trash_dir, rocksdb_sst_mgr_rate_bytes_per_sec, true /* delete_existing_trash */)); std::vector cf_names; rocksdb::Status status; status = rocksdb::DB::ListColumnFamilies(*rocksdb_db_options, rocksdb_datadir, &cf_names); if (!status.ok()) { /* When we start on an empty datadir, ListColumnFamilies returns IOError, and RocksDB doesn't provide any way to check what kind of error it was. Checking system errno happens to work right now. */ if (status.IsIOError() #ifndef _WIN32 && errno == ENOENT #endif ) { sql_print_information("RocksDB: Got ENOENT when listing column families"); sql_print_information( "RocksDB: assuming that we're creating a new database"); } else { rdb_log_status_error(status, "Error listing column families"); DBUG_RETURN(HA_EXIT_FAILURE); } } else sql_print_information("RocksDB: %ld column families found", cf_names.size()); std::vector cf_descr; std::vector cf_handles; rocksdb_tbl_options->index_type = (rocksdb::BlockBasedTableOptions::IndexType)rocksdb_index_type; if (!rocksdb_tbl_options->no_block_cache) { std::shared_ptr block_cache = rocksdb_use_clock_cache ? rocksdb::NewClockCache(rocksdb_block_cache_size) : rocksdb::NewLRUCache(rocksdb_block_cache_size); if (rocksdb_sim_cache_size > 0) { // Simulated cache enabled // Wrap block cache inside a simulated cache and pass it to RocksDB rocksdb_tbl_options->block_cache = rocksdb::NewSimCache(block_cache, rocksdb_sim_cache_size, 6); } else { // Pass block cache to RocksDB rocksdb_tbl_options->block_cache = block_cache; } } // Using newer BlockBasedTable format version for better compression // and better memory allocation. // See: // https://github.com/facebook/rocksdb/commit/9ab5adfc59a621d12357580c94451d9f7320c2dd rocksdb_tbl_options->format_version = 2; if (rocksdb_collect_sst_properties) { properties_collector_factory = std::make_shared(&ddl_manager); rocksdb_set_compaction_options(nullptr, nullptr, nullptr, nullptr); RDB_MUTEX_LOCK_CHECK(rdb_sysvars_mutex); DBUG_ASSERT(rocksdb_table_stats_sampling_pct <= RDB_TBL_STATS_SAMPLE_PCT_MAX); properties_collector_factory->SetTableStatsSamplingPct( rocksdb_table_stats_sampling_pct); RDB_MUTEX_UNLOCK_CHECK(rdb_sysvars_mutex); } if (rocksdb_persistent_cache_size_mb > 0) { std::shared_ptr pcache; uint64_t cache_size_bytes= rocksdb_persistent_cache_size_mb * 1024 * 1024; status = rocksdb::NewPersistentCache( rocksdb::Env::Default(), std::string(rocksdb_persistent_cache_path), cache_size_bytes, myrocks_logger, true, &pcache); if (!status.ok()) { // NO_LINT_DEBUG sql_print_error("RocksDB: Persistent cache returned error: (%s)", status.getState()); DBUG_RETURN(HA_EXIT_FAILURE); } rocksdb_tbl_options->persistent_cache = pcache; } else if (strlen(rocksdb_persistent_cache_path)) { sql_print_error("RocksDB: Must specify rocksdb_persistent_cache_size_mb"); DBUG_RETURN(HA_EXIT_FAILURE); } std::unique_ptr cf_options_map(new Rdb_cf_options()); if (!cf_options_map->init(*rocksdb_tbl_options, properties_collector_factory, rocksdb_default_cf_options, rocksdb_override_cf_options)) { // NO_LINT_DEBUG sql_print_error("RocksDB: Failed to initialize CF options map."); DBUG_RETURN(HA_EXIT_FAILURE); } /* If there are no column families, we're creating the new database. Create one column family named "default". */ if (cf_names.size() == 0) cf_names.push_back(DEFAULT_CF_NAME); std::vector compaction_enabled_cf_indices; sql_print_information("RocksDB: Column Families at start:"); for (size_t i = 0; i < cf_names.size(); ++i) { rocksdb::ColumnFamilyOptions opts; cf_options_map->get_cf_options(cf_names[i], &opts); sql_print_information(" cf=%s", cf_names[i].c_str()); sql_print_information(" write_buffer_size=%ld", opts.write_buffer_size); sql_print_information(" target_file_size_base=%" PRIu64, opts.target_file_size_base); /* Temporarily disable compactions to prevent a race condition where compaction starts before compaction filter is ready. */ if (!opts.disable_auto_compactions) { compaction_enabled_cf_indices.push_back(i); opts.disable_auto_compactions = true; } cf_descr.push_back(rocksdb::ColumnFamilyDescriptor(cf_names[i], opts)); } rocksdb::Options main_opts(*rocksdb_db_options, cf_options_map->get_defaults()); rocksdb::TransactionDBOptions tx_db_options; tx_db_options.transaction_lock_timeout = 2000; // 2 seconds tx_db_options.custom_mutex_factory = std::make_shared(); tx_db_options.write_policy = static_cast(rocksdb_write_policy); status = check_rocksdb_options_compatibility(rocksdb_datadir, main_opts, cf_descr); // We won't start if we'll determine that there's a chance of data corruption // because of incompatible options. if (!status.ok()) { rdb_log_status_error( status, "Compatibility check against existing database options failed"); DBUG_RETURN(HA_EXIT_FAILURE); } status = rocksdb::TransactionDB::Open( main_opts, tx_db_options, rocksdb_datadir, cf_descr, &cf_handles, &rdb); if (!status.ok()) { rdb_log_status_error(status, "Error opening instance"); DBUG_RETURN(HA_EXIT_FAILURE); } cf_manager.init(std::move(cf_options_map), &cf_handles); if (dict_manager.init(rdb, &cf_manager)) { // NO_LINT_DEBUG sql_print_error("RocksDB: Failed to initialize data dictionary."); DBUG_RETURN(HA_EXIT_FAILURE); } if (binlog_manager.init(&dict_manager)) { // NO_LINT_DEBUG sql_print_error("RocksDB: Failed to initialize binlog manager."); DBUG_RETURN(HA_EXIT_FAILURE); } if (ddl_manager.init(&dict_manager, &cf_manager, rocksdb_validate_tables)) { // NO_LINT_DEBUG sql_print_error("RocksDB: Failed to initialize DDL manager."); DBUG_RETURN(HA_EXIT_FAILURE); } Rdb_sst_info::init(rdb); /* Enable auto compaction, things needed for compaction filter are finished initializing */ std::vector compaction_enabled_cf_handles; compaction_enabled_cf_handles.reserve(compaction_enabled_cf_indices.size()); for (const auto &index : compaction_enabled_cf_indices) { compaction_enabled_cf_handles.push_back(cf_handles[index]); } status = rdb->EnableAutoCompaction(compaction_enabled_cf_handles); if (!status.ok()) { rdb_log_status_error(status, "Error enabling compaction"); DBUG_RETURN(HA_EXIT_FAILURE); } auto err = rdb_bg_thread.create_thread(BG_THREAD_NAME #ifdef HAVE_PSI_INTERFACE , rdb_background_psi_thread_key #endif ); if (err != 0) { sql_print_error("RocksDB: Couldn't start the background thread: (errno=%d)", err); DBUG_RETURN(HA_EXIT_FAILURE); } err = rdb_drop_idx_thread.create_thread(INDEX_THREAD_NAME #ifdef HAVE_PSI_INTERFACE , rdb_drop_idx_psi_thread_key #endif ); if (err != 0) { sql_print_error("RocksDB: Couldn't start the drop index thread: (errno=%d)", err); DBUG_RETURN(HA_EXIT_FAILURE); } err = rdb_mc_thread.create_thread(MANUAL_COMPACTION_THREAD_NAME #ifdef HAVE_PSI_INTERFACE , rdb_mc_psi_thread_key #endif ); if (err != 0) { // NO_LINT_DEBUG sql_print_error( "RocksDB: Couldn't start the manual compaction thread: (errno=%d)", err); rdb_open_tables.free_hash(); DBUG_RETURN(HA_EXIT_FAILURE); } rdb_set_collation_exception_list(rocksdb_strict_collation_exceptions); if (rocksdb_pause_background_work) { rdb->PauseBackgroundWork(); } // NO_LINT_DEBUG sql_print_information("RocksDB: global statistics using %s indexer", STRINGIFY_ARG(RDB_INDEXER)); #if defined(HAVE_SCHED_GETCPU) if (sched_getcpu() == -1) { // NO_LINT_DEBUG sql_print_information( "RocksDB: sched_getcpu() failed - " "global statistics will use thread_id_indexer_t instead"); } #endif /** Rocksdb does not always shutdown its threads, when plugin is shut down. Disable server's leak check at exit to avoid crash. */ my_disable_leak_check = true; err = my_error_register(rdb_get_error_messages, HA_ERR_ROCKSDB_FIRST, HA_ERR_ROCKSDB_LAST); if (err != 0) { // NO_LINT_DEBUG sql_print_error("RocksDB: Couldn't initialize error messages"); rdb_open_tables.m_hash.~Rdb_table_set(); DBUG_RETURN(HA_EXIT_FAILURE); } // Creating an instance of HistogramImpl should only happen after RocksDB // has been successfully initialized. commit_latency_stats = new rocksdb::HistogramImpl(); // Construct a list of directories which will be monitored by I/O watchdog // to make sure that we won't lose write access to them. std::vector directories; // 1. Data directory. directories.push_back(mysql_real_data_home); // 2. Transaction logs. if (myrocks::rocksdb_wal_dir && *myrocks::rocksdb_wal_dir) { directories.push_back(myrocks::rocksdb_wal_dir); } #if !defined(_WIN32) && !defined(__APPLE__) io_watchdog = new Rdb_io_watchdog(directories); io_watchdog->reset_timeout(rocksdb_io_write_timeout_secs); #endif // NO_LINT_DEBUG sql_print_information("MyRocks storage engine plugin has been successfully " "initialized."); DBUG_RETURN(HA_EXIT_SUCCESS); } /* Storage Engine deinitialization function, invoked when plugin is unloaded. */ static int rocksdb_done_func(void *const p) { DBUG_ENTER_FUNC(); int error = 0; // signal the drop index thread to stop rdb_drop_idx_thread.signal(true); // Flush all memtables for not losing data, even if WAL is disabled. rocksdb_flush_all_memtables(); // Stop all rocksdb background work CancelAllBackgroundWork(rdb->GetBaseDB(), true); // Signal the background thread to stop and to persist all stats collected // from background flushes and compactions. This will add more keys to a new // memtable, but since the memtables were just flushed, it should not trigger // a flush that can stall due to background threads being stopped. As long // as these keys are stored in a WAL file, they can be retrieved on restart. rdb_bg_thread.signal(true); // Wait for the background thread to finish. auto err = rdb_bg_thread.join(); if (err != 0) { // We'll log the message and continue because we're shutting down and // continuation is the optimal strategy. // NO_LINT_DEBUG sql_print_error("RocksDB: Couldn't stop the background thread: (errno=%d)", err); } // Wait for the drop index thread to finish. err = rdb_drop_idx_thread.join(); if (err != 0) { // NO_LINT_DEBUG sql_print_error("RocksDB: Couldn't stop the index thread: (errno=%d)", err); } // signal the manual compaction thread to stop rdb_mc_thread.signal(true); // Wait for the manual compaction thread to finish. err = rdb_mc_thread.join(); if (err != 0) { // NO_LINT_DEBUG sql_print_error( "RocksDB: Couldn't stop the manual compaction thread: (errno=%d)", err); } if (rdb_open_tables.m_hash.size()) { // Looks like we are getting unloaded and yet we have some open tables // left behind. error = 1; } /* MariaDB: When the plugin is unloaded with UNINSTALL SONAME command, some connections may still have Rdb_transaction objects. These objects are not genuine transactions (as SQL layer makes sure that a plugin that is being unloaded has no open tables), they are empty Rdb_transaction objects that were left there to save on object creation/deletion. Go through the list and delete them. */ { class Rdb_trx_deleter: public Rdb_tx_list_walker { public: std::set rdb_trxs; void process_tran(const Rdb_transaction *const tx) override { /* Check if the transaction is really empty. We only check non-WriteBatch-based transactions, because there is no easy way to check WriteBatch-based transactions. */ if (!tx->is_writebatch_trx()) { const auto tx_impl = static_cast(tx); DBUG_ASSERT(tx_impl); if (tx_impl->get_rdb_trx()) DBUG_ASSERT(0); } rdb_trxs.insert((Rdb_transaction*)tx); }; } deleter; Rdb_transaction::walk_tx_list(&deleter); for (std::set::iterator it= deleter.rdb_trxs.begin(); it != deleter.rdb_trxs.end(); ++it) { // When a transaction is deleted, it removes itself from s_tx_list. delete *it; } } /* destructors for static objects can be called at _exit(), but we want to free the memory at dlclose() */ rdb_open_tables.m_hash.~Rdb_table_set(); mysql_mutex_destroy(&rdb_open_tables.m_mutex); mysql_mutex_destroy(&rdb_sysvars_mutex); mysql_mutex_destroy(&rdb_block_cache_resize_mutex); delete rdb_collation_exceptions; mysql_mutex_destroy(&rdb_collation_data_mutex); mysql_mutex_destroy(&rdb_mem_cmp_space_mutex); Rdb_transaction::term_mutex(); for (auto &it : rdb_collation_data) { delete it; it = nullptr; } ddl_manager.cleanup(); binlog_manager.cleanup(); dict_manager.cleanup(); cf_manager.cleanup(); delete rdb; rdb = nullptr; delete commit_latency_stats; commit_latency_stats = nullptr; #if !defined(_WIN32) && !defined(__APPLE__) delete io_watchdog; io_watchdog = nullptr; #endif // Disown the cache data since we're shutting down. // This results in memory leaks but it improved the shutdown time. // Don't disown when running under valgrind #ifndef HAVE_purify if (rocksdb_tbl_options->block_cache) { rocksdb_tbl_options->block_cache->DisownData(); } #endif /* HAVE_purify */ /* MariaDB: don't clear rocksdb_db_options and rocksdb_tbl_options. MyRocks' plugin variables refer to them. The plugin cannot be loaded again (see prevent_myrocks_loading) but plugin variables are processed before myrocks::rocksdb_init_func is invoked, so they must point to valid memory. */ //rocksdb_db_options = nullptr; rocksdb_db_options->statistics = nullptr; //rocksdb_tbl_options = nullptr; rocksdb_stats = nullptr; my_free(rocksdb_update_cf_options); rocksdb_update_cf_options = nullptr; my_error_unregister(HA_ERR_ROCKSDB_FIRST, HA_ERR_ROCKSDB_LAST); /* Prevent loading the plugin after it has been loaded and then unloaded. This doesn't work currently. */ prevent_myrocks_loading= true; DBUG_RETURN(error); } static inline void rocksdb_smart_seek(bool seek_backward, rocksdb::Iterator *const iter, const rocksdb::Slice &key_slice) { if (seek_backward) { iter->SeekForPrev(key_slice); } else { iter->Seek(key_slice); } } static inline void rocksdb_smart_next(bool seek_backward, rocksdb::Iterator *const iter) { if (seek_backward) { iter->Prev(); } else { iter->Next(); } } #ifndef NDEBUG // simulate that RocksDB has reported corrupted data static void dbug_change_status_to_corrupted(rocksdb::Status *status) { *status = rocksdb::Status::Corruption(); } #endif // If the iterator is not valid it might be because of EOF but might be due // to IOError or corruption. The good practice is always check it. // https://github.com/facebook/rocksdb/wiki/Iterator#error-handling static inline bool is_valid(rocksdb::Iterator *scan_it) { if (scan_it->Valid()) { return true; } else { rocksdb::Status s = scan_it->status(); DBUG_EXECUTE_IF("rocksdb_return_status_corrupted", dbug_change_status_to_corrupted(&s);); if (s.IsIOError() || s.IsCorruption()) { if (s.IsCorruption()) { rdb_persist_corruption_marker(); } rdb_handle_io_error(s, RDB_IO_ERROR_GENERAL); } return false; } } /** @brief Example of simple lock controls. The "table_handler" it creates is a structure we will pass to each ha_rocksdb handler. Do you have to have one of these? Well, you have pieces that are used for locking, and they are needed to function. */ Rdb_table_handler * Rdb_open_tables_map::get_table_handler(const char *const table_name) { Rdb_table_handler *table_handler; uint length; char *tmp_name; DBUG_ASSERT(table_name != nullptr); length = (uint)strlen(table_name); // First, look up the table in the hash map. RDB_MUTEX_LOCK_CHECK(m_mutex); if (!m_hash.size() || !(table_handler = m_hash.find(table_name, length))) { // Since we did not find it in the hash map, attempt to create and add it // to the hash map. if (!(table_handler = reinterpret_cast(my_multi_malloc( MYF(MY_WME | MY_ZEROFILL), &table_handler, sizeof(*table_handler), &tmp_name, length + 1, NullS)))) { // Allocating a new Rdb_table_handler and a new table name failed. RDB_MUTEX_UNLOCK_CHECK(m_mutex); return nullptr; } table_handler->m_ref_count = 0; table_handler->m_table_name_length = length; table_handler->m_table_name = tmp_name; strmov(table_handler->m_table_name, table_name); if (m_hash.insert(table_handler)) { // Inserting into the hash map failed. RDB_MUTEX_UNLOCK_CHECK(m_mutex); my_free(table_handler); return nullptr; } thr_lock_init(&table_handler->m_thr_lock); #ifdef MARIAROCKS_NOT_YET table_handler->m_io_perf_read.init(); table_handler->m_io_perf_write.init(); #endif } DBUG_ASSERT(table_handler->m_ref_count >= 0); table_handler->m_ref_count++; RDB_MUTEX_UNLOCK_CHECK(m_mutex); return table_handler; } std::vector rdb_get_open_table_names(void) { return rdb_open_tables.get_table_names(); } std::vector Rdb_open_tables_map::get_table_names(void) const { size_t i; const Rdb_table_handler *table_handler; std::vector names; RDB_MUTEX_LOCK_CHECK(m_mutex); for (i = 0; (table_handler = m_hash.at(i)); i++) { DBUG_ASSERT(table_handler != nullptr); names.push_back(table_handler->m_table_name); } DBUG_ASSERT(i == m_hash.size()); RDB_MUTEX_UNLOCK_CHECK(m_mutex); return names; } /* Inspired by innobase_get_int_col_max_value from InnoDB. This returns the maximum value a type can take on. */ static ulonglong rdb_get_int_col_max_value(const Field *field) { ulonglong max_value = 0; switch (field->key_type()) { case HA_KEYTYPE_BINARY: max_value = 0xFFULL; break; case HA_KEYTYPE_INT8: max_value = 0x7FULL; break; case HA_KEYTYPE_USHORT_INT: max_value = 0xFFFFULL; break; case HA_KEYTYPE_SHORT_INT: max_value = 0x7FFFULL; break; case HA_KEYTYPE_UINT24: max_value = 0xFFFFFFULL; break; case HA_KEYTYPE_INT24: max_value = 0x7FFFFFULL; break; case HA_KEYTYPE_ULONG_INT: max_value = 0xFFFFFFFFULL; break; case HA_KEYTYPE_LONG_INT: max_value = 0x7FFFFFFFULL; break; case HA_KEYTYPE_ULONGLONG: max_value = 0xFFFFFFFFFFFFFFFFULL; break; case HA_KEYTYPE_LONGLONG: max_value = 0x7FFFFFFFFFFFFFFFULL; break; case HA_KEYTYPE_FLOAT: max_value = 0x1000000ULL; break; case HA_KEYTYPE_DOUBLE: max_value = 0x20000000000000ULL; break; default: abort(); } return max_value; } void ha_rocksdb::load_auto_incr_value() { ulonglong auto_incr = 0; bool validate_last = false, use_datadic = true; #ifndef NDEBUG DBUG_EXECUTE_IF("myrocks_autoinc_upgrade", use_datadic = false;); validate_last = true; #endif if (use_datadic && dict_manager.get_auto_incr_val( m_tbl_def->get_autoincr_gl_index_id(), &auto_incr)) { update_auto_incr_val(auto_incr); } // If we find nothing in the data dictionary, or if we are in debug mode, // then call index_last to get the last value. // // This is needed when upgrading from a server that did not support // persistent auto_increment, of if the table is empty. // // For debug mode, we are just verifying that the data dictionary value is // greater than or equal to the maximum value in the table. if (auto_incr == 0 || validate_last) { auto_incr = load_auto_incr_value_from_index(); update_auto_incr_val(auto_incr); } // If we failed to find anything from the data dictionary and index, then // initialize auto_increment to 1. if (m_tbl_def->m_auto_incr_val == 0) { update_auto_incr_val(1); } } ulonglong ha_rocksdb::load_auto_incr_value_from_index() { const int save_active_index = active_index; active_index = table->s->next_number_index; const uint8 save_table_status = table->status; ulonglong last_val = 0; Rdb_transaction *const tx = get_or_create_tx(table->in_use); const bool is_new_snapshot = !tx->has_snapshot(); if (is_new_snapshot) { tx->acquire_snapshot(true); } // Do a lookup. We only need index column, so it should be index-only. // (another reason to make it index-only is that table->read_set is not set // appropriately and non-index-only lookup will not read the value) const bool save_keyread_only = m_keyread_only; m_keyread_only = true; m_key_requested = true; if (!index_last(table->record[0])) { Field *field = table->key_info[table->s->next_number_index].key_part[0].field; ulonglong max_val = rdb_get_int_col_max_value(field); my_bitmap_map *const old_map = dbug_tmp_use_all_columns(table, table->read_set); last_val = field->val_int(); if (last_val != max_val) { last_val++; } #ifndef NDEBUG ulonglong dd_val; if (last_val <= max_val) { const auto &gl_index_id = m_tbl_def->get_autoincr_gl_index_id(); if (dict_manager.get_auto_incr_val(gl_index_id, &dd_val) && tx->get_auto_incr(gl_index_id) == 0) { DBUG_ASSERT(dd_val >= last_val); } } #endif dbug_tmp_restore_column_map(table->read_set, old_map); } m_keyread_only = save_keyread_only; if (is_new_snapshot) { tx->release_snapshot(); } table->status = save_table_status; active_index = save_active_index; /* Do what ha_rocksdb::index_end() does. (Why don't we use index_init/index_end? class handler defines index_init as private, for some reason). */ release_scan_iterator(); return last_val; } void ha_rocksdb::update_auto_incr_val(ulonglong val) { ulonglong auto_incr_val = m_tbl_def->m_auto_incr_val; while ( auto_incr_val < val && !m_tbl_def->m_auto_incr_val.compare_exchange_weak(auto_incr_val, val)) { // Do nothing - just loop until auto_incr_val is >= val or we successfully // set it } } void ha_rocksdb::update_auto_incr_val_from_field() { Field *field; ulonglong new_val, max_val; field = table->key_info[table->s->next_number_index].key_part[0].field; max_val = rdb_get_int_col_max_value(field); my_bitmap_map *const old_map = dbug_tmp_use_all_columns(table, table->read_set); new_val = field->val_int(); // don't increment if we would wrap around if (new_val != max_val) { new_val++; } dbug_tmp_restore_column_map(table->read_set, old_map); // Only update if positive value was set for auto_incr column. if (new_val <= max_val) { Rdb_transaction *const tx = get_or_create_tx(table->in_use); tx->set_auto_incr(m_tbl_def->get_autoincr_gl_index_id(), new_val); // Update the in memory auto_incr value in m_tbl_def. update_auto_incr_val(new_val); } } int ha_rocksdb::load_hidden_pk_value() { const int save_active_index = active_index; active_index = m_tbl_def->m_key_count - 1; const uint8 save_table_status = table->status; Rdb_transaction *const tx = get_or_create_tx(table->in_use); const bool is_new_snapshot = !tx->has_snapshot(); longlong hidden_pk_id = 1; // Do a lookup. if (!index_last(table->record[0])) { /* Decode PK field from the key */ auto err = read_hidden_pk_id_from_rowkey(&hidden_pk_id); if (err) { if (is_new_snapshot) { tx->release_snapshot(); } return err; } hidden_pk_id++; } longlong old = m_tbl_def->m_hidden_pk_val; while (old < hidden_pk_id && !m_tbl_def->m_hidden_pk_val.compare_exchange_weak(old, hidden_pk_id)) { } if (is_new_snapshot) { tx->release_snapshot(); } table->status = save_table_status; active_index = save_active_index; release_scan_iterator(); return HA_EXIT_SUCCESS; } /* Get PK value from m_tbl_def->m_hidden_pk_info. */ longlong ha_rocksdb::update_hidden_pk_val() { DBUG_ASSERT(has_hidden_pk(table)); const longlong new_val = m_tbl_def->m_hidden_pk_val++; return new_val; } /* Get the id of the hidden pk id from m_last_rowkey */ int ha_rocksdb::read_hidden_pk_id_from_rowkey(longlong *const hidden_pk_id) { DBUG_ASSERT(table != nullptr); DBUG_ASSERT(has_hidden_pk(table)); rocksdb::Slice rowkey_slice(m_last_rowkey.ptr(), m_last_rowkey.length()); // Get hidden primary key from old key slice Rdb_string_reader reader(&rowkey_slice); if ((!reader.read(Rdb_key_def::INDEX_NUMBER_SIZE))) return HA_ERR_ROCKSDB_CORRUPT_DATA; const int length= 8; /* was Field_longlong::PACK_LENGTH in FB MySQL tree */ const uchar *from = reinterpret_cast(reader.read(length)); if (from == nullptr) { /* Mem-comparable image doesn't have enough bytes */ return HA_ERR_ROCKSDB_CORRUPT_DATA; } *hidden_pk_id = rdb_netbuf_read_uint64(&from); return HA_EXIT_SUCCESS; } /** @brief Free lock controls. We call this whenever we close a table. If the table had the last reference to the table_handler, then we free the memory associated with it. */ void Rdb_open_tables_map::release_table_handler( Rdb_table_handler *const table_handler) { RDB_MUTEX_LOCK_CHECK(m_mutex); DBUG_ASSERT(table_handler != nullptr); DBUG_ASSERT(table_handler->m_ref_count > 0); if (!--table_handler->m_ref_count) { // Last reference was released. Tear down the hash entry. const auto ret MY_ATTRIBUTE((__unused__)) = m_hash.remove(table_handler); DBUG_ASSERT(!ret); // the hash entry must actually be found and deleted my_core::thr_lock_delete(&table_handler->m_thr_lock); my_free(table_handler); } RDB_MUTEX_UNLOCK_CHECK(m_mutex); } static handler *rocksdb_create_handler(my_core::handlerton *const hton, my_core::TABLE_SHARE *const table_arg, my_core::MEM_ROOT *const mem_root) { return new (mem_root) ha_rocksdb(hton, table_arg); } ha_rocksdb::ha_rocksdb(my_core::handlerton *const hton, my_core::TABLE_SHARE *const table_arg) : handler(hton, table_arg), m_table_handler(nullptr), m_scan_it(nullptr), m_scan_it_skips_bloom(false), m_scan_it_snapshot(nullptr), m_scan_it_lower_bound(nullptr), m_scan_it_upper_bound(nullptr), m_tbl_def(nullptr), m_pk_descr(nullptr), m_key_descr_arr(nullptr), m_pk_can_be_decoded(false), m_maybe_unpack_info(false), m_pk_tuple(nullptr), m_pk_packed_tuple(nullptr), m_sk_packed_tuple(nullptr), m_end_key_packed_tuple(nullptr), m_sk_match_prefix(nullptr), m_sk_match_prefix_buf(nullptr), m_sk_packed_tuple_old(nullptr), m_dup_sk_packed_tuple(nullptr), m_dup_sk_packed_tuple_old(nullptr), m_pack_buffer(nullptr), m_lock_rows(RDB_LOCK_NONE), m_keyread_only(FALSE), m_encoder_arr(nullptr), m_row_checksums_checked(0), m_in_rpl_delete_rows(false), m_in_rpl_update_rows(false), m_force_skip_unique_check(false) {} const std::string &ha_rocksdb::get_table_basename() const { return m_tbl_def->base_tablename(); } /** @return false OK other Error inpacking the data */ bool ha_rocksdb::init_with_fields() { DBUG_ENTER_FUNC(); const uint pk = table_share->primary_key; if (pk != MAX_KEY) { const uint key_parts = table_share->key_info[pk].user_defined_key_parts; check_keyread_allowed(pk /*PK*/, key_parts - 1, true); } else m_pk_can_be_decoded = false; cached_table_flags = table_flags(); DBUG_RETURN(false); /* Ok */ } /* If the key is a TTL key, we may need to filter it out. The purpose of read filtering for tables with TTL is to ensure that during a transaction a key which has expired already but not removed by compaction yet is not returned to the user. Without this the user might be hit with problems such as disappearing rows within a transaction, etc, because the compaction filter ignores snapshots when filtering keys. */ bool ha_rocksdb::should_hide_ttl_rec(const Rdb_key_def &kd, const rocksdb::Slice &ttl_rec_val, const int64_t curr_ts) { DBUG_ASSERT(kd.has_ttl()); DBUG_ASSERT(kd.m_ttl_rec_offset != UINT_MAX); /* Curr_ts can only be 0 if there are no snapshots open. should_hide_ttl_rec can only be called when there is >=1 snapshots, unless we are filtering on the write path (single INSERT/UPDATE) in which case we are passed in the current time as curr_ts. In the event curr_ts is 0, we always decide not to filter the record. We also log a warning and increment a diagnostic counter. */ if (curr_ts == 0) { update_row_stats(ROWS_HIDDEN_NO_SNAPSHOT); return false; } if (!rdb_is_ttl_read_filtering_enabled() || !rdb_is_ttl_enabled()) { return false; } Rdb_string_reader reader(&ttl_rec_val); /* Find where the 8-byte ttl is for each record in this index. */ uint64 ts; if (!reader.read(kd.m_ttl_rec_offset) || reader.read_uint64(&ts)) { /* This condition should never be reached since all TTL records have an 8 byte ttl field in front. Don't filter the record out, and log an error. */ std::string buf; buf = rdb_hexdump(ttl_rec_val.data(), ttl_rec_val.size(), RDB_MAX_HEXDUMP_LEN); const GL_INDEX_ID gl_index_id = kd.get_gl_index_id(); // NO_LINT_DEBUG sql_print_error("Decoding ttl from PK value failed, " "for index (%u,%u), val: %s", gl_index_id.cf_id, gl_index_id.index_id, buf.c_str()); DBUG_ASSERT(0); return false; } /* Hide record if it has expired before the current snapshot time. */ uint64 read_filter_ts = 0; #ifndef NDEBUG read_filter_ts += rdb_dbug_set_ttl_read_filter_ts(); #endif bool is_hide_ttl = ts + kd.m_ttl_duration + read_filter_ts <= static_cast(curr_ts); if (is_hide_ttl) { update_row_stats(ROWS_FILTERED); } return is_hide_ttl; } void ha_rocksdb::rocksdb_skip_expired_records(const Rdb_key_def &kd, rocksdb::Iterator *const iter, bool seek_backward) { if (kd.has_ttl()) { while (iter->Valid() && should_hide_ttl_rec( kd, iter->value(), get_or_create_tx(table->in_use)->m_snapshot_timestamp)) { rocksdb_smart_next(seek_backward, iter); } } } /** Convert record from table->record[0] form into a form that can be written into rocksdb. @param pk_packed_slice Packed PK tuple. We need it in order to compute and store its CRC. @param packed_rec OUT Data slice with record data. */ int ha_rocksdb::convert_record_to_storage_format( const struct update_row_info &row_info, rocksdb::Slice *const packed_rec) { DBUG_ASSERT_IMP(m_maybe_unpack_info, row_info.new_pk_unpack_info); DBUG_ASSERT(m_pk_descr != nullptr); const rocksdb::Slice &pk_packed_slice = row_info.new_pk_slice; Rdb_string_writer *const pk_unpack_info = row_info.new_pk_unpack_info; bool has_ttl = m_pk_descr->has_ttl(); bool has_ttl_column = !m_pk_descr->m_ttl_column.empty(); bool ttl_in_pk = has_ttl_column && (row_info.ttl_pk_offset != UINT_MAX); m_storage_record.length(0); if (has_ttl) { /* If it's a TTL record, reserve space for 8 byte TTL value in front. */ m_storage_record.fill(ROCKSDB_SIZEOF_TTL_RECORD + m_null_bytes_in_rec, 0); m_ttl_bytes_updated = false; /* If the TTL is contained within the key, we use the offset to find the TTL value and place it in the beginning of the value record. */ if (ttl_in_pk) { Rdb_string_reader reader(&pk_packed_slice); const char *ts; if (!reader.read(row_info.ttl_pk_offset) || !(ts = reader.read(ROCKSDB_SIZEOF_TTL_RECORD))) { std::string buf; buf = rdb_hexdump(pk_packed_slice.data(), pk_packed_slice.size(), RDB_MAX_HEXDUMP_LEN); const GL_INDEX_ID gl_index_id = m_pk_descr->get_gl_index_id(); // NO_LINT_DEBUG sql_print_error("Decoding ttl from PK failed during insert, " "for index (%u,%u), key: %s", gl_index_id.cf_id, gl_index_id.index_id, buf.c_str()); return HA_EXIT_FAILURE; } char *const data = const_cast(m_storage_record.ptr()); memcpy(data, ts, ROCKSDB_SIZEOF_TTL_RECORD); #ifndef NDEBUG // Adjust for test case if needed rdb_netbuf_store_uint64( reinterpret_cast(data), rdb_netbuf_to_uint64(reinterpret_cast(data)) + rdb_dbug_set_ttl_rec_ts()); #endif // Also store in m_ttl_bytes to propagate to update_sk memcpy(m_ttl_bytes, data, ROCKSDB_SIZEOF_TTL_RECORD); } else if (!has_ttl_column) { /* For implicitly generated TTL records we need to copy over the old TTL value from the old record in the event of an update. It was stored in m_ttl_bytes. Otherwise, generate a timestamp using the current time. */ if (!row_info.old_pk_slice.empty()) { char *const data = const_cast(m_storage_record.ptr()); memcpy(data, m_ttl_bytes, sizeof(uint64)); } else { uint64 ts = static_cast(std::time(nullptr)); #ifndef NDEBUG ts += rdb_dbug_set_ttl_rec_ts(); #endif char *const data = const_cast(m_storage_record.ptr()); rdb_netbuf_store_uint64(reinterpret_cast(data), ts); // Also store in m_ttl_bytes to propagate to update_sk memcpy(m_ttl_bytes, data, ROCKSDB_SIZEOF_TTL_RECORD); } } } else { /* All NULL bits are initially 0 */ m_storage_record.fill(m_null_bytes_in_rec, 0); } // If a primary key may have non-empty unpack_info for certain values, // (m_maybe_unpack_info=TRUE), we write the unpack_info block. The block // itself was prepared in Rdb_key_def::pack_record. if (m_maybe_unpack_info) { m_storage_record.append(reinterpret_cast(pk_unpack_info->ptr()), pk_unpack_info->get_current_pos()); } for (uint i = 0; i < table->s->fields; i++) { /* Don't pack decodable PK key parts */ if (m_encoder_arr[i].m_storage_type != Rdb_field_encoder::STORE_ALL) { continue; } Field *const field = table->field[i]; if (m_encoder_arr[i].maybe_null()) { char *data = const_cast(m_storage_record.ptr()); if (has_ttl) { data += ROCKSDB_SIZEOF_TTL_RECORD; } if (field->is_null()) { data[m_encoder_arr[i].m_null_offset] |= m_encoder_arr[i].m_null_mask; /* Don't write anything for NULL values */ continue; } } if (m_encoder_arr[i].m_field_type == MYSQL_TYPE_BLOB) { my_core::Field_blob *blob = (my_core::Field_blob *)field; /* Get the number of bytes needed to store length*/ const uint length_bytes = blob->pack_length() - portable_sizeof_char_ptr; /* Store the length of the value */ m_storage_record.append(reinterpret_cast(blob->ptr), length_bytes); /* Store the blob value itself */ char *data_ptr; memcpy(&data_ptr, blob->ptr + length_bytes, sizeof(uchar **)); m_storage_record.append(data_ptr, blob->get_length()); } else if (m_encoder_arr[i].m_field_type == MYSQL_TYPE_VARCHAR) { Field_varstring *const field_var = (Field_varstring *)field; uint data_len; /* field_var->length_bytes is 1 or 2 */ if (field_var->length_bytes == 1) { data_len = field_var->ptr[0]; } else { DBUG_ASSERT(field_var->length_bytes == 2); data_len = uint2korr(field_var->ptr); } m_storage_record.append(reinterpret_cast(field_var->ptr), field_var->length_bytes + data_len); } else { /* Copy the field data */ const uint len = field->pack_length_in_rec(); m_storage_record.append(reinterpret_cast(field->ptr), len); /* Check if this is the TTL field within the table, if so store the TTL in the front of the record as well here. */ if (has_ttl && has_ttl_column && i == m_pk_descr->get_ttl_field_offset()) { DBUG_ASSERT(len == ROCKSDB_SIZEOF_TTL_RECORD); DBUG_ASSERT(field->real_type() == MYSQL_TYPE_LONGLONG); DBUG_ASSERT(m_pk_descr->get_ttl_field_offset() != UINT_MAX); char *const data = const_cast(m_storage_record.ptr()); uint64 ts = uint8korr(field->ptr); #ifndef NDEBUG ts += rdb_dbug_set_ttl_rec_ts(); #endif rdb_netbuf_store_uint64(reinterpret_cast(data), ts); // If this is an update and the timestamp has been updated, take note // so we can avoid updating SKs unnecessarily. if (!row_info.old_pk_slice.empty()) { m_ttl_bytes_updated = memcmp(m_ttl_bytes, data, ROCKSDB_SIZEOF_TTL_RECORD); } // Store timestamp in m_ttl_bytes to propagate to update_sk memcpy(m_ttl_bytes, data, ROCKSDB_SIZEOF_TTL_RECORD); } } } if (should_store_row_debug_checksums()) { const uint32_t key_crc32 = my_core::crc32( 0, rdb_slice_to_uchar_ptr(&pk_packed_slice), pk_packed_slice.size()); const uint32_t val_crc32 = my_core::crc32(0, rdb_mysql_str_to_uchar_str(&m_storage_record), m_storage_record.length()); uchar key_crc_buf[RDB_CHECKSUM_SIZE]; uchar val_crc_buf[RDB_CHECKSUM_SIZE]; rdb_netbuf_store_uint32(key_crc_buf, key_crc32); rdb_netbuf_store_uint32(val_crc_buf, val_crc32); m_storage_record.append((const char *)&RDB_CHECKSUM_DATA_TAG, 1); m_storage_record.append((const char *)key_crc_buf, RDB_CHECKSUM_SIZE); m_storage_record.append((const char *)val_crc_buf, RDB_CHECKSUM_SIZE); } *packed_rec = rocksdb::Slice(m_storage_record.ptr(), m_storage_record.length()); return HA_EXIT_SUCCESS; } /* @brief Setup which fields will be unpacked when reading rows @detail Three special cases when we still unpack all fields: - When this table is being updated (m_lock_rows==RDB_LOCK_WRITE). - When @@rocksdb_verify_row_debug_checksums is ON (In this mode, we need to read all fields to find whether there is a row checksum at the end. We could skip the fields instead of decoding them, but currently we do decoding.) - On index merge as bitmap is cleared during that operation @seealso ha_rocksdb::setup_field_converters() ha_rocksdb::convert_record_from_storage_format() */ void ha_rocksdb::setup_read_decoders() { m_decoders_vect.clear(); m_key_requested = false; int last_useful = 0; int skip_size = 0; for (uint i = 0; i < table->s->fields; i++) { // bitmap is cleared on index merge, but it still needs to decode columns const bool field_requested = m_lock_rows == RDB_LOCK_WRITE || m_verify_row_debug_checksums || bitmap_is_clear_all(table->read_set) || bitmap_is_set(table->read_set, table->field[i]->field_index); // We only need the decoder if the whole record is stored. if (m_encoder_arr[i].m_storage_type != Rdb_field_encoder::STORE_ALL) { // the field potentially needs unpacking if (field_requested) { // the field is in the read set m_key_requested = true; } continue; } if (field_requested) { // We will need to decode this field m_decoders_vect.push_back({&m_encoder_arr[i], true, skip_size}); last_useful = m_decoders_vect.size(); skip_size = 0; } else { if (m_encoder_arr[i].uses_variable_len_encoding() || m_encoder_arr[i].maybe_null()) { // For variable-length field, we need to read the data and skip it m_decoders_vect.push_back({&m_encoder_arr[i], false, skip_size}); skip_size = 0; } else { // Fixed-width field can be skipped without looking at it. // Add appropriate skip_size to the next field. skip_size += m_encoder_arr[i].m_pack_length_in_rec; } } } // It could be that the last few elements are varchars that just do // skipping. Remove them. m_decoders_vect.erase(m_decoders_vect.begin() + last_useful, m_decoders_vect.end()); } #ifndef NDEBUG void dbug_append_garbage_at_end(rocksdb::PinnableSlice *on_disk_rec) { std::string str(on_disk_rec->data(), on_disk_rec->size()); on_disk_rec->Reset(); str.append("abc"); on_disk_rec->PinSelf(rocksdb::Slice(str)); } void dbug_truncate_record(rocksdb::PinnableSlice *on_disk_rec) { on_disk_rec->remove_suffix(on_disk_rec->size()); } void dbug_modify_rec_varchar12(rocksdb::PinnableSlice *on_disk_rec) { std::string res; // The record is NULL-byte followed by VARCHAR(10). // Put the NULL-byte res.append("\0", 1); // Then, add a valid VARCHAR(12) value. res.append("\xC", 1); res.append("123456789ab", 12); on_disk_rec->Reset(); on_disk_rec->PinSelf(rocksdb::Slice(res)); } void dbug_modify_key_varchar8(String &on_disk_rec) { std::string res; // The key starts with index number res.append(on_disk_rec.ptr(), Rdb_key_def::INDEX_NUMBER_SIZE); // Then, a mem-comparable form of a varchar(8) value. res.append("ABCDE\0\0\0\xFC", 9); on_disk_rec.length(0); on_disk_rec.append(res.data(), res.size()); } void dbug_create_err_inplace_alter() { my_printf_error(ER_UNKNOWN_ERROR, "Intentional failure in inplace alter occurred.", MYF(0)); } #endif int ha_rocksdb::convert_record_from_storage_format( const rocksdb::Slice *const key, uchar *const buf) { DBUG_EXECUTE_IF("myrocks_simulate_bad_row_read1", dbug_append_garbage_at_end(&m_retrieved_record);); DBUG_EXECUTE_IF("myrocks_simulate_bad_row_read2", dbug_truncate_record(&m_retrieved_record);); DBUG_EXECUTE_IF("myrocks_simulate_bad_row_read3", dbug_modify_rec_varchar12(&m_retrieved_record);); return convert_record_from_storage_format(key, &m_retrieved_record, buf); } int ha_rocksdb::convert_blob_from_storage_format( my_core::Field_blob *const blob, Rdb_string_reader *const reader, bool decode) { /* Get the number of bytes needed to store length*/ const uint length_bytes = blob->pack_length() - portable_sizeof_char_ptr; const char *data_len_str; if (!(data_len_str = reader->read(length_bytes))) { return HA_ERR_ROCKSDB_CORRUPT_DATA; } memcpy(blob->ptr, data_len_str, length_bytes); const uint32 data_len = blob->get_length( reinterpret_cast(data_len_str), length_bytes); const char *blob_ptr; if (!(blob_ptr = reader->read(data_len))) { return HA_ERR_ROCKSDB_CORRUPT_DATA; } if (decode) { // set 8-byte pointer to 0, like innodb does (relevant for 32-bit // platforms) memset(blob->ptr + length_bytes, 0, 8); memcpy(blob->ptr + length_bytes, &blob_ptr, sizeof(uchar **)); } return HA_EXIT_SUCCESS; } int ha_rocksdb::convert_varchar_from_storage_format( my_core::Field_varstring *const field_var, Rdb_string_reader *const reader, bool decode) { const char *data_len_str; if (!(data_len_str = reader->read(field_var->length_bytes))) return HA_ERR_ROCKSDB_CORRUPT_DATA; uint data_len; /* field_var->length_bytes is 1 or 2 */ if (field_var->length_bytes == 1) { data_len = (uchar)data_len_str[0]; } else { DBUG_ASSERT(field_var->length_bytes == 2); data_len = uint2korr(data_len_str); } if (data_len > field_var->field_length) { /* The data on disk is longer than table DDL allows? */ return HA_ERR_ROCKSDB_CORRUPT_DATA; } if (!reader->read(data_len)) { return HA_ERR_ROCKSDB_CORRUPT_DATA; } if (decode) { memcpy(field_var->ptr, data_len_str, field_var->length_bytes + data_len); } return HA_EXIT_SUCCESS; } int ha_rocksdb::convert_field_from_storage_format( my_core::Field *const field, Rdb_string_reader *const reader, bool decode, uint len) { const char *data_bytes; if (len > 0) { if ((data_bytes = reader->read(len)) == nullptr) { return HA_ERR_ROCKSDB_CORRUPT_DATA; } if (decode) memcpy(field->ptr, data_bytes, len); } return HA_EXIT_SUCCESS; } /* @brief Unpack the record in this->m_retrieved_record and this->m_last_rowkey from storage format into buf (which can be table->record[0] or table->record[1]). @param key Table record's key in mem-comparable form. @param buf Store record in table->record[0] format here @detail If the table has blobs, the unpacked data in buf may keep pointers to the data in this->m_retrieved_record. The key is only needed to check its checksum value (the checksum is in m_retrieved_record). @seealso ha_rocksdb::setup_read_decoders() Sets up data structures which tell which columns to decode. @return 0 OK other Error inpacking the data */ int ha_rocksdb::convert_record_from_storage_format( const rocksdb::Slice *const key, const rocksdb::Slice *const value, uchar *const buf) { Rdb_string_reader reader(value); /* Decode PK fields from the key */ DBUG_EXECUTE_IF("myrocks_simulate_bad_pk_read1", dbug_modify_key_varchar8(m_last_rowkey);); const rocksdb::Slice rowkey_slice(m_last_rowkey.ptr(), m_last_rowkey.length()); const char *unpack_info = nullptr; uint16 unpack_info_len = 0; rocksdb::Slice unpack_slice; /* If it's a TTL record, skip the 8 byte TTL value */ const char *ttl_bytes; if (m_pk_descr->has_ttl()) { if ((ttl_bytes = reader.read(ROCKSDB_SIZEOF_TTL_RECORD))) { memcpy(m_ttl_bytes, ttl_bytes, ROCKSDB_SIZEOF_TTL_RECORD); } else { return HA_ERR_ROCKSDB_CORRUPT_DATA; } } /* Other fields are decoded from the value */ const char *null_bytes = nullptr; if (m_null_bytes_in_rec && !(null_bytes = reader.read(m_null_bytes_in_rec))) { return HA_ERR_ROCKSDB_CORRUPT_DATA; } if (m_maybe_unpack_info) { unpack_info = reader.get_current_ptr(); if (!unpack_info || !Rdb_key_def::is_unpack_data_tag(unpack_info[0]) || !reader.read(Rdb_key_def::get_unpack_header_size(unpack_info[0]))) { return HA_ERR_ROCKSDB_CORRUPT_DATA; } unpack_info_len = rdb_netbuf_to_uint16(reinterpret_cast(unpack_info + 1)); unpack_slice = rocksdb::Slice(unpack_info, unpack_info_len); reader.read(unpack_info_len - Rdb_key_def::get_unpack_header_size(unpack_info[0])); } int err = HA_EXIT_SUCCESS; if (m_key_requested) { err = m_pk_descr->unpack_record(table, buf, &rowkey_slice, unpack_info ? &unpack_slice : nullptr, false /* verify_checksum */); } if (err != HA_EXIT_SUCCESS) { return err; } for (auto it = m_decoders_vect.begin(); it != m_decoders_vect.end(); it++) { const Rdb_field_encoder *const field_dec = it->m_field_enc; const bool decode = it->m_decode; const bool isNull = field_dec->maybe_null() && ((null_bytes[field_dec->m_null_offset] & field_dec->m_null_mask) != 0); Field *const field = table->field[field_dec->m_field_index]; /* Skip the bytes we need to skip */ if (it->m_skip && !reader.read(it->m_skip)) { return HA_ERR_ROCKSDB_CORRUPT_DATA; } uint field_offset = field->ptr - table->record[0]; uint null_offset = field->null_offset(); bool maybe_null = field->real_maybe_null(); field->move_field(buf + field_offset, maybe_null ? buf + null_offset : nullptr, field->null_bit); // WARNING! - Don't return before restoring field->ptr and field->null_ptr! if (isNull) { if (decode) { /* This sets the NULL-bit of this record */ field->set_null(); /* Besides that, set the field value to default value. CHECKSUM TABLE depends on this. */ memcpy(field->ptr, table->s->default_values + field_offset, field->pack_length()); } } else { if (decode) { field->set_notnull(); } if (field_dec->m_field_type == MYSQL_TYPE_BLOB) { err = convert_blob_from_storage_format( (my_core::Field_blob *) field, &reader, decode); } else if (field_dec->m_field_type == MYSQL_TYPE_VARCHAR) { err = convert_varchar_from_storage_format( (my_core::Field_varstring *) field, &reader, decode); } else { err = convert_field_from_storage_format( field, &reader, decode, field_dec->m_pack_length_in_rec); } } // Restore field->ptr and field->null_ptr field->move_field(table->record[0] + field_offset, maybe_null ? table->record[0] + null_offset : nullptr, field->null_bit); if (err != HA_EXIT_SUCCESS) { return err; } } if (m_verify_row_debug_checksums) { if (reader.remaining_bytes() == RDB_CHECKSUM_CHUNK_SIZE && reader.read(1)[0] == RDB_CHECKSUM_DATA_TAG) { uint32_t stored_key_chksum = rdb_netbuf_to_uint32((const uchar *)reader.read(RDB_CHECKSUM_SIZE)); uint32_t stored_val_chksum = rdb_netbuf_to_uint32((const uchar *)reader.read(RDB_CHECKSUM_SIZE)); const uint32_t computed_key_chksum = my_core::crc32(0, rdb_slice_to_uchar_ptr(key), key->size()); const uint32_t computed_val_chksum = my_core::crc32(0, rdb_slice_to_uchar_ptr(value), value->size() - RDB_CHECKSUM_CHUNK_SIZE); DBUG_EXECUTE_IF("myrocks_simulate_bad_pk_checksum1", stored_key_chksum++;); if (stored_key_chksum != computed_key_chksum) { m_pk_descr->report_checksum_mismatch(true, key->data(), key->size()); return HA_ERR_ROCKSDB_CHECKSUM_MISMATCH; } DBUG_EXECUTE_IF("myrocks_simulate_bad_pk_checksum2", stored_val_chksum++;); if (stored_val_chksum != computed_val_chksum) { m_pk_descr->report_checksum_mismatch(false, value->data(), value->size()); return HA_ERR_ROCKSDB_CHECKSUM_MISMATCH; } m_row_checksums_checked++; } if (reader.remaining_bytes()) return HA_ERR_ROCKSDB_CORRUPT_DATA; } return HA_EXIT_SUCCESS; } void ha_rocksdb::get_storage_type(Rdb_field_encoder *const encoder, const uint &kp) { // STORE_SOME uses unpack_info. if (m_pk_descr->has_unpack_info(kp)) { DBUG_ASSERT(m_pk_descr->can_unpack(kp)); encoder->m_storage_type = Rdb_field_encoder::STORE_SOME; m_maybe_unpack_info = true; } else if (m_pk_descr->can_unpack(kp)) { encoder->m_storage_type = Rdb_field_encoder::STORE_NONE; } } /* Setup data needed to convert table->record[] to and from record storage format. @seealso ha_rocksdb::convert_record_to_storage_format, ha_rocksdb::convert_record_from_storage_format */ void ha_rocksdb::setup_field_converters() { uint i; uint null_bytes = 0; uchar cur_null_mask = 0x1; DBUG_ASSERT(m_encoder_arr == nullptr); m_encoder_arr = static_cast( my_malloc(table->s->fields * sizeof(Rdb_field_encoder), MYF(0))); if (m_encoder_arr == nullptr) { return; } for (i = 0; i < table->s->fields; i++) { Field *const field = table->field[i]; m_encoder_arr[i].m_storage_type = Rdb_field_encoder::STORE_ALL; /* Check if this field is - a part of primary key, and - it can be decoded back from its key image. If both hold, we don't need to store this field in the value part of RocksDB's key-value pair. If hidden pk exists, we skip this check since the field will never be part of the hidden pk. */ if (!has_hidden_pk(table)) { KEY *const pk_info = &table->key_info[table->s->primary_key]; for (uint kp = 0; kp < pk_info->user_defined_key_parts; kp++) { /* key_part->fieldnr is counted from 1 */ if (field->field_index + 1 == pk_info->key_part[kp].fieldnr) { get_storage_type(&m_encoder_arr[i], kp); break; } } } m_encoder_arr[i].m_field_type = field->real_type(); m_encoder_arr[i].m_field_index = i; m_encoder_arr[i].m_pack_length_in_rec = field->pack_length_in_rec(); if (field->real_maybe_null()) { m_encoder_arr[i].m_null_mask = cur_null_mask; m_encoder_arr[i].m_null_offset = null_bytes; if (cur_null_mask == 0x80) { cur_null_mask = 0x1; null_bytes++; } else cur_null_mask = cur_null_mask << 1; } else { m_encoder_arr[i].m_null_mask = 0; } } /* Count the last, unfinished NULL-bits byte */ if (cur_null_mask != 0x1) null_bytes++; m_null_bytes_in_rec = null_bytes; } int ha_rocksdb::alloc_key_buffers(const TABLE *const table_arg, const Rdb_tbl_def *const tbl_def_arg, bool alloc_alter_buffers) { DBUG_ENTER_FUNC(); DBUG_ASSERT(m_pk_tuple == nullptr); std::shared_ptr *const kd_arr = tbl_def_arg->m_key_descr_arr; uint key_len = 0; uint max_packed_sk_len = 0; uint pack_key_len = 0; m_pk_descr = kd_arr[pk_index(table_arg, tbl_def_arg)]; if (has_hidden_pk(table_arg)) { m_pk_key_parts = 1; } else { m_pk_key_parts = table->key_info[table->s->primary_key].user_defined_key_parts; key_len = table->key_info[table->s->primary_key].key_length; } // move this into get_table_handler() ?? m_pk_descr->setup(table_arg, tbl_def_arg); m_pk_tuple = reinterpret_cast(my_malloc(key_len, MYF(0))); pack_key_len = m_pk_descr->max_storage_fmt_length(); m_pk_packed_tuple = reinterpret_cast(my_malloc(pack_key_len, MYF(0))); /* Sometimes, we may use m_sk_packed_tuple for storing packed PK */ max_packed_sk_len = pack_key_len; for (uint i = 0; i < table_arg->s->keys; i++) { if (i == table_arg->s->primary_key) /* Primary key was processed above */ continue; // TODO: move this into get_table_handler() ?? kd_arr[i]->setup(table_arg, tbl_def_arg); const uint packed_len = kd_arr[i]->max_storage_fmt_length(); if (packed_len > max_packed_sk_len) { max_packed_sk_len = packed_len; } } m_sk_packed_tuple = reinterpret_cast(my_malloc(max_packed_sk_len, MYF(0))); m_sk_match_prefix_buf = reinterpret_cast(my_malloc(max_packed_sk_len, MYF(0))); m_sk_packed_tuple_old = reinterpret_cast(my_malloc(max_packed_sk_len, MYF(0))); m_end_key_packed_tuple = reinterpret_cast(my_malloc(max_packed_sk_len, MYF(0))); m_pack_buffer = reinterpret_cast(my_malloc(max_packed_sk_len, MYF(0))); m_scan_it_lower_bound = reinterpret_cast(my_malloc(max_packed_sk_len, MYF(0))); m_scan_it_upper_bound = reinterpret_cast(my_malloc(max_packed_sk_len, MYF(0))); /* If inplace alter is happening, allocate special buffers for unique secondary index duplicate checking. */ if (alloc_alter_buffers) { m_dup_sk_packed_tuple = reinterpret_cast(my_malloc(max_packed_sk_len, MYF(0))); m_dup_sk_packed_tuple_old = reinterpret_cast(my_malloc(max_packed_sk_len, MYF(0))); } if (m_pk_tuple == nullptr || m_pk_packed_tuple == nullptr || m_sk_packed_tuple == nullptr || m_sk_packed_tuple_old == nullptr || m_end_key_packed_tuple == nullptr || m_pack_buffer == nullptr || m_scan_it_upper_bound == nullptr || m_scan_it_lower_bound == nullptr || (alloc_alter_buffers && (m_dup_sk_packed_tuple == nullptr || m_dup_sk_packed_tuple_old == nullptr))) { // One or more of the above allocations failed. Clean up and exit free_key_buffers(); DBUG_RETURN(HA_ERR_OUT_OF_MEM); } DBUG_RETURN(HA_EXIT_SUCCESS); } void ha_rocksdb::free_key_buffers() { my_free(m_pk_tuple); m_pk_tuple = nullptr; my_free(m_pk_packed_tuple); m_pk_packed_tuple = nullptr; my_free(m_sk_packed_tuple); m_sk_packed_tuple = nullptr; my_free(m_sk_match_prefix_buf); m_sk_match_prefix_buf = nullptr; my_free(m_sk_packed_tuple_old); m_sk_packed_tuple_old = nullptr; my_free(m_end_key_packed_tuple); m_end_key_packed_tuple = nullptr; my_free(m_pack_buffer); m_pack_buffer = nullptr; my_free(m_dup_sk_packed_tuple); m_dup_sk_packed_tuple = nullptr; my_free(m_dup_sk_packed_tuple_old); m_dup_sk_packed_tuple_old = nullptr; my_free(m_scan_it_lower_bound); m_scan_it_lower_bound = nullptr; my_free(m_scan_it_upper_bound); m_scan_it_upper_bound = nullptr; } #ifdef MARIAROCKS_NOT_YET void ha_rocksdb::set_use_read_free_rpl(const char *const whitelist) { DBUG_ASSERT(whitelist != nullptr); #if defined(HAVE_PSI_INTERFACE) Regex_list_handler regex_handler(key_rwlock_read_free_rpl_tables); #else Regex_list_handler regex_handler; #endif if (!regex_handler.set_patterns(whitelist)) { warn_about_bad_patterns(®ex_handler, "read_free_rpl_tables"); } m_use_read_free_rpl = regex_handler.matches(m_tbl_def->base_tablename()); } #endif void ha_rocksdb::set_skip_unique_check_tables(const char *const whitelist) { DBUG_ASSERT(whitelist != nullptr); #if defined(HAVE_PSI_INTERFACE) Regex_list_handler regex_handler(key_rwlock_skip_unique_check_tables); #else Regex_list_handler regex_handler; #endif if (!regex_handler.set_patterns(whitelist)) { warn_about_bad_patterns(®ex_handler, "skip_unique_check_tables"); } m_skip_unique_check = regex_handler.matches(m_tbl_def->base_tablename()); } /** @return HA_EXIT_SUCCESS OK other HA_ERR error code (can be SE-specific) */ int ha_rocksdb::open(const char *const name, int mode, uint test_if_locked) { DBUG_ENTER_FUNC(); int err = close(); if (err) { DBUG_RETURN(err); } m_table_handler = rdb_open_tables.get_table_handler(name); if (m_table_handler == nullptr) { DBUG_RETURN(HA_ERR_OUT_OF_MEM); } my_core::thr_lock_data_init(&m_table_handler->m_thr_lock, &m_db_lock, nullptr); m_io_perf.init(&m_table_handler->m_table_perf_context, &m_table_handler->m_io_perf_read, &m_table_handler->m_io_perf_write, &stats); Rdb_perf_context_guard guard(&m_io_perf, rocksdb_perf_context_level(ha_thd())); std::string fullname; err = rdb_normalize_tablename(name, &fullname); if (err != HA_EXIT_SUCCESS) { DBUG_RETURN(err); } m_tbl_def = ddl_manager.find(fullname); if (m_tbl_def == nullptr) { my_error(ER_INTERNAL_ERROR, MYF(0), "Attempt to open a table that is not present in RocksDB-SE data " "dictionary"); DBUG_RETURN(HA_ERR_ROCKSDB_INVALID_TABLE); } m_lock_rows = RDB_LOCK_NONE; m_key_descr_arr = m_tbl_def->m_key_descr_arr; /* Full table scan actually uses primary key (UPDATE needs to know this, otherwise it will go into infinite loop on queries like "UPDATE tbl SET pk=pk+100") */ key_used_on_scan = table->s->primary_key; // close() above has already called free_key_buffers(). No need to do it here. err = alloc_key_buffers(table, m_tbl_def); if (err) { DBUG_RETURN(err); } /* init_with_fields() is used to initialize table flags based on the field definitions in table->field[]. It is called by open_binary_frm(), but that function calls the method for a temporary ha_rocksdb object which is later destroyed. If we are here in ::open(), then init_with_fields() has not been called for this object. Call it ourselves, we want all member variables to be properly initialized. */ init_with_fields(); setup_field_converters(); /* MariaDB: adjust field->part_of_key for PK columns. We can only do it here because SE API is just relying on the HA_PRIMARY_KEY_IN_READ_INDEX which does not allow to distinguish between unpack'able and non-unpack'able columns. Upstream uses handler->init_with_fields() but we don't have that call. */ { if (!has_hidden_pk(table)) { KEY *const pk_info = &table->key_info[table->s->primary_key]; for (uint kp = 0; kp < pk_info->user_defined_key_parts; kp++) { if (!m_pk_descr->can_unpack(kp)) { // uint field_index= pk_info->key_part[kp].field->field_index; table->field[field_index]->part_of_key.clear_all(); table->field[field_index]->part_of_key.set_bit(table->s->primary_key); } } } for (uint key= 0; key < table->s->keys; key++) { KEY *const key_info = &table->key_info[key]; if (key == table->s->primary_key) continue; for (uint kp = 0; kp < key_info->usable_key_parts; kp++) { uint field_index= key_info->key_part[kp].field->field_index; if (m_key_descr_arr[key]->can_unpack(kp)) { table->field[field_index]->part_of_key.set_bit(key); } else { table->field[field_index]->part_of_key.clear_bit(key); } } } } info(HA_STATUS_NO_LOCK | HA_STATUS_VARIABLE | HA_STATUS_CONST); /* The following load_XXX code calls row decode functions, and they do that without having done ::external_lock() or index_init()/rnd_init(). (Note: this also means we're doing a read when there was no setup_field_converters() call) Initialize the necessary variables for them: */ m_verify_row_debug_checksums = false; /* Load auto_increment value only once on first use. */ if (table->found_next_number_field && m_tbl_def->m_auto_incr_val == 0) { load_auto_incr_value(); } /* Load hidden pk only once on first use. */ if (has_hidden_pk(table) && m_tbl_def->m_hidden_pk_val == 0 && (err = load_hidden_pk_value()) != HA_EXIT_SUCCESS) { free_key_buffers(); DBUG_RETURN(err); } /* Index block size in MyRocks: used by MySQL in query optimization */ stats.block_size = rocksdb_tbl_options->block_size; #ifdef MARIAROCKS_NOT_YET // MDEV-10976 /* Determine at open whether we can use Read Free Replication or not */ set_use_read_free_rpl(THDVAR(ha_thd(), read_free_rpl_tables)); #endif /* Determine at open whether we should skip unique checks for this table */ set_skip_unique_check_tables(THDVAR(ha_thd(), skip_unique_check_tables)); DBUG_RETURN(HA_EXIT_SUCCESS); } int ha_rocksdb::close(void) { DBUG_ENTER_FUNC(); m_pk_descr = nullptr; m_key_descr_arr = nullptr; free_key_buffers(); my_free(m_encoder_arr); m_encoder_arr = nullptr; if (m_table_handler != nullptr) { rdb_open_tables.release_table_handler(m_table_handler); m_table_handler = nullptr; } // These are needed to suppress valgrind errors in rocksdb.partition m_storage_record.free(); m_last_rowkey.free(); m_sk_tails.free(); m_sk_tails_old.free(); m_pk_unpack_info.free(); DBUG_RETURN(HA_EXIT_SUCCESS); } static const char *rdb_error_messages[] = { "Table must have a PRIMARY KEY.", "Specifying DATA DIRECTORY for an individual table is not supported.", "Specifying INDEX DIRECTORY for an individual table is not supported.", "RocksDB commit failed.", "Failure during bulk load operation.", "Found data corruption.", "CRC checksum mismatch.", "Invalid table.", "Could not access RocksDB properties.", "File I/O error during merge/sort operation.", "RocksDB status: not found.", "RocksDB status: corruption.", "RocksDB status: invalid argument.", "RocksDB status: io error.", "RocksDB status: no space.", "RocksDB status: merge in progress.", "RocksDB status: incomplete.", "RocksDB status: shutdown in progress.", "RocksDB status: timed out.", "RocksDB status: aborted.", "RocksDB status: lock limit reached.", "RocksDB status: busy.", "RocksDB status: deadlock.", "RocksDB status: expired.", "RocksDB status: try again.", }; static_assert((sizeof(rdb_error_messages) / sizeof(rdb_error_messages[0])) == ((HA_ERR_ROCKSDB_LAST - HA_ERR_ROCKSDB_FIRST) + 1), "Number of error messages doesn't match number of error codes"); //psergey-merge: do we need this in MariaDB: we have get_error_messages //below... #if 0 static const char *rdb_get_error_message(int nr) { return rdb_error_messages[nr - HA_ERR_ROCKSDB_FIRST]; } #endif static const char **rdb_get_error_messages(int nr) { return rdb_error_messages; } bool ha_rocksdb::get_error_message(const int error, String *const buf) { DBUG_ENTER_FUNC(); static_assert(HA_ERR_ROCKSDB_LAST > HA_ERR_FIRST, "HA_ERR_ROCKSDB_LAST > HA_ERR_FIRST"); static_assert(HA_ERR_ROCKSDB_LAST > HA_ERR_LAST, "HA_ERR_ROCKSDB_LAST > HA_ERR_LAST"); if (error == HA_ERR_LOCK_WAIT_TIMEOUT || error == HA_ERR_LOCK_DEADLOCK || error == HA_ERR_ROCKSDB_STATUS_BUSY) { Rdb_transaction *const tx = get_tx_from_thd(ha_thd()); DBUG_ASSERT(tx != nullptr); buf->append(tx->m_detailed_error); DBUG_RETURN(true); } if (error >= HA_ERR_ROCKSDB_FIRST && error <= HA_ERR_ROCKSDB_LAST) { buf->append(rdb_error_messages[error - HA_ERR_ROCKSDB_FIRST]); } // We can be called with the values which are < HA_ERR_FIRST because most // MySQL internal functions will just return HA_EXIT_FAILURE in case of // an error. DBUG_RETURN(false); } /* Generalized way to convert RocksDB status errors into MySQL error code, and print error message. Each error code below maps to a RocksDB status code found in: rocksdb/include/rocksdb/status.h */ int ha_rocksdb::rdb_error_to_mysql(const rocksdb::Status &s, const char *opt_msg) { DBUG_ASSERT(!s.ok()); int err; switch (s.code()) { case rocksdb::Status::Code::kOk: err = HA_EXIT_SUCCESS; break; case rocksdb::Status::Code::kNotFound: err = HA_ERR_ROCKSDB_STATUS_NOT_FOUND; break; case rocksdb::Status::Code::kCorruption: err = HA_ERR_ROCKSDB_STATUS_CORRUPTION; break; case rocksdb::Status::Code::kNotSupported: err = HA_ERR_ROCKSDB_STATUS_NOT_SUPPORTED; break; case rocksdb::Status::Code::kInvalidArgument: err = HA_ERR_ROCKSDB_STATUS_INVALID_ARGUMENT; break; case rocksdb::Status::Code::kIOError: err = (s.IsNoSpace()) ? HA_ERR_ROCKSDB_STATUS_NO_SPACE : HA_ERR_ROCKSDB_STATUS_IO_ERROR; break; case rocksdb::Status::Code::kMergeInProgress: err = HA_ERR_ROCKSDB_STATUS_MERGE_IN_PROGRESS; break; case rocksdb::Status::Code::kIncomplete: err = HA_ERR_ROCKSDB_STATUS_INCOMPLETE; break; case rocksdb::Status::Code::kShutdownInProgress: err = HA_ERR_ROCKSDB_STATUS_SHUTDOWN_IN_PROGRESS; break; case rocksdb::Status::Code::kTimedOut: err = HA_ERR_ROCKSDB_STATUS_TIMED_OUT; break; case rocksdb::Status::Code::kAborted: err = (s.IsLockLimit()) ? HA_ERR_ROCKSDB_STATUS_LOCK_LIMIT : HA_ERR_ROCKSDB_STATUS_ABORTED; break; case rocksdb::Status::Code::kBusy: err = (s.IsDeadlock()) ? HA_ERR_ROCKSDB_STATUS_DEADLOCK : HA_ERR_ROCKSDB_STATUS_BUSY; break; case rocksdb::Status::Code::kExpired: err = HA_ERR_ROCKSDB_STATUS_EXPIRED; break; case rocksdb::Status::Code::kTryAgain: err = HA_ERR_ROCKSDB_STATUS_TRY_AGAIN; break; default: DBUG_ASSERT(0); return -1; } if (opt_msg) { std::string concatenated_error = s.ToString() + " (" + std::string(opt_msg) + ")"; my_error(ER_GET_ERRMSG, MYF(0), s.code(), concatenated_error.c_str(), rocksdb_hton_name); } else { my_error(ER_GET_ERRMSG, MYF(0), s.code(), s.ToString().c_str(), rocksdb_hton_name); } return err; } /* MyRocks supports only the following collations for indexed columns */ static const std::set RDB_INDEX_COLLATIONS = { COLLATION_BINARY, COLLATION_UTF8_BIN, COLLATION_LATIN1_BIN}; static bool rdb_is_index_collation_supported(const my_core::Field *const field) { const my_core::enum_field_types type = field->real_type(); /* Handle [VAR](CHAR|BINARY) or TEXT|BLOB */ if (type == MYSQL_TYPE_VARCHAR || type == MYSQL_TYPE_STRING || type == MYSQL_TYPE_BLOB) { return (RDB_INDEX_COLLATIONS.find(field->charset()->number) != RDB_INDEX_COLLATIONS.end()) || rdb_is_collation_supported(field->charset()); } return true; } static bool rdb_field_uses_nopad_collation(const my_core::Field *const field) { const my_core::enum_field_types type = field->real_type(); /* Handle [VAR](CHAR|BINARY) or TEXT|BLOB */ if (type == MYSQL_TYPE_VARCHAR || type == MYSQL_TYPE_STRING || type == MYSQL_TYPE_BLOB) { /* This is technically a NOPAD collation but it's a binary collation that we can handle. */ if (RDB_INDEX_COLLATIONS.find(field->charset()->number) != RDB_INDEX_COLLATIONS.end()) return false; return (field->charset()->state & MY_CS_NOPAD); } return false; } /* Create structures needed for storing data in rocksdb. This is called when the table is created. The structures will be shared by all TABLE* objects. @param table_arg Table with definition db_table "dbname.tablename" len strlen of the above tbl_def_arg tbl_def whose key_descr is being created/populated old_tbl_def_arg tbl_def from which keys are being copied over from (for use during inplace alter) @return 0 - Ok other - error, either given table ddl is not supported by rocksdb or OOM. */ int ha_rocksdb::create_key_defs( const TABLE *const table_arg, Rdb_tbl_def *const tbl_def_arg, const TABLE *const old_table_arg /* = nullptr */, const Rdb_tbl_def *const old_tbl_def_arg /* = nullptr */) const { DBUG_ENTER_FUNC(); DBUG_ASSERT(table_arg->s != nullptr); uint i; /* These need to be one greater than MAX_INDEXES since the user can create MAX_INDEXES secondary keys and no primary key which would cause us to generate a hidden one. */ std::array cfs; /* NOTE: All new column families must be created before new index numbers are allocated to each key definition. See below for more details. http://github.com/MySQLOnRocksDB/mysql-5.6/issues/86#issuecomment-138515501 */ if (create_cfs(table_arg, tbl_def_arg, &cfs)) { DBUG_RETURN(HA_EXIT_FAILURE); } if (!old_tbl_def_arg) { /* old_tbl_def doesn't exist. this means we are in the process of creating a new table. Get the index numbers (this will update the next_index_number) and create Rdb_key_def structures. */ for (i = 0; i < tbl_def_arg->m_key_count; i++) { if (create_key_def(table_arg, i, tbl_def_arg, &m_key_descr_arr[i], cfs[i])) { DBUG_RETURN(HA_EXIT_FAILURE); } } } else { /* old_tbl_def exists. This means we are creating a new tbl_def as part of in-place alter table. Copy over existing keys from the old_tbl_def and generate the necessary new key definitions if any. */ if (create_inplace_key_defs(table_arg, tbl_def_arg, old_table_arg, old_tbl_def_arg, cfs)) { DBUG_RETURN(HA_EXIT_FAILURE); } } DBUG_RETURN(HA_EXIT_SUCCESS); } /* Checks index parameters and creates column families needed for storing data in rocksdb if necessary. @param in table_arg Table with definition db_table Table name tbl_def_arg Table def structure being populated @param out cfs CF info for each key definition in 'key_info' order @return 0 - Ok other - error */ int ha_rocksdb::create_cfs( const TABLE *const table_arg, Rdb_tbl_def *const tbl_def_arg, std::array *const cfs) const { DBUG_ENTER_FUNC(); DBUG_ASSERT(table_arg->s != nullptr); char tablename_sys[NAME_LEN + 1]; bool tsys_set= false; /* The first loop checks the index parameters and creates column families if necessary. */ for (uint i = 0; i < tbl_def_arg->m_key_count; i++) { rocksdb::ColumnFamilyHandle *cf_handle; if (!is_hidden_pk(i, table_arg, tbl_def_arg) && tbl_def_arg->base_tablename().find(tmp_file_prefix) != 0) { if (!tsys_set) { tsys_set= true; my_core::filename_to_tablename(tbl_def_arg->base_tablename().c_str(), tablename_sys, sizeof(tablename_sys)); } for (uint part = 0; part < table_arg->key_info[i].ext_key_parts; part++) { /* MariaDB: disallow NOPAD collations */ if (rdb_field_uses_nopad_collation( table_arg->key_info[i].key_part[part].field)) { my_error(ER_MYROCKS_CANT_NOPAD_COLLATION, MYF(0)); DBUG_RETURN(HA_EXIT_FAILURE); } if (rocksdb_strict_collation_check && !rdb_is_index_collation_supported( table_arg->key_info[i].key_part[part].field) && !rdb_collation_exceptions->matches(tablename_sys)) { char buf[1024]; my_snprintf(buf, sizeof(buf), "Indexed column %s.%s uses a collation that does not " "allow index-only access in secondary key and has " "reduced disk space efficiency in primary key.", tbl_def_arg->full_tablename().c_str(), table_arg->key_info[i].key_part[part].field->field_name.str); my_error(ER_INTERNAL_ERROR, MYF(ME_JUST_WARNING), buf); } } } // Internal consistency check to make sure that data in TABLE and // Rdb_tbl_def structures matches. Either both are missing or both are // specified. Yes, this is critical enough to make it into SHIP_ASSERT. SHIP_ASSERT(IF_PARTITIONING(!table_arg->part_info,true) == tbl_def_arg->base_partition().empty()); // Generate the name for the column family to use. bool per_part_match_found = false; std::string cf_name = generate_cf_name(i, table_arg, tbl_def_arg, &per_part_match_found); // Prevent create from using the system column family. if (cf_name == DEFAULT_SYSTEM_CF_NAME) { my_error(ER_WRONG_ARGUMENTS, MYF(0), "column family not valid for storing index data."); DBUG_RETURN(HA_EXIT_FAILURE); } // Here's how `get_or_create_cf` will use the input parameters: // // `cf_name` - will be used as a CF name. cf_handle = cf_manager.get_or_create_cf(rdb, cf_name); if (!cf_handle) { DBUG_RETURN(HA_EXIT_FAILURE); } auto &cf = (*cfs)[i]; cf.cf_handle = cf_handle; cf.is_reverse_cf = Rdb_cf_manager::is_cf_name_reverse(cf_name.c_str()); cf.is_per_partition_cf = per_part_match_found; } DBUG_RETURN(HA_EXIT_SUCCESS); } /* Create key definition needed for storing data in rocksdb during ADD index inplace operations. @param in table_arg Table with definition tbl_def_arg New table def structure being populated old_tbl_def_arg Old(current) table def structure cfs Struct array which contains column family information @return 0 - Ok other - error, either given table ddl is not supported by rocksdb or OOM. */ int ha_rocksdb::create_inplace_key_defs( const TABLE *const table_arg, Rdb_tbl_def *const tbl_def_arg, const TABLE *const old_table_arg, const Rdb_tbl_def *const old_tbl_def_arg, const std::array &cfs) const { DBUG_ENTER_FUNC(); std::shared_ptr *const old_key_descr = old_tbl_def_arg->m_key_descr_arr; std::shared_ptr *const new_key_descr = tbl_def_arg->m_key_descr_arr; const std::unordered_map old_key_pos = get_old_key_positions(table_arg, tbl_def_arg, old_table_arg, old_tbl_def_arg); uint i; for (i = 0; i < tbl_def_arg->m_key_count; i++) { const auto &it = old_key_pos.find(get_key_name(i, table_arg, tbl_def_arg)); if (it != old_key_pos.end()) { /* Found matching index in old table definition, so copy it over to the new one created. */ const Rdb_key_def &okd = *old_key_descr[it->second]; const GL_INDEX_ID gl_index_id = okd.get_gl_index_id(); struct Rdb_index_info index_info; if (!dict_manager.get_index_info(gl_index_id, &index_info)) { // NO_LINT_DEBUG sql_print_error("RocksDB: Could not get index information " "for Index Number (%u,%u), table %s", gl_index_id.cf_id, gl_index_id.index_id, old_tbl_def_arg->full_tablename().c_str()); DBUG_RETURN(HA_EXIT_FAILURE); } uint32 ttl_rec_offset = Rdb_key_def::has_index_flag(index_info.m_index_flags, Rdb_key_def::TTL_FLAG) ? Rdb_key_def::calculate_index_flag_offset( index_info.m_index_flags, Rdb_key_def::TTL_FLAG) : UINT_MAX; /* We can't use the copy constructor because we need to update the keynr within the pack_info for each field and the keyno of the keydef itself. */ new_key_descr[i] = std::make_shared( okd.get_index_number(), i, okd.get_cf(), index_info.m_index_dict_version, index_info.m_index_type, index_info.m_kv_version, okd.m_is_reverse_cf, okd.m_is_per_partition_cf, okd.m_name.c_str(), dict_manager.get_stats(gl_index_id), index_info.m_index_flags, ttl_rec_offset, index_info.m_ttl_duration); } else if (create_key_def(table_arg, i, tbl_def_arg, &new_key_descr[i], cfs[i])) { DBUG_RETURN(HA_EXIT_FAILURE); } DBUG_ASSERT(new_key_descr[i] != nullptr); new_key_descr[i]->setup(table_arg, tbl_def_arg); } DBUG_RETURN(HA_EXIT_SUCCESS); } std::unordered_map ha_rocksdb::get_old_key_positions( const TABLE *const table_arg, const Rdb_tbl_def *const tbl_def_arg, const TABLE *const old_table_arg, const Rdb_tbl_def *const old_tbl_def_arg) const { DBUG_ENTER_FUNC(); std::shared_ptr *const old_key_descr = old_tbl_def_arg->m_key_descr_arr; std::unordered_map old_key_pos; std::unordered_map new_key_pos; uint i; for (i = 0; i < tbl_def_arg->m_key_count; i++) { new_key_pos[get_key_name(i, table_arg, tbl_def_arg)] = i; } for (i = 0; i < old_tbl_def_arg->m_key_count; i++) { if (is_hidden_pk(i, old_table_arg, old_tbl_def_arg)) { old_key_pos[old_key_descr[i]->m_name] = i; continue; } /* In case of matching key name, need to check key parts of keys as well, in case a simultaneous drop + add is performed, where the key name is the same but the key parts are different. Example: CREATE TABLE t1 (a INT, b INT, KEY ka(a)) ENGINE=RocksDB; ALTER TABLE t1 DROP INDEX ka, ADD INDEX ka(b), ALGORITHM=INPLACE; */ const KEY *const old_key = &old_table_arg->key_info[i]; const auto &it = new_key_pos.find(old_key->name.str); if (it == new_key_pos.end()) { continue; } KEY *const new_key = &table_arg->key_info[it->second]; /* Check that the key is identical between old and new tables. If not, we still need to create a new index. The exception is if there is an index changed from unique to non-unique, in these cases we don't need to rebuild as they are stored the same way in RocksDB. */ bool unique_to_non_unique = ((old_key->flags ^ new_key->flags) == HA_NOSAME) && (old_key->flags & HA_NOSAME); if (compare_keys(old_key, new_key) && !unique_to_non_unique) { continue; } /* Check to make sure key parts match. */ if (compare_key_parts(old_key, new_key)) { continue; } old_key_pos[old_key->name.str] = i; } DBUG_RETURN(old_key_pos); } /* Check to see if two keys are identical. */ int ha_rocksdb::compare_keys(const KEY *const old_key, const KEY *const new_key) const { DBUG_ENTER_FUNC(); /* Check index name. */ if (strcmp(old_key->name.str, new_key->name.str) != 0) { DBUG_RETURN(HA_EXIT_FAILURE); } /* If index algorithms are different then keys are different. */ if (old_key->algorithm != new_key->algorithm) { DBUG_RETURN(HA_EXIT_FAILURE); } /* Check that the key is identical between old and new tables. */ if ((old_key->flags ^ new_key->flags) & HA_KEYFLAG_MASK) { DBUG_RETURN(HA_EXIT_FAILURE); } /* Check index comment. (for column family changes) */ std::string old_comment(old_key->comment.str, old_key->comment.length); std::string new_comment(new_key->comment.str, new_key->comment.length); if (old_comment.compare(new_comment) != 0) { DBUG_RETURN(HA_EXIT_FAILURE); } DBUG_RETURN(HA_EXIT_SUCCESS); } /* Check two keys to ensure that key parts within keys match */ int ha_rocksdb::compare_key_parts(const KEY *const old_key, const KEY *const new_key) const { DBUG_ENTER_FUNC(); /* Skip if key parts do not match, as it is a different key */ if (new_key->user_defined_key_parts != old_key->user_defined_key_parts) { DBUG_RETURN(HA_EXIT_FAILURE); } /* Check to see that key parts themselves match */ for (uint i = 0; i < old_key->user_defined_key_parts; i++) { if (strcmp(old_key->key_part[i].field->field_name.str, new_key->key_part[i].field->field_name.str) != 0) { DBUG_RETURN(HA_EXIT_FAILURE); } /* Check if prefix index key part length has changed */ if (old_key->key_part[i].length != new_key->key_part[i].length) { DBUG_RETURN(HA_EXIT_FAILURE); } } DBUG_RETURN(HA_EXIT_SUCCESS); } /* Create key definition needed for storing data in rocksdb. This can be called either during CREATE table or doing ADD index operations. @param in table_arg Table with definition i Position of index being created inside table_arg->key_info tbl_def_arg Table def structure being populated cf_info Struct which contains column family information @param out new_key_def Newly created index definition. @return 0 - Ok other - error, either given table ddl is not supported by rocksdb or OOM. */ int ha_rocksdb::create_key_def(const TABLE *const table_arg, const uint &i, const Rdb_tbl_def *const tbl_def_arg, std::shared_ptr *const new_key_def, const struct key_def_cf_info &cf_info) const { DBUG_ENTER_FUNC(); DBUG_ASSERT(*new_key_def == nullptr); uint64 ttl_duration = 0; std::string ttl_column; uint ttl_field_offset; uint err; if ((err = Rdb_key_def::extract_ttl_duration(table_arg, tbl_def_arg, &ttl_duration))) { DBUG_RETURN(err); } if ((err = Rdb_key_def::extract_ttl_col(table_arg, tbl_def_arg, &ttl_column, &ttl_field_offset))) { DBUG_RETURN(err); } /* We don't currently support TTL on tables with hidden primary keys. */ if (ttl_duration > 0 && is_hidden_pk(i, table_arg, tbl_def_arg)) { my_error(ER_RDB_TTL_UNSUPPORTED, MYF(0)); DBUG_RETURN(HA_EXIT_FAILURE); } /* If TTL duration is not specified but TTL column was specified, throw an error because TTL column requires duration. */ if (ttl_duration == 0 && !ttl_column.empty()) { my_error(ER_RDB_TTL_COL_FORMAT, MYF(0), ttl_column.c_str()); DBUG_RETURN(HA_EXIT_FAILURE); } const uint index_id = ddl_manager.get_and_update_next_number(&dict_manager); const uint16_t index_dict_version = Rdb_key_def::INDEX_INFO_VERSION_LATEST; uchar index_type; uint16_t kv_version; if (is_hidden_pk(i, table_arg, tbl_def_arg)) { index_type = Rdb_key_def::INDEX_TYPE_HIDDEN_PRIMARY; kv_version = Rdb_key_def::PRIMARY_FORMAT_VERSION_LATEST; } else if (i == table_arg->s->primary_key) { index_type = Rdb_key_def::INDEX_TYPE_PRIMARY; uint16 pk_latest_version = Rdb_key_def::PRIMARY_FORMAT_VERSION_LATEST; kv_version = pk_latest_version; } else { index_type = Rdb_key_def::INDEX_TYPE_SECONDARY; uint16 sk_latest_version = Rdb_key_def::SECONDARY_FORMAT_VERSION_LATEST; kv_version = sk_latest_version; } // Use PRIMARY_FORMAT_VERSION_UPDATE1 here since it is the same value as // SECONDARY_FORMAT_VERSION_UPDATE1 so it doesn't matter if this is a // primary key or secondary key. DBUG_EXECUTE_IF("MYROCKS_LEGACY_VARBINARY_FORMAT", { kv_version = Rdb_key_def::PRIMARY_FORMAT_VERSION_UPDATE1; }); DBUG_EXECUTE_IF("MYROCKS_NO_COVERED_BITMAP_FORMAT", { if (index_type == Rdb_key_def::INDEX_TYPE_SECONDARY) { kv_version = Rdb_key_def::SECONDARY_FORMAT_VERSION_UPDATE2; } }); uint32 index_flags = (ttl_duration > 0 ? Rdb_key_def::TTL_FLAG : 0); uint32 ttl_rec_offset = Rdb_key_def::has_index_flag(index_flags, Rdb_key_def::TTL_FLAG) ? Rdb_key_def::calculate_index_flag_offset(index_flags, Rdb_key_def::TTL_FLAG) : UINT_MAX; const char *const key_name = get_key_name(i, table_arg, m_tbl_def); *new_key_def = std::make_shared( index_id, i, cf_info.cf_handle, index_dict_version, index_type, kv_version, cf_info.is_reverse_cf, cf_info.is_per_partition_cf, key_name, Rdb_index_stats(), index_flags, ttl_rec_offset, ttl_duration); if (!ttl_column.empty()) { (*new_key_def)->m_ttl_column = ttl_column; } DBUG_RETURN(HA_EXIT_SUCCESS); } int rdb_normalize_tablename(const std::string &tablename, std::string *const strbuf) { if (tablename.size() < 2 || tablename[0] != '.' || (tablename[1] != FN_LIBCHAR && tablename[1] != FN_LIBCHAR2)) { DBUG_ASSERT(0); // We were not passed table name? return HA_ERR_ROCKSDB_INVALID_TABLE; } size_t pos = tablename.find_first_of(FN_LIBCHAR, 2); if (pos == std::string::npos) { pos = tablename.find_first_of(FN_LIBCHAR2, 2); } if (pos == std::string::npos) { DBUG_ASSERT(0); // We were not passed table name? return HA_ERR_ROCKSDB_INVALID_TABLE; } *strbuf = tablename.substr(2, pos - 2) + "." + tablename.substr(pos + 1); return HA_EXIT_SUCCESS; } /* Check to see if the user's original statement includes foreign key references */ bool ha_rocksdb::contains_foreign_key(THD *const thd) { bool success; const char *str = thd_query_string(thd)->str; DBUG_ASSERT(str != nullptr); while (*str != '\0') { // Scan from our current pos looking for 'FOREIGN' str = rdb_find_in_string(str, "FOREIGN", &success); if (!success) { return false; } // Skip past the found "FOREIGN' str = rdb_check_next_token(&my_charset_bin, str, "FOREIGN", &success); DBUG_ASSERT(success); if (!my_isspace(&my_charset_bin, *str)) { return false; } // See if the next token is 'KEY' str = rdb_check_next_token(&my_charset_bin, str, "KEY", &success); if (!success) { continue; } // See if the next token is '(' str = rdb_check_next_token(&my_charset_bin, str, "(", &success); if (!success) { // There is an optional index id after 'FOREIGN KEY', skip it str = rdb_skip_id(&my_charset_bin, str); // Now check for '(' again str = rdb_check_next_token(&my_charset_bin, str, "(", &success); } // If we have found 'FOREIGN KEY [] (' we can be confident we have // a foreign key clause. return success; } // We never found a valid foreign key clause return false; } /** @brief splits the normalized table name of .#P# into the , and components. @param dbbuf returns database name/table_schema @param tablebuf returns tablename @param partitionbuf returns partition suffix if there is one @return HA_EXIT_SUCCESS on success, non-zero on failure to split */ int rdb_split_normalized_tablename(const std::string &fullname, std::string *const db, std::string *const table, std::string *const partition) { DBUG_ASSERT(!fullname.empty()); #define RDB_PARTITION_STR "#P#" /* Normalize returns dbname.tablename. */ size_t dotpos = fullname.find('.'); /* Invalid table name? */ if (dotpos == std::string::npos) { return HA_ERR_ROCKSDB_INVALID_TABLE; } // Table must have a database name associated with it. DBUG_ASSERT(dotpos > 0); if (db != nullptr) { *db = fullname.substr(0, dotpos); } dotpos++; const size_t partpos = fullname.find(RDB_PARTITION_STR, dotpos, strlen(RDB_PARTITION_STR)); if (partpos != std::string::npos) { DBUG_ASSERT(partpos >= dotpos); if (table != nullptr) { *table = fullname.substr(dotpos, partpos - dotpos); } if (partition != nullptr) { *partition = fullname.substr(partpos + strlen(RDB_PARTITION_STR)); } } else if (table != nullptr) { *table = fullname.substr(dotpos); } return HA_EXIT_SUCCESS; } /** @brief create() is called to create a database. The variable name will have the name of the table. @details When create() is called you do not need to worry about opening the table. Also, the .frm file will have already been created so adjusting create_info is not necessary. You can overwrite the .frm file at this point if you wish to change the table definition, but there are no methods currently provided for doing so. Called from handle.cc by ha_create_table(). @return HA_EXIT_SUCCESS OK other HA_ERR error code (can be SE-specific) @see ha_create_table() in handle.cc */ int ha_rocksdb::create(const char *const name, TABLE *const table_arg, HA_CREATE_INFO *const create_info) { DBUG_ENTER_FUNC(); DBUG_ASSERT(table_arg != nullptr); DBUG_ASSERT(create_info != nullptr); if (create_info->data_file_name) { // DATA DIRECTORY is used to create tables under a specific location // outside the MySQL data directory. We don't support this for MyRocks. // The `rocksdb_datadir` setting should be used to configure RocksDB data // directory. print_error(HA_ERR_ROCKSDB_TABLE_DATA_DIRECTORY_NOT_SUPPORTED, MYF(ME_JUST_WARNING)); DBUG_RETURN(HA_WRONG_CREATE_OPTION); } if (create_info->index_file_name) { // Similar check for INDEX DIRECTORY as well. print_error(HA_ERR_ROCKSDB_TABLE_INDEX_DIRECTORY_NOT_SUPPORTED, MYF(ME_JUST_WARNING)); DBUG_RETURN(HA_WRONG_CREATE_OPTION); } std::string str; int err; THD *const thd = my_core::thd_get_current_thd(); if (get_table_if_exists(name)) { if (thd->lex->sql_command == SQLCOM_TRUNCATE) { err = delete_table(name); if (err != HA_EXIT_SUCCESS) { DBUG_RETURN(err); } } else { err = rdb_normalize_tablename(name, &str); if (err != HA_EXIT_SUCCESS) { DBUG_RETURN(err); } my_error(ER_METADATA_INCONSISTENCY, MYF(0), str.c_str(), name); DBUG_RETURN(HA_ERR_ROCKSDB_CORRUPT_DATA); } } /* Construct dbname.tablename ourselves, because parititioning passes strings like "./test/t14#P#p0" for individual partitions, while table_arg->s->table_name has none of that. */ err = rdb_normalize_tablename(name, &str); if (err != HA_EXIT_SUCCESS) { DBUG_RETURN(err); } if (contains_foreign_key(thd)) { my_error(ER_NOT_SUPPORTED_YET, MYF(0), "FOREIGN KEY for the RocksDB storage engine"); DBUG_RETURN(HA_ERR_UNSUPPORTED); } const std::unique_ptr wb = dict_manager.begin(); rocksdb::WriteBatch *const batch = wb.get(); /* Create table/key descriptions and put them into the data dictionary */ m_tbl_def = new Rdb_tbl_def(str); uint n_keys = table_arg->s->keys; /* If no primary key found, create a hidden PK and place it inside table definition */ if (has_hidden_pk(table_arg)) { n_keys += 1; } m_key_descr_arr = new std::shared_ptr[n_keys]; m_tbl_def->m_key_count = n_keys; m_tbl_def->m_key_descr_arr = m_key_descr_arr; err = create_key_defs(table_arg, m_tbl_def); if (err != HA_EXIT_SUCCESS) { goto error; } m_pk_descr = m_key_descr_arr[pk_index(table_arg, m_tbl_def)]; if (create_info->auto_increment_value) { bool autoinc_upgrade_test = false; m_tbl_def->m_auto_incr_val = create_info->auto_increment_value; DBUG_EXECUTE_IF("myrocks_autoinc_upgrade", autoinc_upgrade_test = true;); if (!autoinc_upgrade_test) { auto s = dict_manager.put_auto_incr_val( batch, m_tbl_def->get_autoincr_gl_index_id(), m_tbl_def->m_auto_incr_val); if (!s.ok()) { goto error; } } } dict_manager.lock(); err = ddl_manager.put_and_write(m_tbl_def, batch); if (err != HA_EXIT_SUCCESS) { dict_manager.unlock(); goto error; } err = dict_manager.commit(batch); if (err != HA_EXIT_SUCCESS) { dict_manager.unlock(); goto error; } dict_manager.unlock(); DBUG_RETURN(HA_EXIT_SUCCESS); error: /* Delete what we have allocated so far */ delete m_tbl_def; m_tbl_def = nullptr; m_key_descr_arr = nullptr; DBUG_RETURN(err); } /** @note This function is used only when the table has not yet been opened, and keyread_allowed bitmap doesn't have the correct values yet. See comment in ha_rocksdb::index_flags() for details. */ bool ha_rocksdb::check_keyread_allowed(uint inx, uint part, bool all_parts) const { bool res = true; KEY *const key_info = &table_share->key_info[inx]; Rdb_field_packing dummy1; res = dummy1.setup(nullptr, key_info->key_part[part].field, inx, part, key_info->key_part[part].length); if (res && all_parts) { for (uint i = 0; i < part; i++) { Field *field; if ((field = key_info->key_part[i].field)) { Rdb_field_packing dummy; if (!dummy.setup(nullptr, field, inx, i, key_info->key_part[i].length)) { /* Cannot do index-only reads for this column */ res = false; break; } } } } const uint pk = table_share->primary_key; if (inx == pk && all_parts && part + 1 == table_share->key_info[pk].user_defined_key_parts) { m_pk_can_be_decoded = res; } return res; } int ha_rocksdb::read_key_exact(const Rdb_key_def &kd, rocksdb::Iterator *const iter, const bool &full_key_match, const rocksdb::Slice &key_slice, const int64_t ttl_filter_ts) { /* We are looking for the first record such that index_tuple= lookup_tuple. lookup_tuple may be a prefix of the index. */ rocksdb_smart_seek(kd.m_is_reverse_cf, iter, key_slice); while (iter->Valid() && kd.value_matches_prefix(iter->key(), key_slice)) { /* If TTL is enabled we need to check if the given key has already expired from the POV of the current transaction. If it has, try going to the next key. */ if (kd.has_ttl() && should_hide_ttl_rec(kd, iter->value(), ttl_filter_ts)) { rocksdb_smart_next(kd.m_is_reverse_cf, iter); continue; } return HA_EXIT_SUCCESS; } /* Got a record that is not equal to the lookup value, or even a record from another table.index. */ return HA_ERR_KEY_NOT_FOUND; } int ha_rocksdb::read_before_key(const Rdb_key_def &kd, const bool &full_key_match, const rocksdb::Slice &key_slice, const int64_t ttl_filter_ts) { /* We are looking for record with the biggest t.key such that t.key < lookup_tuple. */ rocksdb_smart_seek(!kd.m_is_reverse_cf, m_scan_it, key_slice); while (is_valid(m_scan_it)) { /* We are using full key and we've hit an exact match, or... If TTL is enabled we need to check if the given key has already expired from the POV of the current transaction. If it has, try going to the next key. */ if ((full_key_match && kd.value_matches_prefix(m_scan_it->key(), key_slice)) || (kd.has_ttl() && should_hide_ttl_rec(kd, m_scan_it->value(), ttl_filter_ts))) { rocksdb_smart_next(!kd.m_is_reverse_cf, m_scan_it); continue; } return HA_EXIT_SUCCESS; } return HA_ERR_KEY_NOT_FOUND; } int ha_rocksdb::read_after_key(const Rdb_key_def &kd, const rocksdb::Slice &key_slice, const int64_t ttl_filter_ts) { /* We are looking for the first record such that index_tuple $GT lookup_tuple with HA_READ_AFTER_KEY, $GT = '>', with HA_READ_KEY_OR_NEXT, $GT = '>=' */ rocksdb_smart_seek(kd.m_is_reverse_cf, m_scan_it, key_slice); /* If TTL is enabled we need to check if the given key has already expired from the POV of the current transaction. If it has, try going to the next key. */ while (is_valid(m_scan_it) && kd.has_ttl() && should_hide_ttl_rec(kd, m_scan_it->value(), ttl_filter_ts)) { rocksdb_smart_next(kd.m_is_reverse_cf, m_scan_it); } return is_valid(m_scan_it) ? HA_EXIT_SUCCESS : HA_ERR_KEY_NOT_FOUND; } int ha_rocksdb::position_to_correct_key( const Rdb_key_def &kd, const enum ha_rkey_function &find_flag, const bool &full_key_match, const uchar *const key, const key_part_map &keypart_map, const rocksdb::Slice &key_slice, bool *const move_forward, const int64_t ttl_filter_ts) { int rc = 0; *move_forward = true; switch (find_flag) { case HA_READ_KEY_EXACT: rc = read_key_exact(kd, m_scan_it, full_key_match, key_slice, ttl_filter_ts); break; case HA_READ_BEFORE_KEY: *move_forward = false; rc = read_before_key(kd, full_key_match, key_slice, ttl_filter_ts); if (rc == 0 && !kd.covers_key(m_scan_it->key())) { /* The record we've got is not from this index */ rc = HA_ERR_KEY_NOT_FOUND; } break; case HA_READ_AFTER_KEY: case HA_READ_KEY_OR_NEXT: rc = read_after_key(kd, key_slice, ttl_filter_ts); if (rc == 0 && !kd.covers_key(m_scan_it->key())) { /* The record we've got is not from this index */ rc = HA_ERR_KEY_NOT_FOUND; } break; case HA_READ_KEY_OR_PREV: case HA_READ_PREFIX: /* This flag is not used by the SQL layer, so we don't support it yet. */ rc = HA_ERR_UNSUPPORTED; break; case HA_READ_PREFIX_LAST: case HA_READ_PREFIX_LAST_OR_PREV: *move_forward = false; /* Find the last record with the specified index prefix lookup. - HA_READ_PREFIX_LAST requires that the record has the prefix=lookup (if there are no such records, HA_ERR_KEY_NOT_FOUND should be returned). - HA_READ_PREFIX_LAST_OR_PREV has no such requirement. If there are no records with prefix=lookup, we should return the last record before that. */ rc = read_before_key(kd, full_key_match, key_slice, ttl_filter_ts); if (rc == 0) { const rocksdb::Slice &rkey = m_scan_it->key(); if (!kd.covers_key(rkey)) { /* The record we've got is not from this index */ rc = HA_ERR_KEY_NOT_FOUND; } else if (find_flag == HA_READ_PREFIX_LAST) { uint size = kd.pack_index_tuple(table, m_pack_buffer, m_sk_packed_tuple, key, keypart_map); rocksdb::Slice lookup_tuple(reinterpret_cast(m_sk_packed_tuple), size); // We need to compare the key we've got with the original search prefix. if (!kd.value_matches_prefix(rkey, lookup_tuple)) { rc = HA_ERR_KEY_NOT_FOUND; } } } break; default: DBUG_ASSERT(0); break; } return rc; } int ha_rocksdb::calc_eq_cond_len(const Rdb_key_def &kd, const enum ha_rkey_function &find_flag, const rocksdb::Slice &slice, const int &bytes_changed_by_succ, const key_range *const end_key, uint *const end_key_packed_size) { if (find_flag == HA_READ_KEY_EXACT) return slice.size(); if (find_flag == HA_READ_PREFIX_LAST) { /* We have made the kd.successor(m_sk_packed_tuple) call above. The slice is at least Rdb_key_def::INDEX_NUMBER_SIZE bytes long. */ return slice.size() - bytes_changed_by_succ; } if (end_key) { *end_key_packed_size = kd.pack_index_tuple(table, m_pack_buffer, m_end_key_packed_tuple, end_key->key, end_key->keypart_map); /* Calculating length of the equal conditions here. 4 byte index id is included. Example1: id1 BIGINT, id2 INT, id3 BIGINT, PRIMARY KEY (id1, id2, id3) WHERE id1=1 AND id2=1 AND id3>=2 => eq_cond_len= 4+8+4= 16 WHERE id1=1 AND id2>=1 AND id3>=2 => eq_cond_len= 4+8= 12 Example2: id1 VARCHAR(30), id2 INT, PRIMARY KEY (id1, id2) WHERE id1 = 'AAA' and id2 < 3; => eq_cond_len=13 (varchar used 9 bytes) */ rocksdb::Slice end_slice(reinterpret_cast(m_end_key_packed_tuple), *end_key_packed_size); return slice.difference_offset(end_slice); } /* On range scan without any end key condition, there is no eq cond, and eq cond length is the same as index_id size (4 bytes). Example1: id1 BIGINT, id2 INT, id3 BIGINT, PRIMARY KEY (id1, id2, id3) WHERE id1>=1 AND id2 >= 2 and id2 <= 5 => eq_cond_len= 4 */ return Rdb_key_def::INDEX_NUMBER_SIZE; } int ha_rocksdb::read_row_from_primary_key(uchar *const buf) { int rc; const rocksdb::Slice &rkey = m_scan_it->key(); const uint pk_size = rkey.size(); const char *pk_data = rkey.data(); memcpy(m_pk_packed_tuple, pk_data, pk_size); m_last_rowkey.copy(pk_data, pk_size, &my_charset_bin); if (m_lock_rows != RDB_LOCK_NONE) { /* We need to put a lock and re-read */ rc = get_row_by_rowid(buf, m_pk_packed_tuple, pk_size); } else { /* Unpack from the row we've read */ const rocksdb::Slice &value = m_scan_it->value(); rc = convert_record_from_storage_format(&rkey, &value, buf); } return rc; } int ha_rocksdb::read_row_from_secondary_key(uchar *const buf, const Rdb_key_def &kd, bool move_forward) { int rc = 0; uint pk_size; /* Get the key columns and primary key value */ const rocksdb::Slice &rkey = m_scan_it->key(); const rocksdb::Slice &value = m_scan_it->value(); bool covered_lookup = m_keyread_only || kd.covers_lookup(table, &value, &m_lookup_bitmap); if (covered_lookup && m_lock_rows == RDB_LOCK_NONE) { pk_size = kd.get_primary_key_tuple(table, *m_pk_descr, &rkey, m_pk_packed_tuple); if (pk_size == RDB_INVALID_KEY_LEN) { rc = HA_ERR_ROCKSDB_CORRUPT_DATA; } else { rc = kd.unpack_record(table, buf, &rkey, &value, m_verify_row_debug_checksums); global_stats.covered_secondary_key_lookups.inc(); } } else { if (kd.m_is_reverse_cf) move_forward = !move_forward; rc = find_icp_matching_index_rec(move_forward, buf); if (!rc) { const rocksdb::Slice &rkey = m_scan_it->key(); pk_size = kd.get_primary_key_tuple(table, *m_pk_descr, &rkey, m_pk_packed_tuple); if (pk_size == RDB_INVALID_KEY_LEN) { rc = HA_ERR_ROCKSDB_CORRUPT_DATA; } else { rc = get_row_by_rowid(buf, m_pk_packed_tuple, pk_size); } } } if (!rc) { m_last_rowkey.copy((const char *)m_pk_packed_tuple, pk_size, &my_charset_bin); } return rc; } /** @note The problem with this function is that SQL layer calls it, when - the table has not been yet opened (no ::open() call done) - this->table_share already exists, but it is in the process of being filled, so some of fields are still NULL. - In particular, table_share->key_info[inx].key_part[] is filled only up to part #part. Subsequent key parts are not yet filled. To complicate things further, SQL layer will call index_flags() with all_parts=TRUE. Essentially, we're asked to provide flags for reading keyparts whose datatype is not yet known. We walk around this problem by using check_keyread_allowed(), which uses table_share object and is careful not to step on unitialized data. When we get a call with all_parts=TRUE, we try to analyze all parts but ignore those that have key_part->field==nullptr (these are not initialized yet). */ ulong ha_rocksdb::index_flags(uint inx, uint part, bool all_parts) const { DBUG_ENTER_FUNC(); ulong base_flags = HA_READ_NEXT | // doesn't seem to be used HA_READ_ORDER | HA_READ_RANGE | HA_READ_PREV; if (check_keyread_allowed(inx, part, all_parts)) base_flags |= HA_KEYREAD_ONLY; if (inx == table_share->primary_key) { /* Index-only reads on primary key are the same as table scan for us. Still, we need to explicitly "allow" them, otherwise SQL layer will miss some plans. */ base_flags |= HA_KEYREAD_ONLY | HA_CLUSTERED_INDEX; } else { /* We can Index Condition Pushdown any key except the primary. With primary key, we get (pk, record) pair immediately, there is no place to put the ICP check. */ base_flags |= HA_DO_INDEX_COND_PUSHDOWN; } DBUG_RETURN(base_flags); } /** @brief Read next index tuple through the secondary index. @details m_scan_it points at the index key-value pair that we should read the (pk,row) pair for. */ int ha_rocksdb::secondary_index_read(const int keyno, uchar *const buf) { DBUG_ASSERT(table != nullptr); #ifdef MARIAROCKS_NOT_YET stats.rows_requested++; #endif /* Use STATUS_NOT_FOUND when record not found or some error occurred */ table->status = STATUS_NOT_FOUND; if (is_valid(m_scan_it)) { rocksdb::Slice key = m_scan_it->key(); /* Check if we've ran out of records of this index */ if (m_key_descr_arr[keyno]->covers_key(key)) { int rc = 0; // TODO: We could here check if we have ran out of range we're scanning const uint size = m_key_descr_arr[keyno]->get_primary_key_tuple( table, *m_pk_descr, &key, m_pk_packed_tuple); if (size == RDB_INVALID_KEY_LEN) { return HA_ERR_ROCKSDB_CORRUPT_DATA; } m_last_rowkey.copy((const char *)m_pk_packed_tuple, size, &my_charset_bin); rocksdb::Slice value = m_scan_it->value(); bool covered_lookup = m_keyread_only || m_key_descr_arr[keyno]->covers_lookup( table, &value, &m_lookup_bitmap); if (covered_lookup && m_lock_rows == RDB_LOCK_NONE) { rc = m_key_descr_arr[keyno]->unpack_record( table, buf, &key, &value, m_verify_row_debug_checksums); global_stats.covered_secondary_key_lookups.inc(); } else { rc = get_row_by_rowid(buf, m_pk_packed_tuple, size); } if (!rc) { table->status = 0; #ifdef MARIAROCKS_NOT_YET stats.rows_read++; stats.rows_index_next++; #endif update_row_stats(ROWS_READ); } return rc; } } return HA_ERR_END_OF_FILE; } bool ha_rocksdb::is_using_full_key(key_part_map keypart_map, uint actual_key_parts) { return (keypart_map == HA_WHOLE_KEY) || (keypart_map == ((key_part_map(1) << actual_key_parts) - 1)); } /* ha_rocksdb::read_range_first overrides handler::read_range_first. The only difference from handler::read_range_first is that ha_rocksdb::read_range_first passes end_key to ha_rocksdb::index_read_map_impl function. @return HA_EXIT_SUCCESS OK other HA_ERR error code (can be SE-specific) */ int ha_rocksdb::read_range_first(const key_range *const start_key, const key_range *const end_key, bool eq_range_arg, bool sorted) { DBUG_ENTER_FUNC(); int result; eq_range = eq_range_arg; /* MariaDB: Pass RANGE_SCAN_ASC when we support ICP on backwards scans */ set_end_range(end_key); range_key_part = table->key_info[active_index].key_part; if (!start_key) // Read first record result = ha_index_first(table->record[0]); else { #ifdef MARIAROCKS_NOT_YET if (is_using_prohibited_gap_locks( is_using_full_unique_key(active_index, start_key->keypart_map, start_key->flag))) { DBUG_RETURN(HA_ERR_LOCK_DEADLOCK); } #endif increment_statistics(&SSV::ha_read_key_count); MYSQL_TABLE_IO_WAIT(m_psi, PSI_TABLE_FETCH_ROW, active_index, 0, { result = index_read_map_impl(table->record[0], start_key->key, start_key->keypart_map, start_key->flag, end_key); }) } if (result) DBUG_RETURN((result == HA_ERR_KEY_NOT_FOUND) ? HA_ERR_END_OF_FILE : result); if (compare_key(end_range) <= 0) { DBUG_RETURN(HA_EXIT_SUCCESS); } else { /* The last read row does not fall in the range. So request storage engine to release row lock if possible. */ unlock_row(); DBUG_RETURN(HA_ERR_END_OF_FILE); } } int ha_rocksdb::prepare_index_scan() { range_key_part= table->key_info[active_index].key_part; set_end_range(NULL); return 0; } int ha_rocksdb::prepare_range_scan(const key_range *start_key, const key_range *end_key) { range_key_part= table->key_info[active_index].key_part; if (start_key) { m_save_start_range= *start_key; m_start_range= &m_save_start_range; } else m_start_range= NULL; set_end_range(end_key); return 0; } /** @return HA_EXIT_SUCCESS OK other HA_ERR error code (can be SE-specific) */ int ha_rocksdb::index_read_map(uchar *const buf, const uchar *const key, key_part_map keypart_map, enum ha_rkey_function find_flag) { DBUG_ENTER_FUNC(); DBUG_RETURN(index_read_map_impl(buf, key, keypart_map, find_flag, nullptr)); } /* See storage/rocksdb/rocksdb-range-access.txt for description of how MySQL index navigation commands are converted into RocksDB lookup commands. This function takes end_key as an argument, and it is set on range scan. MyRocks needs to decide whether prefix bloom filter can be used or not. To decide to use prefix bloom filter or not, calculating equal condition length is needed. On equal lookups (find_flag == HA_READ_KEY_EXACT), equal condition length is the same as rocksdb::Slice.size() of the start key. On range scan, equal condition length is MIN(start_key, end_key) of the rocksdb::Slice expression. @return HA_EXIT_SUCCESS OK other HA_ERR error code (can be SE-specific) */ int ha_rocksdb::index_read_map_impl(uchar *const buf, const uchar *const key, key_part_map keypart_map, enum ha_rkey_function find_flag, const key_range *end_key) { DBUG_ENTER_FUNC(); int rc = 0; const Rdb_key_def &kd = *m_key_descr_arr[active_index]; const uint actual_key_parts = kd.get_key_parts(); bool using_full_key = is_using_full_key(keypart_map, actual_key_parts); if (!end_key) end_key = end_range; /* By default, we don't need the retrieved records to match the prefix */ m_sk_match_prefix = nullptr; #ifdef MARIAROCKS_NOT_YET stats.rows_requested++; #endif if (active_index == table->s->primary_key && find_flag == HA_READ_KEY_EXACT && using_full_key) { /* Equality lookup over primary key, using full tuple. This is a special case, use DB::Get. */ const uint size = kd.pack_index_tuple(table, m_pack_buffer, m_pk_packed_tuple, key, keypart_map); bool skip_lookup = is_blind_delete_enabled(); rc = get_row_by_rowid(buf, m_pk_packed_tuple, size, skip_lookup, false); if (!rc && !skip_lookup) { #ifdef MARIAROCKS_NOT_YET stats.rows_read++; stats.rows_index_first++; #endif update_row_stats(ROWS_READ); } DBUG_RETURN(rc); } /* Unique secondary index performs lookups without the extended key fields */ uint packed_size; if (active_index != table->s->primary_key && table->key_info[active_index].flags & HA_NOSAME && find_flag == HA_READ_KEY_EXACT && using_full_key) { key_part_map tmp_map = (key_part_map(1) << table->key_info[active_index] .user_defined_key_parts) - 1; packed_size = kd.pack_index_tuple(table, m_pack_buffer, m_sk_packed_tuple, key, tmp_map); if (table->key_info[active_index].user_defined_key_parts != kd.get_key_parts()) using_full_key = false; } else { packed_size = kd.pack_index_tuple(table, m_pack_buffer, m_sk_packed_tuple, key, keypart_map); } if ((pushed_idx_cond && pushed_idx_cond_keyno == active_index) && (find_flag == HA_READ_KEY_EXACT || find_flag == HA_READ_PREFIX_LAST)) { /* We are doing a point index lookup, and ICP is enabled. It is possible that this call will be followed by ha_rocksdb->index_next_same() call. Do what InnoDB does: save the lookup tuple now. We will need it in index_next_same/find_icp_matching_index_rec in order to stop scanning as soon as index record doesn't match the lookup tuple. When not using ICP, handler::index_next_same() will make sure that rows that don't match the lookup prefix are not returned. row matches the lookup prefix. */ m_sk_match_prefix = m_sk_match_prefix_buf; m_sk_match_length = packed_size; memcpy(m_sk_match_prefix, m_sk_packed_tuple, packed_size); } int bytes_changed_by_succ = 0; if (find_flag == HA_READ_PREFIX_LAST_OR_PREV || find_flag == HA_READ_PREFIX_LAST || find_flag == HA_READ_AFTER_KEY) { /* See below */ bytes_changed_by_succ = kd.successor(m_sk_packed_tuple, packed_size); } rocksdb::Slice slice(reinterpret_cast(m_sk_packed_tuple), packed_size); uint end_key_packed_size = 0; /* In MariaDB, the end_key is always the bigger end of the range. If we are doing a reverse-ordered scan (that is, walking from the bigger key values to smaller), we should use the smaller end of range as end_key. */ const key_range *cur_end_key= end_key; if (find_flag == HA_READ_PREFIX_LAST_OR_PREV || find_flag == HA_READ_BEFORE_KEY) { cur_end_key= m_start_range; } const uint eq_cond_len = calc_eq_cond_len(kd, find_flag, slice, bytes_changed_by_succ, cur_end_key, &end_key_packed_size); bool use_all_keys = false; if (find_flag == HA_READ_KEY_EXACT && my_count_bits(keypart_map) == kd.get_key_parts()) use_all_keys = true; Rdb_transaction *const tx = get_or_create_tx(table->in_use); const bool is_new_snapshot = !tx->has_snapshot(); // Loop as long as we get a deadlock error AND we end up creating the // snapshot here (i.e. it did not exist prior to this) for (;;) { /* This will open the iterator and position it at a record that's equal or greater than the lookup tuple. */ setup_scan_iterator(kd, &slice, use_all_keys, eq_cond_len); /* Once we are positioned on from above, move to the position we really want: See storage/rocksdb/rocksdb-range-access.txt */ bool move_forward; rc = position_to_correct_key(kd, find_flag, using_full_key, key, keypart_map, slice, &move_forward, tx->m_snapshot_timestamp); if (rc) { /* This status is returned on any error */ table->status = STATUS_NOT_FOUND; DBUG_RETURN(rc); } m_skip_scan_it_next_call = false; /* Now get the data for the row into 'buf'. If we were using a primary key then we have all the rows we need. For a secondary key we now need to lookup the primary key. */ if (active_index == table->s->primary_key) rc = read_row_from_primary_key(buf); else rc = read_row_from_secondary_key(buf, kd, move_forward); if (rc != HA_ERR_ROCKSDB_STATUS_BUSY || !is_new_snapshot) break; /* Exit the loop */ // release the snapshot and iterator so they will be regenerated tx->release_snapshot(); release_scan_iterator(); } if (rc) { /* the only possible error condition is record-not-found */ table->status = STATUS_NOT_FOUND; } else { table->status = 0; #ifdef MARIAROCKS_NOT_YET stats.rows_read++; stats.rows_index_first++; #endif update_row_stats(ROWS_READ); } DBUG_RETURN(rc); } /* @brief Scan the secondary index until we find an index record that satisfies ICP @param move_forward TRUE <=> move m_scan_it forward FALSE <=> move m_scan_it backward @param buf Record buffer (must be the same buffer that pushed index condition points to, in practice it is table->record[0]) @detail Move the current iterator m_scan_it until we get an index tuple that satisfies the pushed Index Condition. (if there is no pushed index condition, return right away) @return 0 - Index tuple satisfies ICP, can do index read. other - error code */ int ha_rocksdb::find_icp_matching_index_rec(const bool &move_forward, uchar *const buf) { if (pushed_idx_cond && pushed_idx_cond_keyno == active_index) { const Rdb_key_def &kd = *m_key_descr_arr[active_index]; while (1) { rocksdb_skip_expired_records(kd, m_scan_it, !move_forward); if (!is_valid(m_scan_it)) { table->status = STATUS_NOT_FOUND; return HA_ERR_END_OF_FILE; } const rocksdb::Slice rkey = m_scan_it->key(); if (!kd.covers_key(rkey)) { table->status = STATUS_NOT_FOUND; return HA_ERR_END_OF_FILE; } if (m_sk_match_prefix) { const rocksdb::Slice prefix((const char *)m_sk_match_prefix, m_sk_match_length); if (!kd.value_matches_prefix(rkey, prefix)) { table->status = STATUS_NOT_FOUND; return HA_ERR_END_OF_FILE; } } const rocksdb::Slice value = m_scan_it->value(); int err = kd.unpack_record(table, buf, &rkey, &value, m_verify_row_debug_checksums); if (err != HA_EXIT_SUCCESS) { return err; } const enum icp_result icp_status= handler_index_cond_check(this); if (icp_status == ICP_NO_MATCH) { rocksdb_smart_next(!move_forward, m_scan_it); continue; /* Get the next (or prev) index tuple */ } else if (icp_status == ICP_OUT_OF_RANGE || icp_status == ICP_ABORTED_BY_USER) { /* We have walked out of range we are scanning */ table->status = STATUS_NOT_FOUND; return HA_ERR_END_OF_FILE; } else /* icp_status == ICP_MATCH */ { /* Index Condition is satisfied. We have rc==0, proceed to fetch the * row. */ break; } /* TODO: should we have this here, or RockDB handles this internally? if (my_core::thd_killed(current_thd)) { rc= HA_ERR_INTERNAL_ERROR; // doesn't matter break; } */ } } return HA_EXIT_SUCCESS; } /** @return HA_EXIT_SUCCESS OK other HA_ERR error code (can be SE-specific) */ /** @return HA_ADMIN_OK OK other HA_ADMIN error code */ int ha_rocksdb::check(THD *const thd, HA_CHECK_OPT *const check_opt) { DBUG_ENTER_FUNC(); DBUG_ASSERT(thd != nullptr); DBUG_ASSERT(check_opt != nullptr); const uint pk = pk_index(table, m_tbl_def); String rowkey_copy; String sec_key_copy; const char *const table_name = table->s->table_name.str; // Only when debugging: don't use snapshot when reading // Rdb_transaction *tx= get_or_create_tx(table->in_use); // tx->snapshot= nullptr; const bool save_verify_row_debug_checksums = m_verify_row_debug_checksums; m_verify_row_debug_checksums = true; /* For each secondary index, check that we can get a PK value from it */ // NO_LINT_DEBUG sql_print_verbose_info("CHECKTABLE %s: Checking table %s", table_name, table_name); ha_rows UNINIT_VAR(row_checksums_at_start); // set/used iff first_index==true ha_rows row_checksums = ha_rows(-1); bool first_index = true; for (uint keyno = 0; keyno < table->s->keys; keyno++) { if (keyno != pk) { extra(HA_EXTRA_KEYREAD); ha_index_init(keyno, true); ha_rows rows = 0; ha_rows checksums = 0; if (first_index) row_checksums_at_start = m_row_checksums_checked; int res; // NO_LINT_DEBUG sql_print_verbose_info("CHECKTABLE %s: Checking index %s", table_name, table->key_info[keyno].name); while (1) { if (!rows) res = index_first(table->record[0]); else res = index_next(table->record[0]); if (res == HA_ERR_END_OF_FILE) break; if (res) { // error // NO_LINT_DEBUG sql_print_error("CHECKTABLE %s: .. row %lld: index scan error %d", table_name, rows, res); goto error; } rocksdb::Slice key = m_scan_it->key(); sec_key_copy.copy(key.data(), key.size(), &my_charset_bin); rowkey_copy.copy(m_last_rowkey.ptr(), m_last_rowkey.length(), &my_charset_bin); if (m_key_descr_arr[keyno]->unpack_info_has_checksum( m_scan_it->value())) { checksums++; } if ((res = get_row_by_rowid(table->record[0], rowkey_copy.ptr(), rowkey_copy.length()))) { // NO_LINT_DEBUG sql_print_error("CHECKTABLE %s: .. row %lld: " "failed to fetch row by rowid", table_name, rows); goto error; } longlong hidden_pk_id = 0; if (has_hidden_pk(table) && read_hidden_pk_id_from_rowkey(&hidden_pk_id)) goto error; /* Check if we get the same PK value */ uint packed_size = m_pk_descr->pack_record( table, m_pack_buffer, table->record[0], m_pk_packed_tuple, nullptr, false, hidden_pk_id); if (packed_size != rowkey_copy.length() || memcmp(m_pk_packed_tuple, rowkey_copy.ptr(), packed_size)) { // NO_LINT_DEBUG sql_print_error("CHECKTABLE %s: .. row %lld: PK value mismatch", table_name, rows); goto print_and_error; } /* Check if we get the same secondary key value */ packed_size = m_key_descr_arr[keyno]->pack_record( table, m_pack_buffer, table->record[0], m_sk_packed_tuple, &m_sk_tails, false, hidden_pk_id); if (packed_size != sec_key_copy.length() || memcmp(m_sk_packed_tuple, sec_key_copy.ptr(), packed_size)) { // NO_LINT_DEBUG sql_print_error("CHECKTABLE %s: .. row %lld: " "secondary index value mismatch", table_name, rows); goto print_and_error; } rows++; continue; print_and_error : { std::string buf; buf = rdb_hexdump(rowkey_copy.ptr(), rowkey_copy.length(), RDB_MAX_HEXDUMP_LEN); // NO_LINT_DEBUG sql_print_error("CHECKTABLE %s: rowkey: %s", table_name, buf.c_str()); buf = rdb_hexdump(m_retrieved_record.data(), m_retrieved_record.size(), RDB_MAX_HEXDUMP_LEN); // NO_LINT_DEBUG sql_print_error("CHECKTABLE %s: record: %s", table_name, buf.c_str()); buf = rdb_hexdump(sec_key_copy.ptr(), sec_key_copy.length(), RDB_MAX_HEXDUMP_LEN); // NO_LINT_DEBUG sql_print_error("CHECKTABLE %s: index: %s", table_name, buf.c_str()); goto error; } } // NO_LINT_DEBUG sql_print_verbose_info("CHECKTABLE %s: ... %lld index entries checked " "(%lld had checksums)", table_name, rows, checksums); if (first_index) { row_checksums = m_row_checksums_checked - row_checksums_at_start; first_index = false; } ha_index_end(); } } if (row_checksums != ha_rows(-1)) { // NO_LINT_DEBUG sql_print_verbose_info("CHECKTABLE %s: %lld table records had checksums", table_name, row_checksums); } extra(HA_EXTRA_NO_KEYREAD); m_verify_row_debug_checksums = save_verify_row_debug_checksums; /* TODO: we should check also for PK records that are missing in the secondary indexes. For that, need to walk through the PK and check that every PK record has a proper counterpart in each secondary index. */ DBUG_RETURN(HA_ADMIN_OK); error: m_verify_row_debug_checksums = save_verify_row_debug_checksums; ha_index_or_rnd_end(); extra(HA_EXTRA_NO_KEYREAD); DBUG_RETURN(HA_ADMIN_CORRUPT); } static void dbug_dump_str(FILE *const out, const char *const str, int len) { fprintf(out, "\""); for (int i = 0; i < len; i++) { if (str[i] > 32) fprintf(out, "%c", str[i]); else fprintf(out, "\\%d", str[i]); } fprintf(out, "\""); } /* Debugging help: dump the whole database into a human-readable file. Usage: dbug_dump_database(rdb); */ void dbug_dump_database(rocksdb::DB *const db) { FILE *const out = fopen("/tmp/rocksdb.dump", "wt"); if (!out) return; rocksdb::Iterator *it = db->NewIterator(rocksdb::ReadOptions()); for (it->SeekToFirst(); it->Valid(); it->Next()) { rocksdb::Slice key = it->key(); rocksdb::Slice val = it->value(); dbug_dump_str(out, key.data(), key.size()); fprintf(out, " -> "); dbug_dump_str(out, val.data(), val.size()); fprintf(out, "\n"); } delete it; fclose(out); } rocksdb::Status ha_rocksdb::get_for_update( Rdb_transaction *const tx, rocksdb::ColumnFamilyHandle *const column_family, const rocksdb::Slice &key, rocksdb::PinnableSlice *const value) const { DBUG_ASSERT(m_lock_rows != RDB_LOCK_NONE); const bool exclusive = m_lock_rows != RDB_LOCK_READ; rocksdb::Status s = tx->get_for_update(column_family, key, value, exclusive); // If we have a lock conflict and we are running in READ COMMITTTED mode // release and reacquire the snapshot and then retry the get_for_update(). if (s.IsBusy() && !s.IsDeadlock() && my_core::thd_tx_isolation(ha_thd()) == ISO_READ_COMMITTED) { tx->release_snapshot(); tx->acquire_snapshot(false); s = tx->get_for_update(column_family, key, value, exclusive); } return s; } bool ha_rocksdb::is_blind_delete_enabled() { THD *thd = ha_thd(); return (THDVAR(thd, blind_delete_primary_key) && thd->lex->sql_command == SQLCOM_DELETE && thd->lex->table_count == 1 && table->s->keys == 1 && !has_hidden_pk(table) && !thd->rgi_slave); } /* Given a rowid (i.e. packed PK) as a parameter, get the record. @return HA_EXIT_SUCCESS OK other HA_ERR error code (can be SE-specific) */ int ha_rocksdb::get_row_by_rowid(uchar *const buf, const char *const rowid, const uint rowid_size, const bool skip_lookup, const bool skip_ttl_check) { DBUG_ENTER_FUNC(); DBUG_ASSERT(table != nullptr); int rc; rocksdb::Slice key_slice(rowid, rowid_size); Rdb_transaction *const tx = get_or_create_tx(table->in_use); DBUG_ASSERT(tx != nullptr); DEBUG_SYNC(ha_thd(), "rocksdb.get_row_by_rowid"); DBUG_EXECUTE_IF("dbug.rocksdb.get_row_by_rowid", { THD *thd = ha_thd(); const char act[] = "now signal Reached " "wait_for signal.rocksdb.get_row_by_rowid_let_running"; DBUG_ASSERT(opt_debug_sync_timeout > 0); DBUG_ASSERT(!debug_sync_set_action(thd, STRING_WITH_LEN(act))); };); bool found; rocksdb::Status s; /* Pretend row found without looking up */ if (skip_lookup) { #ifdef MARIAROCKS_NOT_YET stats.rows_deleted_blind++; #endif update_row_stats(ROWS_DELETED_BLIND); m_last_rowkey.copy((const char *)rowid, rowid_size, &my_charset_bin); table->status = 0; DBUG_RETURN(0); } if (m_lock_rows == RDB_LOCK_NONE) { tx->acquire_snapshot(true); s = tx->get(m_pk_descr->get_cf(), key_slice, &m_retrieved_record); } else { s = get_for_update(tx, m_pk_descr->get_cf(), key_slice, &m_retrieved_record); } DBUG_EXECUTE_IF("rocksdb_return_status_corrupted", dbug_change_status_to_corrupted(&s);); if (!s.IsNotFound() && !s.ok()) { DBUG_RETURN(tx->set_status_error(table->in_use, s, *m_pk_descr, m_tbl_def, m_table_handler)); } found = !s.IsNotFound(); table->status = STATUS_NOT_FOUND; if (found) { /* If we found the record, but it's expired, pretend we didn't find it. */ if (!skip_ttl_check && m_pk_descr->has_ttl() && should_hide_ttl_rec(*m_pk_descr, m_retrieved_record, tx->m_snapshot_timestamp)) { DBUG_RETURN(HA_ERR_KEY_NOT_FOUND); } m_last_rowkey.copy((const char *)rowid, rowid_size, &my_charset_bin); rc = convert_record_from_storage_format(&key_slice, buf); if (!rc) { table->status = 0; } } else { /* Note: we don't need to unlock the row. It is intentional that we keep locks on rows that don't exist. */ rc = HA_ERR_KEY_NOT_FOUND; } DBUG_RETURN(rc); } /** @return HA_EXIT_SUCCESS OK other HA_ERR error code (can be SE-specific) */ int ha_rocksdb::index_next(uchar *const buf) { DBUG_ENTER_FUNC(); bool moves_forward = true; if (m_key_descr_arr[active_index]->m_is_reverse_cf) { moves_forward = false; } int rc = index_next_with_direction(buf, moves_forward); if (rc == HA_ERR_KEY_NOT_FOUND) rc = HA_ERR_END_OF_FILE; DBUG_RETURN(rc); } /** @return HA_EXIT_SUCCESS OK other HA_ERR error code (can be SE-specific) */ int ha_rocksdb::index_prev(uchar *const buf) { DBUG_ENTER_FUNC(); bool moves_forward = false; if (m_key_descr_arr[active_index]->m_is_reverse_cf) { moves_forward = true; } int rc = index_next_with_direction(buf, moves_forward); if (rc == HA_ERR_KEY_NOT_FOUND) rc = HA_ERR_END_OF_FILE; DBUG_RETURN(rc); } int ha_rocksdb::index_next_with_direction(uchar *const buf, bool move_forward) { DBUG_ENTER_FUNC(); int rc; if (active_index == pk_index(table, m_tbl_def)) { rc = rnd_next_with_direction(buf, move_forward); } else { if (m_skip_scan_it_next_call) { m_skip_scan_it_next_call = false; } else { if (move_forward) m_scan_it->Next(); /* this call cannot fail */ else m_scan_it->Prev(); } rocksdb_skip_expired_records(*m_key_descr_arr[active_index], m_scan_it, !move_forward); rc = find_icp_matching_index_rec(move_forward, buf); if (!rc) rc = secondary_index_read(active_index, buf); } DBUG_RETURN(rc); } /** @return HA_EXIT_SUCCESS OK other HA_ERR error code (can be SE-specific) */ int ha_rocksdb::index_first(uchar *const buf) { DBUG_ENTER_FUNC(); m_sk_match_prefix = nullptr; int rc = m_key_descr_arr[active_index]->m_is_reverse_cf ? index_last_intern(buf) : index_first_intern(buf); if (rc == HA_ERR_KEY_NOT_FOUND) rc = HA_ERR_END_OF_FILE; DBUG_RETURN(rc); } /** @return HA_EXIT_SUCCESS OK other HA_ERR error code (can be SE-specific) */ int ha_rocksdb::index_last(uchar *const buf) { DBUG_ENTER_FUNC(); m_sk_match_prefix = nullptr; int rc = m_key_descr_arr[active_index]->m_is_reverse_cf ? index_first_intern(buf) : index_last_intern(buf); if (rc == HA_ERR_KEY_NOT_FOUND) rc = HA_ERR_END_OF_FILE; DBUG_RETURN(rc); } /* Start scanning from the "first" value. The 'first' here means "the first from start of the key space". For reverse-ordered key spaces, the first value will be the biggest, 'zzzz'. An picture of a forward-ordered keyspace (remember, the keys have form 'indexnr-keyval'. Suppose the index we are at has number n) (n-1) - ... ( n ) <--- 1. (n) doesn't exist in the db but it would be here. ( n ) - aaa <--- 2. Seek("n") will put us here on the first index ( n ) - bbb record. ( n ) - cc So, need to do: Seek(n); A backward-ordered keyspace: (n+1) - bbb (n+1) - aaa (n+1) <--- (n+1) doesn't exist in the db but would be here. ( n ) - ccc <--- 1. We need to be here. ( n ) - bbb ( n ) - aaa ( n ) So, need to: Seek(n+1); */ int ha_rocksdb::index_first_intern(uchar *const buf) { DBUG_ENTER_FUNC(); uchar *key; uint key_size; int rc; if (is_pk(active_index, table, m_tbl_def)) { key = m_pk_packed_tuple; } else { key = m_sk_packed_tuple; } DBUG_ASSERT(key != nullptr); const Rdb_key_def &kd = *m_key_descr_arr[active_index]; int key_start_matching_bytes = kd.get_first_key(key, &key_size); rocksdb::Slice index_key((const char *)key, key_size); Rdb_transaction *const tx = get_or_create_tx(table->in_use); DBUG_ASSERT(tx != nullptr); const bool is_new_snapshot = !tx->has_snapshot(); // Loop as long as we get a deadlock error AND we end up creating the // snapshot here (i.e. it did not exist prior to this) for (;;) { setup_scan_iterator(kd, &index_key, false, key_start_matching_bytes); m_scan_it->Seek(index_key); m_skip_scan_it_next_call = true; rc = index_next_with_direction(buf, true); if (rc != HA_ERR_ROCKSDB_STATUS_BUSY || !is_new_snapshot) break; // exit the loop // release the snapshot and iterator so they will be regenerated tx->release_snapshot(); release_scan_iterator(); } if (!rc) { /* index_next is always incremented on success, so decrement if it is index_first instead */ #ifdef MARIAROCKS_NOT_YET stats.rows_index_first++; stats.rows_index_next--; #endif } DBUG_RETURN(rc); } /** @details Start scanning from the "last" value The 'last' here means "the last from start of the key space". For reverse-ordered key spaces, we will actually read the smallest value. An picture of a forward-ordered keyspace (remember, the keys have form 'indexnr-keyval'. Suppose the we are at a key that has number n) (n-1)-something ( n )-aaa ( n )-bbb ( n )-ccc <----------- Need to seek to here. (n+1) <---- Doesn't exist, but would be here. (n+1)-smth, or no value at all RocksDB's Iterator::SeekForPrev($val) seeks to "at $val or last value that's smaller". We can't seek to "(n)-ccc" directly, because we don't know what is the value of 'ccc' (the biggest record with prefix (n)). Instead, we seek to "(n+1)", which is the least possible value that's greater than any value in index #n. So, need to: it->SeekForPrev(n+1) A backward-ordered keyspace: (n+1)-something ( n ) - ccc ( n ) - bbb ( n ) - aaa <---------------- (*) Need to seek here. ( n ) <--- Doesn't exist, but would be here. (n-1)-smth, or no value at all So, need to: it->SeekForPrev(n) */ int ha_rocksdb::index_last_intern(uchar *const buf) { DBUG_ENTER_FUNC(); uchar *key; uint key_size; int rc; if (is_pk(active_index, table, m_tbl_def)) { key = m_pk_packed_tuple; } else { key = m_sk_packed_tuple; } DBUG_ASSERT(key != nullptr); const Rdb_key_def &kd = *m_key_descr_arr[active_index]; if (kd.m_is_reverse_cf) { kd.get_infimum_key(key, &key_size); } else { kd.get_supremum_key(key, &key_size); } rocksdb::Slice index_key((const char *)key, key_size); Rdb_transaction *const tx = get_or_create_tx(table->in_use); DBUG_ASSERT(tx != nullptr); bool is_new_snapshot = !tx->has_snapshot(); // Loop as long as we get a deadlock error AND we end up creating the // snapshot here (i.e. it did not exist prior to this) for (;;) { setup_scan_iterator(kd, &index_key, false, Rdb_key_def::INDEX_NUMBER_SIZE); m_scan_it->SeekForPrev(index_key); m_skip_scan_it_next_call = false; if (is_pk(active_index, table, m_tbl_def)) { m_skip_scan_it_next_call = true; rc = rnd_next_with_direction(buf, false); } else { rc = find_icp_matching_index_rec(false /*move_forward*/, buf); if (!rc) rc = secondary_index_read(active_index, buf); } if (rc != HA_ERR_ROCKSDB_STATUS_BUSY || !is_new_snapshot) break; /* exit the loop */ // release the snapshot and iterator so they will be regenerated tx->release_snapshot(); release_scan_iterator(); } if (!rc) { /* index_next is always incremented on success, so decrement if it is index_first instead */ #ifdef MARIAROCKS_NOT_YET stats.rows_index_first++; stats.rows_index_next--; #endif } DBUG_RETURN(rc); } void ha_rocksdb::unlock_row() { DBUG_ENTER_FUNC(); if (m_lock_rows != RDB_LOCK_NONE) { Rdb_transaction *const tx = get_or_create_tx(table->in_use); tx->release_lock(m_pk_descr->get_cf(), std::string(m_last_rowkey.ptr(), m_last_rowkey.length())); } DBUG_VOID_RETURN; } /* Returning true if SingleDelete can be used. - Secondary Indexes can always use SingleDelete. - If the index is PRIMARY KEY, and if all of the columns of the table are covered by the PRIMARY KEY, SingleDelete can be used. */ bool ha_rocksdb::can_use_single_delete(const uint &index) const { return (index != pk_index(table, m_tbl_def) || (!has_hidden_pk(table) && table->key_info[index].ext_key_parts == table->s->fields)); } bool ha_rocksdb::skip_unique_check() const { /* We want to skip unique checks if: 1) bulk_load is on 2) this table is in the whitelist of tables to skip and the replication lag has reached a large enough value (see unique_check_lag_threshold and unique_check_lage_reset_threshold) 3) the user set unique_checks option to 0, and the table does not have any indexes. If the table has secondary keys, then those might becomes inconsisted/corrupted */ return THDVAR(table->in_use, bulk_load) || (m_force_skip_unique_check && m_skip_unique_check) || (my_core::thd_test_options(table->in_use, OPTION_RELAXED_UNIQUE_CHECKS) && m_tbl_def->m_key_count == 1); } #ifdef MARIAROCKS_NOT_YET // MDEV-10975 void ha_rocksdb::set_force_skip_unique_check(bool skip) { DBUG_ENTER_FUNC(); m_force_skip_unique_check = skip; DBUG_VOID_RETURN; } #endif bool ha_rocksdb::commit_in_the_middle() { return THDVAR(table->in_use, bulk_load) || THDVAR(table->in_use, commit_in_the_middle); } /* Executing bulk commit if it should. @retval true if bulk commit failed @retval false if bulk commit was skipped or succeeded */ bool ha_rocksdb::do_bulk_commit(Rdb_transaction *const tx) { return commit_in_the_middle() && tx->get_write_count() >= THDVAR(table->in_use, bulk_load_size) && tx->flush_batch(); } /* If table was created without primary key, SQL layer represents the primary key number as MAX_INDEXES. Hence, this function returns true if the table does not contain a primary key. (In which case we generate a hidden 'auto-incremented' pk.) */ bool ha_rocksdb::has_hidden_pk(const TABLE *const table) const { return Rdb_key_def::table_has_hidden_pk(table); } /* Returns true if given index number is a hidden_pk. - This is used when a table is created with no primary key. */ bool ha_rocksdb::is_hidden_pk(const uint index, const TABLE *const table_arg, const Rdb_tbl_def *const tbl_def_arg) { DBUG_ASSERT(table_arg->s != nullptr); return (table_arg->s->primary_key == MAX_INDEXES && index == tbl_def_arg->m_key_count - 1); } /* Returns index of primary key */ uint ha_rocksdb::pk_index(const TABLE *const table_arg, const Rdb_tbl_def *const tbl_def_arg) { DBUG_ASSERT(table_arg->s != nullptr); return table_arg->s->primary_key == MAX_INDEXES ? tbl_def_arg->m_key_count - 1 : table_arg->s->primary_key; } /* Returns true if given index number is a primary key */ bool ha_rocksdb::is_pk(const uint index, const TABLE *const table_arg, const Rdb_tbl_def *const tbl_def_arg) { DBUG_ASSERT(table_arg->s != nullptr); return index == table_arg->s->primary_key || is_hidden_pk(index, table_arg, tbl_def_arg); } uint ha_rocksdb::max_supported_key_part_length() const { DBUG_ENTER_FUNC(); DBUG_RETURN(rocksdb_large_prefix ? MAX_INDEX_COL_LEN_LARGE : MAX_INDEX_COL_LEN_SMALL); } const char *ha_rocksdb::get_key_name(const uint index, const TABLE *const table_arg, const Rdb_tbl_def *const tbl_def_arg) { if (is_hidden_pk(index, table_arg, tbl_def_arg)) { return HIDDEN_PK_NAME; } DBUG_ASSERT(table_arg->key_info != nullptr); DBUG_ASSERT(table_arg->key_info[index].name.str != nullptr); return table_arg->key_info[index].name.str; } const char *ha_rocksdb::get_key_comment(const uint index, const TABLE *const table_arg, const Rdb_tbl_def *const tbl_def_arg) { if (is_hidden_pk(index, table_arg, tbl_def_arg)) { return nullptr; } DBUG_ASSERT(table_arg->key_info != nullptr); return table_arg->key_info[index].comment.str; } const std::string ha_rocksdb::generate_cf_name(const uint index, const TABLE *const table_arg, const Rdb_tbl_def *const tbl_def_arg, bool *per_part_match_found) { DBUG_ASSERT(table_arg != nullptr); DBUG_ASSERT(tbl_def_arg != nullptr); DBUG_ASSERT(per_part_match_found != nullptr); // When creating CF-s the caller needs to know if there was a custom CF name // specified for a given paritition. *per_part_match_found = false; // Index comment is used to define the column family name specification(s). // If there was no comment, we get an emptry string, and it means "use the // default column family". const char *const comment = get_key_comment(index, table_arg, tbl_def_arg); // `get_key_comment` can return `nullptr`, that's why this. std::string key_comment = comment ? comment : ""; std::string cf_name = Rdb_key_def::parse_comment_for_qualifier( key_comment, table_arg, tbl_def_arg, per_part_match_found, RDB_CF_NAME_QUALIFIER); if (IF_PARTITIONING(table_arg->part_info,nullptr) != nullptr && !*per_part_match_found) { // At this point we tried to search for a custom CF name for a partition, // but none was specified. Therefore default one will be used. return ""; } // If we didn't find any partitioned/non-partitioned qualifiers, return the // comment itself. NOTE: this currently handles returning the cf name // specified in the index comment in the case of no partitions, which doesn't // use any qualifiers at the moment. (aka its a special case) if (cf_name.empty() && !key_comment.empty()) { return key_comment; } return cf_name; } const std::string ha_rocksdb::get_table_comment(const TABLE *const table_arg) { DBUG_ASSERT(table_arg->s != nullptr); return table_arg->s->comment.str; } /** @return HA_EXIT_SUCCESS OK other HA_ERR error code (can be SE-specific) */ int ha_rocksdb::write_row(uchar *const buf) { DBUG_ENTER_FUNC(); DBUG_ASSERT(buf != nullptr); DBUG_ASSERT(buf == table->record[0]); DBUG_ASSERT(m_lock_rows == RDB_LOCK_WRITE); /* Note: "buf == table->record[0]" is copied from innodb. I am not aware of any use cases where this condition is not true. */ if (table->next_number_field && buf == table->record[0]) { int err; if ((err = update_auto_increment())) { DBUG_RETURN(err); } } const int rv = update_write_row(nullptr, buf, skip_unique_check()); if (rv == 0) { #ifdef MARIAROCKS_NOT_YET stats.rows_inserted++; #endif update_row_stats(ROWS_INSERTED); } DBUG_RETURN(rv); } /** Constructing m_last_rowkey (MyRocks key expression) from before_update|delete image (MySQL row expression). m_last_rowkey is normally set during lookup phase, such as rnd_next_with_direction() and rnd_pos(). With Read Free Replication, these read functions are skipped and update_rows(), delete_rows() are called without setting m_last_rowkey. This function sets m_last_rowkey for Read Free Replication. */ void ha_rocksdb::set_last_rowkey(const uchar *const old_data) { #ifdef MARIAROCKS_NOT_YET if (old_data && use_read_free_rpl()) { const int old_pk_size = m_pk_descr->pack_record( table, m_pack_buffer, old_data, m_pk_packed_tuple, nullptr, false); m_last_rowkey.copy((const char *)m_pk_packed_tuple, old_pk_size, &my_charset_bin); } #endif } int ha_rocksdb::get_pk_for_update(struct update_row_info *const row_info) { int size; /* Get new row key for any insert, and any update where the pk is not hidden. Row key for updates with hidden pk is handled below. */ if (!has_hidden_pk(table)) { row_info->hidden_pk_id = 0; row_info->new_pk_unpack_info = &m_pk_unpack_info; size = m_pk_descr->pack_record(table, m_pack_buffer, row_info->new_data, m_pk_packed_tuple, row_info->new_pk_unpack_info, false, 0, 0, nullptr, &row_info->ttl_pk_offset); } else if (row_info->old_data == nullptr) { row_info->hidden_pk_id = update_hidden_pk_val(); size = m_pk_descr->pack_hidden_pk(row_info->hidden_pk_id, m_pk_packed_tuple); } else { /* If hidden primary key, rowkey for new record will always be the same as before */ size = row_info->old_pk_slice.size(); memcpy(m_pk_packed_tuple, row_info->old_pk_slice.data(), size); int err = read_hidden_pk_id_from_rowkey(&row_info->hidden_pk_id); if (err) { return err; } } row_info->new_pk_slice = rocksdb::Slice((const char *)m_pk_packed_tuple, size); return HA_EXIT_SUCCESS; } int ha_rocksdb::check_and_lock_unique_pk(const uint &key_id, const struct update_row_info &row_info, bool *const found, bool *const pk_changed) { DBUG_ASSERT(found != nullptr); DBUG_ASSERT(pk_changed != nullptr); *pk_changed = false; /* For UPDATEs, if the key has changed, we need to obtain a lock. INSERTs always require locking. */ if (row_info.old_pk_slice.size() > 0) { /* If the keys are the same, then no lock is needed */ if (!row_info.new_pk_slice.compare(row_info.old_pk_slice)) { *found = false; return HA_EXIT_SUCCESS; } *pk_changed = true; } /* Perform a read to determine if a duplicate entry exists. For primary keys, a point lookup will be sufficient. note: we intentionally don't set options.snapshot here. We want to read the latest committed data. */ /* To prevent race conditions like below, it is necessary to take a lock for a target row. get_for_update() holds a gap lock if target key does not exist, so below conditions should never happen. 1) T1 Get(empty) -> T2 Get(empty) -> T1 Put(insert) -> T1 commit -> T2 Put(overwrite) -> T2 commit 2) T1 Get(empty) -> T1 Put(insert, not committed yet) -> T2 Get(empty) -> T2 Put(insert, blocked) -> T1 commit -> T2 commit(overwrite) */ const rocksdb::Status s = get_for_update(row_info.tx, m_pk_descr->get_cf(), row_info.new_pk_slice, &m_retrieved_record); if (!s.ok() && !s.IsNotFound()) { return row_info.tx->set_status_error( table->in_use, s, *m_key_descr_arr[key_id], m_tbl_def, m_table_handler); } *found = !s.IsNotFound(); return HA_EXIT_SUCCESS; } int ha_rocksdb::check_and_lock_sk(const uint &key_id, const struct update_row_info &row_info, bool *const found) { DBUG_ASSERT(found != nullptr); *found = false; /* Can skip checking this key if none of the key fields have changed. */ if (row_info.old_data != nullptr && !m_update_scope.is_set(key_id)) { return HA_EXIT_SUCCESS; } KEY *key_info = nullptr; uint n_null_fields = 0; uint user_defined_key_parts = 1; key_info = &table->key_info[key_id]; user_defined_key_parts = key_info->user_defined_key_parts; /* If there are no uniqueness requirements, there's no need to obtain a lock for this key. */ if (!(key_info->flags & HA_NOSAME)) { return HA_EXIT_SUCCESS; } const Rdb_key_def &kd = *m_key_descr_arr[key_id]; /* Calculate the new key for obtaining the lock For unique secondary indexes, the key used for locking does not include the extended fields. */ int size = kd.pack_record(table, m_pack_buffer, row_info.new_data, m_sk_packed_tuple, nullptr, false, 0, user_defined_key_parts, &n_null_fields); if (n_null_fields > 0) { /* If any fields are marked as NULL this will never match another row as to NULL never matches anything else including another NULL. */ return HA_EXIT_SUCCESS; } const rocksdb::Slice new_slice = rocksdb::Slice((const char *)m_sk_packed_tuple, size); /* Acquire lock on the old key in case of UPDATE */ if (row_info.old_data != nullptr) { size = kd.pack_record(table, m_pack_buffer, row_info.old_data, m_sk_packed_tuple_old, nullptr, false, 0, user_defined_key_parts); const rocksdb::Slice old_slice = rocksdb::Slice((const char *)m_sk_packed_tuple_old, size); const rocksdb::Status s = get_for_update(row_info.tx, kd.get_cf(), old_slice, nullptr); if (!s.ok()) { return row_info.tx->set_status_error(table->in_use, s, kd, m_tbl_def, m_table_handler); } /* If the old and new keys are the same we're done since we've already taken the lock on the old key */ if (!new_slice.compare(old_slice)) { return HA_EXIT_SUCCESS; } } /* Perform a read to determine if a duplicate entry exists - since this is a secondary indexes a range scan is needed. note: we intentionally don't set options.snapshot here. We want to read the latest committed data. */ const bool all_parts_used = (user_defined_key_parts == kd.get_key_parts()); /* This iterator seems expensive since we need to allocate and free memory for each unique index. If this needs to be optimized, for keys without NULL fields, the extended primary key fields can be migrated to the value portion of the key. This enables using Get() instead of Seek() as in the primary key case. The bloom filter may need to be disabled for this lookup. */ uchar lower_bound_buf[Rdb_key_def::INDEX_NUMBER_SIZE]; uchar upper_bound_buf[Rdb_key_def::INDEX_NUMBER_SIZE]; rocksdb::Slice lower_bound_slice; rocksdb::Slice upper_bound_slice; const bool total_order_seek = !check_bloom_and_set_bounds( ha_thd(), kd, new_slice, all_parts_used, Rdb_key_def::INDEX_NUMBER_SIZE, lower_bound_buf, upper_bound_buf, &lower_bound_slice, &upper_bound_slice); const bool fill_cache = !THDVAR(ha_thd(), skip_fill_cache); const rocksdb::Status s = get_for_update(row_info.tx, kd.get_cf(), new_slice, nullptr); if (!s.ok() && !s.IsNotFound()) { return row_info.tx->set_status_error(table->in_use, s, kd, m_tbl_def, m_table_handler); } rocksdb::Iterator *const iter = row_info.tx->get_iterator( kd.get_cf(), total_order_seek, fill_cache, lower_bound_slice, upper_bound_slice, true /* read current data */, false /* acquire snapshot */); /* Need to scan the transaction to see if there is a duplicate key. Also need to scan RocksDB and verify the key has not been deleted in the transaction. */ iter->Seek(new_slice); *found = !read_key_exact(kd, iter, all_parts_used, new_slice, row_info.tx->m_snapshot_timestamp); delete iter; return HA_EXIT_SUCCESS; } int ha_rocksdb::check_uniqueness_and_lock( const struct update_row_info &row_info, bool *const pk_changed) { /* Go through each index and determine if the index has uniqueness requirements. If it does, then try to obtain a row lock on the new values. Once all locks have been obtained, then perform the changes needed to update/insert the row. */ for (uint key_id = 0; key_id < m_tbl_def->m_key_count; key_id++) { bool found; int rc; if (is_pk(key_id, table, m_tbl_def)) { rc = check_and_lock_unique_pk(key_id, row_info, &found, pk_changed); } else { rc = check_and_lock_sk(key_id, row_info, &found); } if (rc != HA_EXIT_SUCCESS) { return rc; } /* If the pk key has ttl, we may need to pretend the row wasn't found if it is already expired. The pk record is read into m_retrieved_record by check_and_lock_unique_pk(). */ if (is_pk(key_id, table, m_tbl_def) && found && m_pk_descr->has_ttl() && should_hide_ttl_rec(*m_pk_descr, m_retrieved_record, (row_info.tx->m_snapshot_timestamp ? row_info.tx->m_snapshot_timestamp : static_cast(std::time(nullptr))))) { found = false; } if (found) { /* There is a row with this key already, so error out. */ errkey = key_id; m_dupp_errkey = errkey; return HA_ERR_FOUND_DUPP_KEY; } } return HA_EXIT_SUCCESS; } int ha_rocksdb::check_duplicate_sk(const TABLE *table_arg, const Rdb_key_def &index, const rocksdb::Slice *key, struct unique_sk_buf_info *sk_info) { uint n_null_fields = 0; const rocksdb::Comparator *index_comp = index.get_cf()->GetComparator(); /* Get proper SK buffer. */ uchar *sk_buf = sk_info->swap_and_get_sk_buf(); /* Get memcmp form of sk without extended pk tail */ uint sk_memcmp_size = index.get_memcmp_sk_parts(table_arg, *key, sk_buf, &n_null_fields); sk_info->sk_memcmp_key = rocksdb::Slice(reinterpret_cast(sk_buf), sk_memcmp_size); if (sk_info->sk_memcmp_key_old.size() > 0 && n_null_fields == 0 && index_comp->Compare(sk_info->sk_memcmp_key, sk_info->sk_memcmp_key_old) == 0) { return 1; } sk_info->sk_memcmp_key_old = sk_info->sk_memcmp_key; return 0; } int ha_rocksdb::bulk_load_key(Rdb_transaction *const tx, const Rdb_key_def &kd, const rocksdb::Slice &key, const rocksdb::Slice &value, bool sort) { DBUG_ENTER_FUNC(); int res; rocksdb::ColumnFamilyHandle *cf = kd.get_cf(); // In the case of unsorted inserts, m_sst_info allocated here is not // used to store the keys. It is still used to indicate when tables // are switched. if (m_sst_info == nullptr || m_sst_info->is_committed()) { m_sst_info.reset(new Rdb_sst_info(rdb, m_table_handler->m_table_name, kd.get_name(), cf, *rocksdb_db_options, THDVAR(ha_thd(), trace_sst_api))); res = tx->start_bulk_load(this, m_sst_info); if (res != HA_EXIT_SUCCESS) { DBUG_RETURN(res); } } DBUG_ASSERT(m_sst_info); if (sort) { Rdb_index_merge *key_merge; DBUG_ASSERT(cf != nullptr); res = tx->get_key_merge(kd.get_gl_index_id(), cf, &key_merge); if (res == HA_EXIT_SUCCESS) { res = key_merge->add(key, value); } } else { res = m_sst_info->put(key, value); } DBUG_RETURN(res); } int ha_rocksdb::finalize_bulk_load(bool print_client_error) { DBUG_ENTER_FUNC(); int res = HA_EXIT_SUCCESS; /* Skip if there are no possible ongoing bulk loads */ if (m_sst_info) { res = m_sst_info->commit(print_client_error); m_sst_info.reset(); } DBUG_RETURN(res); } int ha_rocksdb::update_pk(const Rdb_key_def &kd, const struct update_row_info &row_info, const bool &pk_changed) { const uint key_id = kd.get_keyno(); const bool hidden_pk = is_hidden_pk(key_id, table, m_tbl_def); ulonglong bytes_written = 0; /* If the PK has changed, or if this PK uses single deletes and this is an update, the old key needs to be deleted. In the single delete case, it might be possible to have this sequence of keys: PUT(X), PUT(X), SD(X), resulting in the first PUT(X) showing up. */ if (!hidden_pk && (pk_changed || ((row_info.old_pk_slice.size() > 0) && can_use_single_delete(key_id)))) { const rocksdb::Status s = delete_or_singledelete( key_id, row_info.tx, kd.get_cf(), row_info.old_pk_slice); if (!s.ok()) { return row_info.tx->set_status_error(table->in_use, s, kd, m_tbl_def, m_table_handler); } else { bytes_written = row_info.old_pk_slice.size(); } } if (table->found_next_number_field) { update_auto_incr_val_from_field(); } int rc = HA_EXIT_SUCCESS; rocksdb::Slice value_slice; /* Prepare the new record to be written into RocksDB */ if ((rc = convert_record_to_storage_format(row_info, &value_slice))) { return rc; } const auto cf = m_pk_descr->get_cf(); if (rocksdb_enable_bulk_load_api && THDVAR(table->in_use, bulk_load) && !hidden_pk) { /* Write the primary key directly to an SST file using an SstFileWriter */ rc = bulk_load_key(row_info.tx, kd, row_info.new_pk_slice, value_slice, THDVAR(table->in_use, bulk_load_allow_unsorted)); } else if (row_info.skip_unique_check || row_info.tx->m_ddl_transaction) { /* It is responsibility of the user to make sure that the data being inserted doesn't violate any unique keys. */ row_info.tx->get_indexed_write_batch()->Put(cf, row_info.new_pk_slice, value_slice); } else { const auto s = row_info.tx->put(cf, row_info.new_pk_slice, value_slice); if (!s.ok()) { if (s.IsBusy()) { errkey = table->s->primary_key; m_dupp_errkey = errkey; rc = HA_ERR_FOUND_DUPP_KEY; } else { rc = row_info.tx->set_status_error(table->in_use, s, *m_pk_descr, m_tbl_def, m_table_handler); } } } if (rc == HA_EXIT_SUCCESS) { row_info.tx->update_bytes_written( bytes_written + row_info.new_pk_slice.size() + value_slice.size()); } return rc; } int ha_rocksdb::update_sk(const TABLE *const table_arg, const Rdb_key_def &kd, const struct update_row_info &row_info, const bool bulk_load_sk) { int new_packed_size; int old_packed_size; int rc = HA_EXIT_SUCCESS; rocksdb::Slice new_key_slice; rocksdb::Slice new_value_slice; rocksdb::Slice old_key_slice; const uint key_id = kd.get_keyno(); ulonglong bytes_written = 0; /* Can skip updating this key if none of the key fields have changed and, if this table has TTL, the TTL timestamp has not changed. */ if (row_info.old_data != nullptr && !m_update_scope.is_set(key_id) && (!kd.has_ttl() || !m_ttl_bytes_updated)) { return HA_EXIT_SUCCESS; } const bool store_row_debug_checksums = should_store_row_debug_checksums(); new_packed_size = kd.pack_record(table_arg, m_pack_buffer, row_info.new_data, m_sk_packed_tuple, &m_sk_tails, store_row_debug_checksums, row_info.hidden_pk_id, 0, nullptr, nullptr, m_ttl_bytes); if (row_info.old_data != nullptr) { // The old value old_packed_size = kd.pack_record( table_arg, m_pack_buffer, row_info.old_data, m_sk_packed_tuple_old, &m_sk_tails_old, store_row_debug_checksums, row_info.hidden_pk_id, 0, nullptr, nullptr, m_ttl_bytes); /* Check if we are going to write the same value. This can happen when one does UPDATE tbl SET col='foo' and we are looking at the row that already has col='foo'. We also need to compare the unpack info. Suppose, the collation is case-insensitive, and unpack info contains information about whether the letters were uppercase and lowercase. Then, both 'foo' and 'FOO' will have the same key value, but different data in unpack_info. (note: anyone changing bytewise_compare should take this code into account) */ if (old_packed_size == new_packed_size && m_sk_tails_old.get_current_pos() == m_sk_tails.get_current_pos() && !(kd.has_ttl() && m_ttl_bytes_updated) && memcmp(m_sk_packed_tuple_old, m_sk_packed_tuple, old_packed_size) == 0 && memcmp(m_sk_tails_old.ptr(), m_sk_tails.ptr(), m_sk_tails.get_current_pos()) == 0) { return HA_EXIT_SUCCESS; } /* Deleting entries from secondary index should skip locking, but be visible to the transaction. (also note that DDL statements do not delete rows, so this is not a DDL statement) */ old_key_slice = rocksdb::Slice( reinterpret_cast(m_sk_packed_tuple_old), old_packed_size); row_info.tx->get_indexed_write_batch()->SingleDelete(kd.get_cf(), old_key_slice); bytes_written = old_key_slice.size(); } new_key_slice = rocksdb::Slice( reinterpret_cast(m_sk_packed_tuple), new_packed_size); new_value_slice = rocksdb::Slice(reinterpret_cast(m_sk_tails.ptr()), m_sk_tails.get_current_pos()); if (bulk_load_sk && row_info.old_data == nullptr) { rc = bulk_load_key(row_info.tx, kd, new_key_slice, new_value_slice, true); } else { row_info.tx->get_indexed_write_batch()->Put(kd.get_cf(), new_key_slice, new_value_slice); } row_info.tx->update_bytes_written(bytes_written + new_key_slice.size() + new_value_slice.size()); return rc; } int ha_rocksdb::update_indexes(const struct update_row_info &row_info, const bool &pk_changed) { int rc; bool bulk_load_sk; // The PK must be updated first to pull out the TTL value. rc = update_pk(*m_pk_descr, row_info, pk_changed); if (rc != HA_EXIT_SUCCESS) { return rc; } // Update the remaining indexes. Allow bulk loading only if // allow_sk is enabled bulk_load_sk = rocksdb_enable_bulk_load_api && THDVAR(table->in_use, bulk_load) && THDVAR(table->in_use, bulk_load_allow_sk); for (uint key_id = 0; key_id < m_tbl_def->m_key_count; key_id++) { if (is_pk(key_id, table, m_tbl_def)) { continue; } rc = update_sk(table, *m_key_descr_arr[key_id], row_info, bulk_load_sk); if (rc != HA_EXIT_SUCCESS) { return rc; } } return HA_EXIT_SUCCESS; } int ha_rocksdb::update_write_row(const uchar *const old_data, const uchar *const new_data, const bool skip_unique_check) { DBUG_ENTER_FUNC(); bool pk_changed = false; struct update_row_info row_info; row_info.old_data = old_data; row_info.new_data = new_data; row_info.skip_unique_check = skip_unique_check; row_info.new_pk_unpack_info = nullptr; set_last_rowkey(old_data); row_info.tx = get_or_create_tx(table->in_use); if (old_data != nullptr) { row_info.old_pk_slice = rocksdb::Slice(m_last_rowkey.ptr(), m_last_rowkey.length()); /* Determine which indexes need updating. */ calc_updated_indexes(); } /* Get the new row key into row_info.new_pk_slice */ int rc = get_pk_for_update(&row_info); if (rc != HA_EXIT_SUCCESS) { DBUG_RETURN(rc); } if (!skip_unique_check) { /* Check to see if we are going to have failures because of unique keys. Also lock the appropriate key values. */ rc = check_uniqueness_and_lock(row_info, &pk_changed); if (rc != HA_EXIT_SUCCESS) { DBUG_RETURN(rc); } } DEBUG_SYNC(ha_thd(), "rocksdb.update_write_row_after_unique_check"); /* At this point, all locks have been obtained, and all checks for duplicate keys have been performed. No further errors can be allowed to occur from here because updates to the transaction will be made and those updates cannot be easily removed without rolling back the entire transaction. */ rc = update_indexes(row_info, pk_changed); if (rc != HA_EXIT_SUCCESS) { DBUG_RETURN(rc); } if (old_data != nullptr) { row_info.tx->incr_update_count(); } else { row_info.tx->incr_insert_count(); } if (do_bulk_commit(row_info.tx)) { DBUG_RETURN(HA_ERR_ROCKSDB_BULK_LOAD); } DBUG_RETURN(HA_EXIT_SUCCESS); } /* Setting iterator upper/lower bounds for Seek/SeekForPrev. This makes RocksDB to avoid scanning tombstones outside of the given key ranges, when prefix_same_as_start=true was not passed (when prefix bloom filter can not be used). Inversing upper/lower bound is necessary on reverse order CF. This covers HA_READ_PREFIX_LAST* case as well. For example, if given query eq condition was 12 bytes and condition was 0x0000b3eb003f65c5e78858b8, and if doing HA_READ_PREFIX_LAST, eq_cond_len was 11 (see calc_eq_cond_len() for details). If the index was reverse order, upper bound would be 0x0000b3eb003f65c5e78857, and lower bound would be 0x0000b3eb003f65c5e78859. These cover given eq condition range. @param lower_bound_buf IN Buffer for lower bound @param upper_bound_buf IN Buffer for upper bound @param outer_u */ void ha_rocksdb::setup_iterator_bounds( const Rdb_key_def &kd, const rocksdb::Slice &eq_cond, size_t bound_len, uchar *const lower_bound, uchar *const upper_bound, rocksdb::Slice *lower_bound_slice, rocksdb::Slice *upper_bound_slice) { uint min_len = std::min(eq_cond.size(), bound_len); memcpy(upper_bound, eq_cond.data(), min_len); kd.successor(upper_bound, min_len); memcpy(lower_bound, eq_cond.data(), min_len); kd.predecessor(lower_bound, min_len); if (kd.m_is_reverse_cf) { *upper_bound_slice = rocksdb::Slice((const char *)lower_bound, min_len); *lower_bound_slice = rocksdb::Slice((const char *)upper_bound, min_len); } else { *upper_bound_slice = rocksdb::Slice((const char *)upper_bound, min_len); *lower_bound_slice = rocksdb::Slice((const char *)lower_bound, min_len); } } /* Open a cursor */ void ha_rocksdb::setup_scan_iterator(const Rdb_key_def &kd, rocksdb::Slice *const slice, const bool use_all_keys, const uint eq_cond_len) { DBUG_ASSERT(slice->size() >= eq_cond_len); Rdb_transaction *const tx = get_or_create_tx(table->in_use); bool skip_bloom = true; const rocksdb::Slice eq_cond(slice->data(), eq_cond_len); if (check_bloom_and_set_bounds( ha_thd(), kd, eq_cond, use_all_keys, eq_cond_len, m_scan_it_lower_bound, m_scan_it_upper_bound, &m_scan_it_lower_bound_slice, &m_scan_it_upper_bound_slice)) { skip_bloom = false; } /* In some cases, setup_scan_iterator() is called multiple times from the same query but bloom filter can not always be used. Suppose the following query example. id2 is VARCHAR(30) and PRIMARY KEY (id1, id2). select count(*) from t2 WHERE id1=100 and id2 IN ('00000000000000000000', '100'); In this case, setup_scan_iterator() is called twice, the first time is for (id1, id2)=(100, '00000000000000000000') and the second time is for (100, '100'). If prefix bloom filter length is 24 bytes, prefix bloom filter can be used for the first condition but not for the second condition. If bloom filter condition is changed, currently it is necessary to destroy and re-create Iterator. */ if (m_scan_it_skips_bloom != skip_bloom) { release_scan_iterator(); } /* SQL layer can call rnd_init() multiple times in a row. In that case, re-use the iterator, but re-position it at the table start. */ if (!m_scan_it) { const bool fill_cache = !THDVAR(ha_thd(), skip_fill_cache); if (commit_in_the_middle()) { DBUG_ASSERT(m_scan_it_snapshot == nullptr); m_scan_it_snapshot = rdb->GetSnapshot(); auto read_opts = rocksdb::ReadOptions(); read_opts.total_order_seek = true; // TODO: set based on WHERE conditions read_opts.snapshot = m_scan_it_snapshot; m_scan_it = rdb->NewIterator(read_opts, kd.get_cf()); } else { m_scan_it = tx->get_iterator(kd.get_cf(), skip_bloom, fill_cache, m_scan_it_lower_bound_slice, m_scan_it_upper_bound_slice); } m_scan_it_skips_bloom = skip_bloom; } } void ha_rocksdb::release_scan_iterator() { delete m_scan_it; m_scan_it = nullptr; if (m_scan_it_snapshot) { rdb->ReleaseSnapshot(m_scan_it_snapshot); m_scan_it_snapshot = nullptr; } } void ha_rocksdb::setup_iterator_for_rnd_scan() { uint key_size; int key_start_matching_bytes = m_pk_descr->get_first_key(m_pk_packed_tuple, &key_size); rocksdb::Slice table_key((const char *)m_pk_packed_tuple, key_size); setup_scan_iterator(*m_pk_descr, &table_key, false, key_start_matching_bytes); m_scan_it->Seek(table_key); m_skip_scan_it_next_call = true; } /** @return HA_EXIT_SUCCESS OK other HA_ERR error code (can be SE-specific) */ int ha_rocksdb::rnd_init(bool scan) { DBUG_ENTER_FUNC(); Rdb_transaction *const tx = get_or_create_tx(table->in_use); setup_read_decoders(); if (scan) { m_rnd_scan_is_new_snapshot = !tx->has_snapshot(); setup_iterator_for_rnd_scan(); } else { /* We don't need any preparations for rnd_pos() calls. */ } // If m_lock_rows is on then we will be doing a get_for_update when accessing // the index, so don't acquire the snapshot right away. Otherwise acquire // the snapshot immediately. tx->acquire_snapshot(m_lock_rows == RDB_LOCK_NONE); DBUG_RETURN(HA_EXIT_SUCCESS); } /** @return HA_EXIT_SUCCESS OK other HA_ERR error code (can be SE-specific) */ int ha_rocksdb::rnd_next(uchar *const buf) { DBUG_ENTER_FUNC(); int rc; for (;;) { rc = rnd_next_with_direction(buf, true); if (rc != HA_ERR_ROCKSDB_STATUS_BUSY || !m_rnd_scan_is_new_snapshot) break; /* exit the loop */ // release the snapshot and iterator and then regenerate them Rdb_transaction *tx = get_or_create_tx(table->in_use); tx->release_snapshot(); release_scan_iterator(); setup_iterator_for_rnd_scan(); } m_rnd_scan_is_new_snapshot = false; if (rc == HA_ERR_KEY_NOT_FOUND) rc = HA_ERR_END_OF_FILE; DBUG_RETURN(rc); } /* See also secondary_index_read(). */ int ha_rocksdb::rnd_next_with_direction(uchar *const buf, bool move_forward) { DBUG_ENTER_FUNC(); int rc; table->status = STATUS_NOT_FOUND; #ifdef MARIAROCKS_NOT_YET stats.rows_requested++; #endif if (!m_scan_it || !is_valid(m_scan_it)) { /* We can get here when SQL layer has called h->index_init(PRIMARY); h->index_read_map(full index tuple, HA_READ_KEY_EXACT); In this case, we should return EOF. */ DBUG_RETURN(HA_ERR_END_OF_FILE); } for (;;) { if (m_skip_scan_it_next_call) { m_skip_scan_it_next_call = false; } else { if (move_forward) m_scan_it->Next(); /* this call cannot fail */ else m_scan_it->Prev(); /* this call cannot fail */ } if (!is_valid(m_scan_it)) { rc = HA_ERR_END_OF_FILE; break; } /* check if we're out of this table */ const rocksdb::Slice key = m_scan_it->key(); if (!m_pk_descr->covers_key(key)) { rc = HA_ERR_END_OF_FILE; break; } if (m_lock_rows != RDB_LOCK_NONE) { /* Lock the row we've just read. Now we call get_for_update which will 1) Take a lock and 2) Will fail if the row was deleted since the snapshot was taken. */ Rdb_transaction *const tx = get_or_create_tx(table->in_use); DEBUG_SYNC(ha_thd(), "rocksdb_concurrent_delete"); if (m_pk_descr->has_ttl() && should_hide_ttl_rec(*m_pk_descr, m_scan_it->value(), tx->m_snapshot_timestamp)) { continue; } const rocksdb::Status s = get_for_update(tx, m_pk_descr->get_cf(), key, &m_retrieved_record); if (s.IsNotFound() && my_core::thd_tx_isolation(ha_thd()) == ISO_READ_COMMITTED) { // This occurs if we accessed a row, tried to lock it, failed, // released and reacquired the snapshot (because of READ COMMITTED // mode) and the row was deleted by someone else in the meantime. // If so, we just want to move on to the next row. continue; } if (!s.ok()) { DBUG_RETURN(tx->set_status_error(table->in_use, s, *m_pk_descr, m_tbl_def, m_table_handler)); } // If we called get_for_update() use the value from that call not from // the iterator as it may be stale since we don't have a snapshot // when m_lock_rows is not RDB_LOCK_NONE. m_last_rowkey.copy(key.data(), key.size(), &my_charset_bin); rc = convert_record_from_storage_format(&key, buf); } else { // Use the value from the iterator rocksdb::Slice value = m_scan_it->value(); if (m_pk_descr->has_ttl() && should_hide_ttl_rec( *m_pk_descr, value, get_or_create_tx(table->in_use)->m_snapshot_timestamp)) { continue; } m_last_rowkey.copy(key.data(), key.size(), &my_charset_bin); rc = convert_record_from_storage_format(&key, &value, buf); } table->status = 0; break; } if (!rc) { #ifdef MARIAROCKS_NOT_YET stats.rows_read++; stats.rows_index_next++; #endif update_row_stats(ROWS_READ); } DBUG_RETURN(rc); } int ha_rocksdb::rnd_end() { DBUG_ENTER_FUNC(); release_scan_iterator(); DBUG_RETURN(HA_EXIT_SUCCESS); } /** @return HA_EXIT_SUCCESS OK other HA_ERR error code (can be SE-specific) */ int ha_rocksdb::index_init(uint idx, bool sorted) { DBUG_ENTER_FUNC(); Rdb_transaction *const tx = get_or_create_tx(table->in_use); DBUG_ASSERT(tx != nullptr); setup_read_decoders(); m_start_range= NULL; if (!m_keyread_only) { m_key_descr_arr[idx]->get_lookup_bitmap(table, &m_lookup_bitmap); } // If m_lock_rows is not RDB_LOCK_NONE then we will be doing a get_for_update // when accessing the index, so don't acquire the snapshot right away. // Otherwise acquire the snapshot immediately. tx->acquire_snapshot(m_lock_rows == RDB_LOCK_NONE); active_index = idx; DBUG_RETURN(HA_EXIT_SUCCESS); } /** @return HA_EXIT_SUCCESS OK */ int ha_rocksdb::index_end() { DBUG_ENTER_FUNC(); release_scan_iterator(); bitmap_free(&m_lookup_bitmap); active_index = MAX_KEY; in_range_check_pushed_down = FALSE; m_start_range= NULL; DBUG_RETURN(HA_EXIT_SUCCESS); } /** @return HA_EXIT_SUCCESS OK other HA_ERR error code (can be SE-specific) */ int ha_rocksdb::truncate() { DBUG_ENTER_FUNC(); DBUG_ASSERT(m_tbl_def != nullptr); m_tbl_def->m_auto_incr_val = 1; m_tbl_def->m_hidden_pk_val = 1; DBUG_RETURN(remove_rows(m_tbl_def)); } /* Delete the row we've last read. The row is also passed as parameter. @detail The caller guarantees table buf points to the row that was just read. The row is either table->record[0] or table->record[1]. (Check out InnoDB: row_update_for_mysql() has "UT_NOT_USED(mysql_rec)" @return HA_EXIT_SUCCESS OK other HA_ERR error code (can be SE-specific) */ int ha_rocksdb::delete_row(const uchar *const buf) { DBUG_ENTER_FUNC(); DBUG_ASSERT(buf != nullptr); set_last_rowkey(buf); rocksdb::Slice key_slice(m_last_rowkey.ptr(), m_last_rowkey.length()); Rdb_transaction *const tx = get_or_create_tx(table->in_use); ulonglong bytes_written = 0; const uint index = pk_index(table, m_tbl_def); rocksdb::Status s = delete_or_singledelete(index, tx, m_pk_descr->get_cf(), key_slice); if (!s.ok()) { DBUG_RETURN(tx->set_status_error(table->in_use, s, *m_pk_descr, m_tbl_def, m_table_handler)); } else { bytes_written = key_slice.size(); } longlong hidden_pk_id = 0; if (m_tbl_def->m_key_count > 1 && has_hidden_pk(table)) { int err = read_hidden_pk_id_from_rowkey(&hidden_pk_id); if (err) { DBUG_RETURN(err); } } // Delete the record for every secondary index for (uint i = 0; i < m_tbl_def->m_key_count; i++) { if (!is_pk(i, table, m_tbl_def)) { int packed_size; const Rdb_key_def &kd = *m_key_descr_arr[i]; packed_size = kd.pack_record(table, m_pack_buffer, buf, m_sk_packed_tuple, nullptr, false, hidden_pk_id); rocksdb::Slice secondary_key_slice( reinterpret_cast(m_sk_packed_tuple), packed_size); /* Deleting on secondary key doesn't need any locks: */ tx->get_indexed_write_batch()->SingleDelete(kd.get_cf(), secondary_key_slice); bytes_written += secondary_key_slice.size(); } } tx->incr_delete_count(); if (do_bulk_commit(tx)) { DBUG_RETURN(HA_ERR_ROCKSDB_BULK_LOAD); } #ifdef MARIAROCKS_NOT_YET stats.rows_deleted++; #endif update_row_stats(ROWS_DELETED); tx->update_bytes_written(bytes_written); DBUG_RETURN(HA_EXIT_SUCCESS); } rocksdb::Status ha_rocksdb::delete_or_singledelete( uint index, Rdb_transaction *const tx, rocksdb::ColumnFamilyHandle *const column_family, const rocksdb::Slice &key) { if (can_use_single_delete(index)) return tx->single_delete(column_family, key); return tx->delete_key(column_family, key); } void ha_rocksdb::update_stats(void) { DBUG_ENTER_FUNC(); stats.records = 0; stats.index_file_length = 0ul; stats.data_file_length = 0ul; stats.mean_rec_length = 0; for (uint i = 0; i < m_tbl_def->m_key_count; i++) { if (is_pk(i, table, m_tbl_def)) { stats.data_file_length = m_pk_descr->m_stats.m_actual_disk_size; stats.records = m_pk_descr->m_stats.m_rows; } else { stats.index_file_length += m_key_descr_arr[i]->m_stats.m_actual_disk_size; } } DBUG_VOID_RETURN; } /** @return HA_EXIT_SUCCESS OK HA_EXIT_FAILURE Error */ int ha_rocksdb::info(uint flag) { DBUG_ENTER_FUNC(); if (!table) DBUG_RETURN(HA_EXIT_FAILURE); if (flag & HA_STATUS_VARIABLE) { /* Test only to simulate corrupted stats */ DBUG_EXECUTE_IF("myrocks_simulate_negative_stats", m_pk_descr->m_stats.m_actual_disk_size = -m_pk_descr->m_stats.m_actual_disk_size;); update_stats(); /* If any stats are negative due to bad cached stats, re-run analyze table and re-retrieve the stats. */ if (static_cast(stats.data_file_length) < 0 || static_cast(stats.index_file_length) < 0 || static_cast(stats.records) < 0) { if (calculate_stats_for_table()) { DBUG_RETURN(HA_EXIT_FAILURE); } update_stats(); } // if number of records is hardcoded, we do not want to force computation // of memtable cardinalities if (stats.records == 0 || (rocksdb_force_compute_memtable_stats && rocksdb_debug_optimizer_n_rows == 0)) { // First, compute SST files stats uchar buf[Rdb_key_def::INDEX_NUMBER_SIZE * 2]; auto r = get_range(pk_index(table, m_tbl_def), buf); uint64_t sz = 0; uint8_t include_flags = rocksdb::DB::INCLUDE_FILES; // recompute SST files stats only if records count is 0 if (stats.records == 0) { rdb->GetApproximateSizes(m_pk_descr->get_cf(), &r, 1, &sz, include_flags); stats.records+= sz/ROCKSDB_ASSUMED_KEY_VALUE_DISK_SIZE; stats.data_file_length+= sz; } // Second, compute memtable stats. This call is expensive, so cache // values computed for some time. uint64_t cachetime = rocksdb_force_compute_memtable_stats_cachetime; uint64_t time = (cachetime == 0) ? 0 : my_interval_timer() / 1000; if (cachetime == 0 || time > m_table_handler->m_mtcache_last_update + cachetime) { uint64_t memtableCount; uint64_t memtableSize; // the stats below are calculated from skiplist wich is a probablistic // data structure, so the results vary between test runs // it also can return 0 for quite a large tables which means that // cardinality for memtable only indxes will be reported as 0 rdb->GetApproximateMemTableStats(m_pk_descr->get_cf(), r, &memtableCount, &memtableSize); // Atomically update all of these fields at the same time if (cachetime > 0) { if (m_table_handler->m_mtcache_lock.fetch_add( 1, std::memory_order_acquire) == 0) { m_table_handler->m_mtcache_count = memtableCount; m_table_handler->m_mtcache_size = memtableSize; m_table_handler->m_mtcache_last_update = time; } m_table_handler->m_mtcache_lock.fetch_sub(1, std::memory_order_release); } stats.records += memtableCount; stats.data_file_length += memtableSize; } else { // Cached data is still valid, so use it instead stats.records += m_table_handler->m_mtcache_count; stats.data_file_length += m_table_handler->m_mtcache_size; } if (rocksdb_debug_optimizer_n_rows > 0) stats.records = rocksdb_debug_optimizer_n_rows; } if (stats.records != 0) stats.mean_rec_length = stats.data_file_length / stats.records; } if (flag & HA_STATUS_CONST) { ref_length = m_pk_descr->max_storage_fmt_length(); for (uint i = 0; i < m_tbl_def->m_key_count; i++) { if (is_hidden_pk(i, table, m_tbl_def)) { continue; } KEY *const k = &table->key_info[i]; for (uint j = 0; j < k->ext_key_parts; j++) { const Rdb_index_stats &k_stats = m_key_descr_arr[i]->m_stats; uint x; if (k_stats.m_distinct_keys_per_prefix.size() > j && k_stats.m_distinct_keys_per_prefix[j] > 0) { x = k_stats.m_rows / k_stats.m_distinct_keys_per_prefix[j]; /* If the number of rows is less than the number of prefixes (due to sampling), the average number of rows with the same prefix is 1. */ if (x == 0) { x = 1; } } else { x = 0; } if (x > stats.records) x = stats.records; if ((x == 0 && rocksdb_debug_optimizer_no_zero_cardinality) || rocksdb_debug_optimizer_n_rows > 0) { // Fake cardinality implementation. For example, (idx1, idx2, idx3) // index /* Make MariaRocks behave the same way as MyRocks does: 1. SQL layer thinks that unique secondary indexes are not extended with PK columns (both in MySQL and MariaDB) 2. MariaDB also thinks that indexes with partially-covered columns are not extended with PK columns. Use the same number of keyparts that MyRocks would use. */ uint ext_key_parts2; if (k->flags & HA_NOSAME) ext_key_parts2= k->ext_key_parts; // This is #1 else ext_key_parts2= m_key_descr_arr[i]->get_key_parts(); // This is #2. // will have rec_per_key for (idx1)=4, (idx1,2)=2, and (idx1,2,3)=1. // rec_per_key for the whole index is 1, and multiplied by 2^n if // n suffix columns of the index are not used. x = 1 << (ext_key_parts2 - j - 1); } k->rec_per_key[j] = x; } } } if (flag & HA_STATUS_ERRKEY) { /* Currently we support only primary keys so we know which key had a uniqueness violation. */ errkey = m_dupp_errkey; dup_ref = m_pk_tuple; // TODO(?): this should store packed PK. } if (flag & HA_STATUS_AUTO) { stats.auto_increment_value = m_tbl_def->m_auto_incr_val; } DBUG_RETURN(HA_EXIT_SUCCESS); } void ha_rocksdb::position(const uchar *const record) { DBUG_ENTER_FUNC(); longlong hidden_pk_id = 0; if (has_hidden_pk(table) && read_hidden_pk_id_from_rowkey(&hidden_pk_id)) DBUG_ASSERT(false); // should never reach here /* Get packed primary key value from the record. (Note: m_last_rowkey has the packed PK of last-read-row, which allows to handle most cases, but there is an exception to this: when slave applies RBR events, it fills the record and then calls position(); rnd_pos()) Also note that we *can* rely on PK column values being available. This is because table_flags() includes HA_PRIMARY_KEY_REQUIRED_FOR_POSITION bit. When that is true, table->prepare_for_position() adds PK columns into the read set (this may potentially disable index-only access when PK column cannot be restored from its mem-comparable form in the secondary indexes). */ const uint packed_size = m_pk_descr->pack_record( table, m_pack_buffer, record, ref, nullptr, false, hidden_pk_id); /* It could be that mem-comparable form of PK occupies less than ref_length bytes. Fill the remainder with zeros. */ if (ref_length > packed_size) memset(ref + packed_size, 0, ref_length - packed_size); DBUG_VOID_RETURN; } /** @return HA_EXIT_SUCCESS OK other HA_ERR error code (can be SE-specific) */ int ha_rocksdb::rnd_pos(uchar *const buf, uchar *const pos) { DBUG_ENTER_FUNC(); int rc; size_t len; #ifdef MARIAROCKS_NOT_YET stats.rows_requested++; #endif len = m_pk_descr->key_length(table, rocksdb::Slice((const char *)pos, ref_length)); if (len == size_t(-1)) { DBUG_RETURN(HA_ERR_ROCKSDB_CORRUPT_DATA); /* Data corruption? */ } rc = get_row_by_rowid(buf, pos, len); if (!rc) { #ifdef MARIAROCKS_NOT_YET stats.rows_read++; #endif update_row_stats(ROWS_READ); } DBUG_RETURN(rc); } /* @brief Calculate (if needed) the bitmap of indexes that are modified by the current query @detail The calculation done by checking index definitions against the table->write_set column bitmap. */ void ha_rocksdb::calc_updated_indexes() { if (!m_update_scope_is_valid) { m_update_scope_is_valid = true; m_update_scope.clear_all(); for (uint keynr = 0; keynr < table->s->keys; keynr++) { const Rdb_key_def &kd = *m_key_descr_arr[keynr]; /* Walk over all key parts, including the "extended key" suffix */ const uint key_parts = kd.get_key_parts(); for (uint kp = 0; kp < key_parts; kp++) { if (has_hidden_pk(table) && kp + 1 == key_parts) break; Field *const field = kd.get_table_field_for_part_no(table, kp); if (bitmap_is_set(table->write_set, field->field_index)) { m_update_scope.set_bit(keynr); break; } } } } } /** @return HA_EXIT_SUCCESS OK other HA_ERR error code (can be SE-specific) */ int ha_rocksdb::update_row(const uchar *const old_data, const uchar *const new_data) { DBUG_ENTER_FUNC(); DBUG_ASSERT(old_data != nullptr); DBUG_ASSERT(new_data != nullptr); DBUG_ASSERT(m_lock_rows == RDB_LOCK_WRITE); /* old_data points to record we're updating. It is the same as the record we've just read (for multi-table UPDATE, too, because SQL layer will make an rnd_pos() call to re-read the record before calling update_row()) */ DBUG_ASSERT(new_data == table->record[0]); const int rv = update_write_row(old_data, new_data, false); if (rv == 0) { #ifdef MARIAROCKS_NOT_YET stats.rows_updated++; #endif update_row_stats(ROWS_UPDATED); } DBUG_RETURN(rv); } /* MariaDB's temporary: MyRocks has this function in sql/handler.cc: */ bool can_hold_read_locks_on_select(THD *thd, thr_lock_type lock_type) { return (lock_type == TL_READ_WITH_SHARED_LOCKS || lock_type == TL_READ_NO_INSERT || (lock_type != TL_IGNORE && thd->lex->sql_command != SQLCOM_SELECT)); } /* The following function was copied from ha_blackhole::store_lock: */ THR_LOCK_DATA **ha_rocksdb::store_lock(THD *const thd, THR_LOCK_DATA **to, enum thr_lock_type lock_type) { DBUG_ENTER_FUNC(); DBUG_ASSERT(thd != nullptr); DBUG_ASSERT(to != nullptr); bool in_lock_tables = my_core::thd_in_lock_tables(thd); /* First, make a decision about MyRocks's internal locking */ if (lock_type >= TL_WRITE_ALLOW_WRITE) { m_lock_rows = RDB_LOCK_WRITE; } else if (lock_type == TL_READ_WITH_SHARED_LOCKS) { m_lock_rows = RDB_LOCK_READ; } else if (lock_type != TL_IGNORE) { m_lock_rows = RDB_LOCK_NONE; if (THDVAR(thd, lock_scanned_rows)) { /* The following logic was copied directly from ha_innobase::store_lock_with_x_type() in storage/innobase/handler/ha_innodb.cc and causes MyRocks to leave locks in place on rows that are in a table that is not being updated. */ const uint sql_command = my_core::thd_sql_command(thd); if ((lock_type == TL_READ && in_lock_tables) || (lock_type == TL_READ_HIGH_PRIORITY && in_lock_tables) || can_hold_read_locks_on_select(thd, lock_type)) { ulong tx_isolation = my_core::thd_tx_isolation(thd); if (sql_command != SQLCOM_CHECKSUM && ((my_core::thd_test_options(thd, OPTION_BIN_LOG) && tx_isolation > ISO_READ_COMMITTED) || tx_isolation == ISO_SERIALIZABLE || (lock_type != TL_READ && lock_type != TL_READ_NO_INSERT) || (sql_command != SQLCOM_INSERT_SELECT && sql_command != SQLCOM_REPLACE_SELECT && sql_command != SQLCOM_UPDATE && sql_command != SQLCOM_DELETE && sql_command != SQLCOM_CREATE_TABLE))) { m_lock_rows = RDB_LOCK_READ; } } } } /* Then, tell the SQL layer what kind of locking it should use: */ if (lock_type != TL_IGNORE && m_db_lock.type == TL_UNLOCK) { /* Here is where we get into the guts of a row level lock. If TL_UNLOCK is set If we are not doing a LOCK TABLE or DISCARD/IMPORT TABLESPACE, then allow multiple writers */ if ((lock_type >= TL_WRITE_CONCURRENT_INSERT && lock_type <= TL_WRITE) && !in_lock_tables && !my_core::thd_tablespace_op(thd)) { lock_type = TL_WRITE_ALLOW_WRITE; } /* In queries of type INSERT INTO t1 SELECT ... FROM t2 ... MySQL would use the lock TL_READ_NO_INSERT on t2, and that would conflict with TL_WRITE_ALLOW_WRITE, blocking all inserts to t2. Convert the lock to a normal read lock to allow concurrent inserts to t2. */ if (lock_type == TL_READ_NO_INSERT && !in_lock_tables) { lock_type = TL_READ; } m_db_lock.type = lock_type; } *to++ = &m_db_lock; DBUG_RETURN(to); } void ha_rocksdb::read_thd_vars(THD *const thd) { m_store_row_debug_checksums = THDVAR(thd, store_row_debug_checksums); m_verify_row_debug_checksums = THDVAR(thd, verify_row_debug_checksums); m_checksums_pct = THDVAR(thd, checksums_pct); } /** @return HA_EXIT_SUCCESS OK other HA_ERR error code (cannot be SE-specific) */ int ha_rocksdb::external_lock(THD *const thd, int lock_type) { DBUG_ENTER_FUNC(); DBUG_ASSERT(thd != nullptr); int res = HA_EXIT_SUCCESS; int binlog_format = my_core::thd_binlog_format(thd); bool unsafe_for_binlog = THDVAR(ha_thd(), unsafe_for_binlog); /* A note about (*) below: In fb/mysql, LOCK TABLE myrocks_table WRITE - produces an error when @@binlog_format='ROW'. - does not produce an error with @@binlog_format='STATEMENT' We cannot easily provide this behavior in MariaDB, because LOCK ROWS changes @@binlog_format for the duration of the statement. Instead we will make the check in ha_rocksdb::start_stmt and will produce an error in the first DML statement done under LOCK TABLE (**) - The same happens for other "DDL-type" commands, so allow them here as well. */ if (lock_type == F_WRLCK && !thd->rgi_slave && !unsafe_for_binlog && binlog_format != BINLOG_FORMAT_ROW && binlog_format != BINLOG_FORMAT_UNSPEC && thd->lex->sql_command != SQLCOM_LOCK_TABLES && // (*) thd->lex->sql_command != SQLCOM_ANALYZE && // (**) thd->lex->sql_command != SQLCOM_OPTIMIZE && // (**) thd->lex->sql_command != SQLCOM_FLUSH && // (**) my_core::thd_binlog_filter_ok(thd)) { my_error(ER_REQUIRE_ROW_BINLOG_FORMAT, MYF(0)); DBUG_RETURN(HA_ERR_UNSUPPORTED); } if (lock_type == F_UNLCK) { Rdb_transaction *const tx = get_tx_from_thd(thd); if (tx) { tx->io_perf_end_and_record(&m_io_perf); tx->m_n_mysql_tables_in_use--; if (tx->m_n_mysql_tables_in_use == 0 && !my_core::thd_test_options(thd, OPTION_NOT_AUTOCOMMIT | OPTION_BEGIN)) { /* Do like InnoDB: when we get here, it's time to commit a single-statement transaction. If the statement involved multiple tables, this code will be executed for each of them, but that's ok because non-first tx->commit() calls will be no-ops. */ if (tx->commit_or_rollback()) { res = HA_ERR_INTERNAL_ERROR; } } } } else { if (my_core::thd_tx_isolation(thd) < ISO_READ_COMMITTED || my_core::thd_tx_isolation(thd) > ISO_REPEATABLE_READ) { my_error(ER_ISOLATION_MODE_NOT_SUPPORTED, MYF(0), tx_isolation_names[my_core::thd_tx_isolation(thd)]); DBUG_RETURN(HA_ERR_UNSUPPORTED); } /* It's nice to do the following on start of every statement. The problem is, handler->start_stmt() is not called for INSERTs. So, we put this code here. */ Rdb_transaction *const tx = get_or_create_tx(thd); read_thd_vars(thd); m_update_scope_is_valid = false; if (skip_unique_check()) { if ((thd->lex->sql_command == SQLCOM_INSERT || thd->lex->sql_command == SQLCOM_LOAD || thd->lex->sql_command == SQLCOM_REPLACE) && (thd->lex->duplicates == DUP_REPLACE || thd->lex->duplicates == DUP_UPDATE)) { my_error(ER_ON_DUPLICATE_DISABLED, MYF(0), thd->query()); DBUG_RETURN(HA_ERR_UNSUPPORTED); } } if (lock_type == F_WRLCK) { if (tx->is_tx_read_only()) { my_error(ER_UPDATES_WITH_CONSISTENT_SNAPSHOT, MYF(0)); DBUG_RETURN(HA_ERR_UNSUPPORTED); } #ifdef MARIADB_NOT_YET if (thd->get_explicit_snapshot()) { my_error(ER_UPDATES_WITH_EXPLICIT_SNAPSHOT, MYF(0)); DBUG_RETURN(HA_ERR_UNSUPPORTED); } #endif /* SQL layer signals us to take a write lock. It does so when starting DML statement. We should put locks on the rows we're reading. Note: sometimes, external_lock() can be called without a prior ::store_lock call. That's why we need to set lock_* members here, too. */ m_lock_rows = RDB_LOCK_WRITE; if (thd->lex->sql_command == SQLCOM_CREATE_INDEX || thd->lex->sql_command == SQLCOM_DROP_INDEX || thd->lex->sql_command == SQLCOM_ALTER_TABLE) { tx->m_ddl_transaction = true; } } tx->m_n_mysql_tables_in_use++; rocksdb_register_tx(rocksdb_hton, thd, tx); tx->io_perf_start(&m_io_perf); } DBUG_RETURN(res); } /** @note A quote from ha_innobase::start_stmt(): MySQL calls this function at the start of each SQL statement inside LOCK TABLES. Inside LOCK TABLES the ::external_lock method does not work to mark SQL statement borders. @return HA_EXIT_SUCCESS OK */ int ha_rocksdb::start_stmt(THD *const thd, thr_lock_type lock_type) { DBUG_ENTER_FUNC(); /* MariaDB: the following is a copy of the check in ha_rocksdb::external_lock: */ int binlog_format = my_core::thd_binlog_format(thd); bool unsafe_for_binlog = THDVAR(ha_thd(), unsafe_for_binlog); if (lock_type >= TL_WRITE_ALLOW_WRITE && !thd->rgi_slave && !unsafe_for_binlog && binlog_format != BINLOG_FORMAT_ROW && binlog_format != BINLOG_FORMAT_UNSPEC && my_core::thd_binlog_filter_ok(thd)) { my_error(ER_REQUIRE_ROW_BINLOG_FORMAT, MYF(0)); DBUG_RETURN(HA_ERR_UNSUPPORTED); } DBUG_ASSERT(thd != nullptr); Rdb_transaction *const tx = get_or_create_tx(thd); read_thd_vars(thd); rocksdb_register_tx(ht, thd, tx); tx->io_perf_start(&m_io_perf); DBUG_RETURN(HA_EXIT_SUCCESS); } rocksdb::Range get_range(uint32_t i, uchar buf[Rdb_key_def::INDEX_NUMBER_SIZE * 2], int offset1, int offset2) { uchar *buf_begin = buf; uchar *buf_end = buf + Rdb_key_def::INDEX_NUMBER_SIZE; rdb_netbuf_store_index(buf_begin, i + offset1); rdb_netbuf_store_index(buf_end, i + offset2); return rocksdb::Range( rocksdb::Slice((const char *)buf_begin, Rdb_key_def::INDEX_NUMBER_SIZE), rocksdb::Slice((const char *)buf_end, Rdb_key_def::INDEX_NUMBER_SIZE)); } static rocksdb::Range get_range(const Rdb_key_def &kd, uchar buf[Rdb_key_def::INDEX_NUMBER_SIZE * 2], int offset1, int offset2) { return get_range(kd.get_index_number(), buf, offset1, offset2); } rocksdb::Range get_range(const Rdb_key_def &kd, uchar buf[Rdb_key_def::INDEX_NUMBER_SIZE * 2]) { if (kd.m_is_reverse_cf) { return myrocks::get_range(kd, buf, 1, 0); } else { return myrocks::get_range(kd, buf, 0, 1); } } rocksdb::Range ha_rocksdb::get_range(const int &i, uchar buf[Rdb_key_def::INDEX_NUMBER_SIZE * 2]) const { return myrocks::get_range(*m_key_descr_arr[i], buf); } /* This function is called with total_order_seek=true, but upper/lower bound setting is not necessary. Boundary set is useful when there is no matching key, but in drop_index_thread's case, it means index is marked as removed, so no further seek will happen for the index id. */ static bool is_myrocks_index_empty( rocksdb::ColumnFamilyHandle *cfh, const bool is_reverse_cf, const rocksdb::ReadOptions &read_opts, const uint index_id) { bool index_removed = false; uchar key_buf[Rdb_key_def::INDEX_NUMBER_SIZE] = {0}; rdb_netbuf_store_uint32(key_buf, index_id); const rocksdb::Slice key = rocksdb::Slice(reinterpret_cast(key_buf), sizeof(key_buf)); std::unique_ptr it(rdb->NewIterator(read_opts, cfh)); rocksdb_smart_seek(is_reverse_cf, it.get(), key); if (!it->Valid()) { index_removed = true; } else { if (memcmp(it->key().data(), key_buf, Rdb_key_def::INDEX_NUMBER_SIZE)) { // Key does not have same prefix index_removed = true; } } return index_removed; } /* Drop index thread's main logic */ void Rdb_drop_index_thread::run() { RDB_MUTEX_LOCK_CHECK(m_signal_mutex); for (;;) { // The stop flag might be set by shutdown command // after drop_index_thread releases signal_mutex // (i.e. while executing expensive Seek()). To prevent drop_index_thread // from entering long cond_timedwait, checking if stop flag // is true or not is needed, with drop_index_interrupt_mutex held. if (m_stop) { break; } timespec ts; int sec= dict_manager.is_drop_index_empty() ? 24 * 60 * 60 // no filtering : 60; // filtering set_timespec(ts,sec); const auto ret MY_ATTRIBUTE((__unused__)) = mysql_cond_timedwait(&m_signal_cond, &m_signal_mutex, &ts); if (m_stop) { break; } // make sure, no program error is returned DBUG_ASSERT(ret == 0 || ret == ETIMEDOUT); RDB_MUTEX_UNLOCK_CHECK(m_signal_mutex); std::unordered_set indices; dict_manager.get_ongoing_drop_indexes(&indices); if (!indices.empty()) { std::unordered_set finished; rocksdb::ReadOptions read_opts; read_opts.total_order_seek = true; // disable bloom filter for (const auto d : indices) { uint32 cf_flags = 0; if (!dict_manager.get_cf_flags(d.cf_id, &cf_flags)) { sql_print_error("RocksDB: Failed to get column family flags " "from cf id %u. MyRocks data dictionary may " "get corrupted.", d.cf_id); abort(); } rocksdb::ColumnFamilyHandle *cfh = cf_manager.get_cf(d.cf_id); DBUG_ASSERT(cfh); const bool is_reverse_cf = cf_flags & Rdb_key_def::REVERSE_CF_FLAG; if (is_myrocks_index_empty(cfh, is_reverse_cf, read_opts, d.index_id)) { finished.insert(d); continue; } uchar buf[Rdb_key_def::INDEX_NUMBER_SIZE * 2]; rocksdb::Range range = get_range(d.index_id, buf, is_reverse_cf ? 1 : 0, is_reverse_cf ? 0 : 1); rocksdb::Status status = DeleteFilesInRange(rdb->GetBaseDB(), cfh, &range.start, &range.limit); if (!status.ok()) { if (status.IsShutdownInProgress()) { break; } rdb_handle_io_error(status, RDB_IO_ERROR_BG_THREAD); } status = rdb->CompactRange(getCompactRangeOptions(), cfh, &range.start, &range.limit); if (!status.ok()) { if (status.IsShutdownInProgress()) { break; } rdb_handle_io_error(status, RDB_IO_ERROR_BG_THREAD); } if (is_myrocks_index_empty(cfh, is_reverse_cf, read_opts, d.index_id)) { finished.insert(d); } } if (!finished.empty()) { dict_manager.finish_drop_indexes(finished); } } RDB_MUTEX_LOCK_CHECK(m_signal_mutex); } RDB_MUTEX_UNLOCK_CHECK(m_signal_mutex); } Rdb_tbl_def *ha_rocksdb::get_table_if_exists(const char *const tablename) { std::string str; if (rdb_normalize_tablename(tablename, &str) != HA_EXIT_SUCCESS) { // We were not passed table name? DBUG_ASSERT(0); return nullptr; } return ddl_manager.find(str); } /* Note: the following function is called when the table is not open. That is, this->table==nullptr, pk_key_descr==nullptr, etc. tablename points to line in form "./dbname/tablename". @return HA_EXIT_SUCCESS OK other HA_ERR error code (can be SE-specific) */ int ha_rocksdb::delete_table(const char *const tablename) { DBUG_ENTER_FUNC(); DBUG_ASSERT(tablename != nullptr); const std::unique_ptr wb = dict_manager.begin(); rocksdb::WriteBatch *const batch = wb.get(); /* Find the table in the hash */ Rdb_tbl_def *const tbl = get_table_if_exists(tablename); if (!tbl) DBUG_RETURN(HA_ERR_NO_SUCH_TABLE); dict_manager.add_drop_table(tbl->m_key_descr_arr, tbl->m_key_count, batch); /* Remove the table entry in data dictionary (this will also remove it from the persistent data dictionary). */ ddl_manager.remove(tbl, batch, true); int err = dict_manager.commit(batch); if (err) { DBUG_RETURN(err); } rdb_drop_idx_thread.signal(); DBUG_RETURN(HA_EXIT_SUCCESS); } int ha_rocksdb::remove_rows(Rdb_tbl_def *const tbl) { const rocksdb::WriteOptions wo = rdb_get_rocksdb_write_options(handler::ha_thd()); rocksdb::ReadOptions opts; opts.total_order_seek = true; Rdb_transaction *const tx = get_or_create_tx(table->in_use); char key_buf[MAX_KEY_LENGTH]; uint key_len; ulonglong bytes_written = 0; uchar lower_bound_buf[Rdb_key_def::INDEX_NUMBER_SIZE]; uchar upper_bound_buf[Rdb_key_def::INDEX_NUMBER_SIZE]; rocksdb::Slice lower_bound_slice; rocksdb::Slice upper_bound_slice; /* Remove all records in each index. (This is is not crash-safe, but it doesn't matter, because bulk row deletion will be handled on rocksdb side) */ for (uint i = 0; i < tbl->m_key_count; i++) { const Rdb_key_def &kd = *tbl->m_key_descr_arr[i]; kd.get_infimum_key(reinterpret_cast(key_buf), &key_len); rocksdb::ColumnFamilyHandle *cf = kd.get_cf(); const rocksdb::Slice table_key(key_buf, key_len); setup_iterator_bounds(kd, table_key, Rdb_key_def::INDEX_NUMBER_SIZE, lower_bound_buf, upper_bound_buf, &lower_bound_slice, &upper_bound_slice); DBUG_ASSERT(key_len == Rdb_key_def::INDEX_NUMBER_SIZE); opts.iterate_lower_bound = &lower_bound_slice; opts.iterate_upper_bound = &upper_bound_slice; std::unique_ptr it(rdb->NewIterator(opts, cf)); it->Seek(table_key); while (it->Valid()) { const rocksdb::Slice key = it->key(); if (!kd.covers_key(key)) { break; } rocksdb::Status s; if (can_use_single_delete(i)) { s = rdb->SingleDelete(wo, cf, key); } else { s = rdb->Delete(wo, cf, key); } if (!s.ok()) { return tx->set_status_error(table->in_use, s, *m_pk_descr, m_tbl_def, m_table_handler); } bytes_written += key.size(); it->Next(); } } tx->update_bytes_written(bytes_written); return HA_EXIT_SUCCESS; } /** @return HA_EXIT_SUCCESS OK other HA_ERR error code (cannot be SE-specific) */ int ha_rocksdb::rename_table(const char *const from, const char *const to) { DBUG_ENTER_FUNC(); std::string from_str; std::string to_str; std::string from_db; std::string to_db; int rc = rdb_normalize_tablename(from, &from_str); if (rc != HA_EXIT_SUCCESS) { DBUG_RETURN(rc); } rc = rdb_split_normalized_tablename(from_str, &from_db); if (rc != HA_EXIT_SUCCESS) { DBUG_RETURN(rc); } rc = rdb_normalize_tablename(to, &to_str); if (rc != HA_EXIT_SUCCESS) { DBUG_RETURN(rc); } rc = rdb_split_normalized_tablename(to_str, &to_db); if (rc != HA_EXIT_SUCCESS) { DBUG_RETURN(rc); } // If the user changed the database part of the name then validate that the // 'to' database exists. if (from_db != to_db && !rdb_database_exists(to_db)) { // If we return a RocksDB specific error code here we get // "error: 206 - Unknown error 206". InnoDB gets // "error -1 - Unknown error -1" so let's match them. DBUG_RETURN(-1); } DBUG_EXECUTE_IF("gen_sql_table_name", to_str = to_str + "#sql-test";); const std::unique_ptr wb = dict_manager.begin(); rocksdb::WriteBatch *const batch = wb.get(); dict_manager.lock(); if (ddl_manager.rename(from_str, to_str, batch)) { rc = HA_ERR_NO_SUCH_TABLE; } else { rc = dict_manager.commit(batch); } dict_manager.unlock(); DBUG_RETURN(rc); } /** check_if_incompatible_data() called if ALTER TABLE can't detect otherwise if new and old definition are compatible @details If there are no other explicit signs like changed number of fields this function will be called by compare_tables() (sql/sql_tables.cc) to decide should we rewrite whole table or only .frm file. */ bool ha_rocksdb::check_if_incompatible_data(HA_CREATE_INFO *const info, uint table_changes) { DBUG_ENTER_FUNC(); DBUG_ASSERT(info != nullptr); // this function is needed only for online alter-table DBUG_RETURN(COMPATIBLE_DATA_NO); } /** @return HA_EXIT_SUCCESS OK */ int ha_rocksdb::extra(enum ha_extra_function operation) { DBUG_ENTER_FUNC(); switch (operation) { case HA_EXTRA_KEYREAD: m_keyread_only = true; break; case HA_EXTRA_NO_KEYREAD: m_keyread_only = false; break; case HA_EXTRA_FLUSH: /* If the table has blobs, then they are part of m_retrieved_record. This call invalidates them. */ m_retrieved_record.Reset(); break; default: break; } DBUG_RETURN(HA_EXIT_SUCCESS); } /* Given a starting key and an ending key, estimate the number of rows that will exist between the two keys. */ ha_rows ha_rocksdb::records_in_range(uint inx, key_range *const min_key, key_range *const max_key) { DBUG_ENTER_FUNC(); ha_rows ret = THDVAR(ha_thd(), records_in_range); if (ret) { DBUG_RETURN(ret); } if (table->force_index) { const ha_rows force_rows = THDVAR(ha_thd(), force_index_records_in_range); if (force_rows) { DBUG_RETURN(force_rows); } } const Rdb_key_def &kd = *m_key_descr_arr[inx]; uint size1 = 0; if (min_key) { size1 = kd.pack_index_tuple(table, m_pack_buffer, m_sk_packed_tuple, min_key->key, min_key->keypart_map); if (min_key->flag == HA_READ_PREFIX_LAST_OR_PREV || min_key->flag == HA_READ_PREFIX_LAST || min_key->flag == HA_READ_AFTER_KEY) { kd.successor(m_sk_packed_tuple, size1); } } else { kd.get_infimum_key(m_sk_packed_tuple, &size1); } uint size2 = 0; if (max_key) { size2 = kd.pack_index_tuple(table, m_pack_buffer, m_sk_packed_tuple_old, max_key->key, max_key->keypart_map); if (max_key->flag == HA_READ_PREFIX_LAST_OR_PREV || max_key->flag == HA_READ_PREFIX_LAST || max_key->flag == HA_READ_AFTER_KEY) { kd.successor(m_sk_packed_tuple_old, size2); } // pad the upper key with FFFFs to make sure it is more than the lower if (size1 > size2) { memset(m_sk_packed_tuple_old + size2, 0xff, size1 - size2); size2 = size1; } } else { kd.get_supremum_key(m_sk_packed_tuple_old, &size2); } const rocksdb::Slice slice1((const char *)m_sk_packed_tuple, size1); const rocksdb::Slice slice2((const char *)m_sk_packed_tuple_old, size2); // slice1 >= slice2 means no row will match if (slice1.compare(slice2) >= 0) { DBUG_RETURN(HA_EXIT_SUCCESS); } rocksdb::Range r(kd.m_is_reverse_cf ? slice2 : slice1, kd.m_is_reverse_cf ? slice1 : slice2); uint64_t sz = 0; auto disk_size = kd.m_stats.m_actual_disk_size; if (disk_size == 0) disk_size = kd.m_stats.m_data_size; auto rows = kd.m_stats.m_rows; if (rows == 0 || disk_size == 0) { rows = 1; disk_size = ROCKSDB_ASSUMED_KEY_VALUE_DISK_SIZE; } // Getting statistics, including from Memtables uint8_t include_flags = rocksdb::DB::INCLUDE_FILES; rdb->GetApproximateSizes(kd.get_cf(), &r, 1, &sz, include_flags); ret = rows * sz / disk_size; uint64_t memTableCount; rdb->GetApproximateMemTableStats(kd.get_cf(), r, &memTableCount, &sz); ret += memTableCount; /* GetApproximateSizes() gives estimates so ret might exceed stats.records. MySQL then decides to use full index scan rather than range scan, which is not efficient for most cases. To prevent this, changing estimated records slightly smaller than stats.records. */ if (ret >= stats.records) { ret = stats.records * 0.99; } if (rocksdb_debug_optimizer_n_rows > 0) { ret = rocksdb_debug_optimizer_n_rows; } else if (ret == 0) { ret = 1; } DBUG_RETURN(ret); } void ha_rocksdb::update_create_info(HA_CREATE_INFO *const create_info) { DBUG_ENTER_FUNC(); DBUG_ASSERT(create_info != nullptr); if (!(create_info->used_fields & HA_CREATE_USED_AUTO)) { create_info->auto_increment_value = m_tbl_def->m_auto_incr_val; } DBUG_VOID_RETURN; } /** @brief Doing manual compaction on OPTIMIZE TABLE in RocksDB. Compaction itself is executed by background thread in RocksDB, but CompactRange() waits until compaction completes so this function may take a long time. Since RocksDB dataset is allocated per index id, OPTIMIZE TABLE triggers manual compaction for all indexes of the table. @details Compaction range is from the beginning of the index id to the first row of the next index id. When using reverse order column family, the first row of the next index id should be the last row of the previous index id. @return HA_ADMIN_OK OK other HA_ADMIN error code */ int ha_rocksdb::optimize(THD *const thd, HA_CHECK_OPT *const check_opt) { DBUG_ENTER_FUNC(); DBUG_ASSERT(thd != nullptr); DBUG_ASSERT(check_opt != nullptr); for (uint i = 0; i < table->s->keys; i++) { uchar buf[Rdb_key_def::INDEX_NUMBER_SIZE * 2]; auto range = get_range(i, buf); const rocksdb::Status s = rdb->CompactRange(getCompactRangeOptions(), m_key_descr_arr[i]->get_cf(), &range.start, &range.limit); if (!s.ok()) { DBUG_RETURN(rdb_error_to_mysql(s)); } } DBUG_RETURN(HA_EXIT_SUCCESS); } static int calculate_stats( const std::unordered_map> &to_recalc, bool include_memtables) { DBUG_ENTER_FUNC(); // find per column family key ranges which need to be queried std::unordered_map> ranges; std::unordered_map stats; std::vector buf(to_recalc.size() * 2 * Rdb_key_def::INDEX_NUMBER_SIZE); uchar *bufp = buf.data(); for (const auto &it : to_recalc) { const GL_INDEX_ID index_id = it.first; auto &kd = it.second; ranges[kd->get_cf()].push_back(myrocks::get_range(*kd, bufp)); bufp += 2 * Rdb_key_def::INDEX_NUMBER_SIZE; stats[index_id] = Rdb_index_stats(index_id); DBUG_ASSERT(kd->get_key_parts() > 0); stats[index_id].m_distinct_keys_per_prefix.resize(kd->get_key_parts()); } // get RocksDB table properties for these ranges rocksdb::TablePropertiesCollection props; for (auto it : ranges) { const auto old_size MY_ATTRIBUTE((__unused__)) = props.size(); const auto status = rdb->GetPropertiesOfTablesInRange( it.first, &it.second[0], it.second.size(), &props); DBUG_ASSERT(props.size() >= old_size); if (!status.ok()) { DBUG_RETURN(ha_rocksdb::rdb_error_to_mysql( status, "Could not access RocksDB properties")); } } int num_sst = 0; for (const auto &it : props) { std::vector sst_stats; Rdb_tbl_prop_coll::read_stats_from_tbl_props(it.second, &sst_stats); /* sst_stats is a list of index statistics for indexes that have entries in the current SST file. */ for (const auto &it1 : sst_stats) { /* Only update statistics for indexes that belong to this SQL table. The reason is: We are walking through all SST files that have entries from this table (and so can compute good statistics). For other SQL tables, it can be that we're only seeing a small fraction of table's entries (and so we can't update statistics based on that). */ if (stats.find(it1.m_gl_index_id) == stats.end()) { continue; } auto it_index = to_recalc.find(it1.m_gl_index_id); DBUG_ASSERT(it_index != to_recalc.end()); if (it_index == to_recalc.end()) { continue; } stats[it1.m_gl_index_id].merge( it1, true, it_index->second->max_storage_fmt_length()); } num_sst++; } if (include_memtables) { // calculate memtable cardinality Rdb_tbl_card_coll cardinality_collector(rocksdb_table_stats_sampling_pct); auto read_opts = rocksdb::ReadOptions(); read_opts.read_tier = rocksdb::ReadTier::kMemtableTier; for (const auto &it_kd : to_recalc) { const std::shared_ptr &kd = it_kd.second; Rdb_index_stats &stat = stats[kd->get_gl_index_id()]; uchar r_buf[Rdb_key_def::INDEX_NUMBER_SIZE * 2]; auto r = myrocks::get_range(*kd, r_buf); uint64_t memtableCount; uint64_t memtableSize; rdb->GetApproximateMemTableStats(kd->get_cf(), r, &memtableCount, &memtableSize); if (memtableCount < (uint64_t)stat.m_rows / 10) { // skip tables that already have enough stats from SST files to reduce // overhead and avoid degradation of big tables stats by sampling from // relatively tiny (less than 10% of full data set) memtable dataset continue; } std::unique_ptr it = std::unique_ptr( rdb->NewIterator(read_opts, kd->get_cf())); rocksdb::Slice first_index_key((const char *)r_buf, Rdb_key_def::INDEX_NUMBER_SIZE); cardinality_collector.Reset(); for (it->Seek(first_index_key); is_valid(it.get()); it->Next()) { const rocksdb::Slice key = it->key(); if (!kd->covers_key(key)) { break; // end of this index } stat.m_rows++; cardinality_collector.ProcessKey(key, kd.get(), &stat); } cardinality_collector.AdjustStats(&stat); } } // set and persist new stats ddl_manager.set_stats(stats); ddl_manager.persist_stats(true); DBUG_RETURN(HA_EXIT_SUCCESS); } int ha_rocksdb::calculate_stats_for_table() { DBUG_ENTER_FUNC(); std::unordered_map> ids_to_check; for (uint i = 0; i < table->s->keys; i++) { ids_to_check.insert(std::make_pair(m_key_descr_arr[i]->get_gl_index_id(), m_key_descr_arr[i])); } DBUG_RETURN(calculate_stats(ids_to_check, true)); } /* @return HA_ADMIN_OK OK other HA_ADMIN error code */ int ha_rocksdb::analyze(THD *const thd, HA_CHECK_OPT *const check_opt) { DBUG_ENTER_FUNC(); if (table) { if (calculate_stats_for_table() != HA_EXIT_SUCCESS) { DBUG_RETURN(HA_ADMIN_FAILED); } } // A call to ::info is needed to repopulate some SQL level structs. This is // necessary for online analyze because we cannot rely on another ::open // call to call info for us. if (info(HA_STATUS_CONST | HA_STATUS_VARIABLE) != HA_EXIT_SUCCESS) { DBUG_RETURN(HA_ADMIN_FAILED); } DBUG_RETURN(HA_ADMIN_OK); } void ha_rocksdb::get_auto_increment(ulonglong off, ulonglong inc, ulonglong nb_desired_values, ulonglong *const first_value, ulonglong *const nb_reserved_values) { /* MySQL has a somewhat complicated way of handling the auto-increment value. The first time get_auto_increment is called for a statement, nb_desired_values is the estimate for how many values will be needed. The engine can then reserve some values, and those will be automatically used by MySQL, until a hard-coded value shows up in the insert statement, after which MySQL again calls this function to reset its starting value. * For simplicity we will just ignore nb_desired_values - we aren't going to reserve any extra values for a multi-insert statement. Each row will simply acquire the next value as needed and we will always tell MySQL that we only reserved 1 value. Since we are using an atomic value for m_auto_incr_val this should be safe - if we had to grab a mutex, doing an actual reserve of some values might be a better solution. */ DEBUG_SYNC(ha_thd(), "rocksdb.autoinc_vars"); if (off > inc) { off = 1; } Field *field; ulonglong new_val, max_val; field = table->key_info[table->s->next_number_index].key_part[0].field; max_val = rdb_get_int_col_max_value(field); // Local variable reference to simplify code below auto &auto_incr = m_tbl_def->m_auto_incr_val; if (inc == 1) { DBUG_ASSERT(off == 1); // Optimization for the standard case where we are always simply // incrementing from the last position // Use CAS operation in a loop to make sure automically get the next auto // increment value while ensuring that we don't wrap around to a negative // number. // // We set auto_incr to the min of max_val and new_val + 1. This means that // if we're at the maximum, we should be returning the same value for // multiple rows, resulting in duplicate key errors (as expected). // // If we return values greater than the max, the SQL layer will "truncate" // the value anyway, but it means that we store invalid values into // auto_incr that will be visible in SHOW CREATE TABLE. new_val = auto_incr; while (new_val != std::numeric_limits::max()) { if (auto_incr.compare_exchange_weak(new_val, std::min(new_val + 1, max_val))) { break; } } } else { // The next value can be more complicated if either 'inc' or 'off' is not 1 ulonglong last_val = auto_incr; // Loop until we can correctly update the atomic value do { DBUG_ASSERT(last_val > 0); // Calculate the next value in the auto increment series: offset // + N * increment where N is 0, 1, 2, ... // // For further information please visit: // http://dev.mysql.com/doc/refman/5.7/en/replication-options-master.html // // The following is confusing so here is an explanation: // To get the next number in the sequence above you subtract out the // offset, calculate the next sequence (N * increment) and then add the // offset back in. // // The additions are rearranged to avoid overflow. The following is // equivalent to (last_val - 1 + inc - off) / inc. This uses the fact // that (a+b)/c = a/c + b/c + (a%c + b%c)/c. To show why: // // (a+b)/c // = (a - a%c + a%c + b - b%c + b%c) / c // = (a - a%c) / c + (b - b%c) / c + (a%c + b%c) / c // = a/c + b/c + (a%c + b%c) / c // // Now, substitute a = last_val - 1, b = inc - off, c = inc to get the // following statement. ulonglong n = (last_val - 1) / inc + ((last_val - 1) % inc + inc - off) / inc; // Check if n * inc + off will overflow. This can only happen if we have // an UNSIGNED BIGINT field. if (n > (std::numeric_limits::max() - off) / inc) { DBUG_ASSERT(max_val == std::numeric_limits::max()); // The 'last_val' value is already equal to or larger than the largest // value in the sequence. Continuing would wrap around (technically // the behavior would be undefined). What should we do? // We could: // 1) set the new value to the last possible number in our sequence // as described above. The problem with this is that this // number could be smaller than a value in an existing row. // 2) set the new value to the largest possible number. This number // may not be in our sequence, but it is guaranteed to be equal // to or larger than any other value already inserted. // // For now I'm going to take option 2. // // Returning ULLONG_MAX from get_auto_increment will cause the SQL // layer to fail with ER_AUTOINC_READ_FAILED. This means that due to // the SE API for get_auto_increment, inserts will fail with // ER_AUTOINC_READ_FAILED if the column is UNSIGNED BIGINT, but // inserts will fail with ER_DUP_ENTRY for other types (or no failure // if the column is in a non-unique SK). new_val = std::numeric_limits::max(); auto_incr = new_val; // Store the largest value into auto_incr break; } new_val = n * inc + off; // Attempt to store the new value (plus 1 since m_auto_incr_val contains // the next available value) into the atomic value. If the current // value no longer matches what we have in 'last_val' this will fail and // we will repeat the loop (`last_val` will automatically get updated // with the current value). // // See above explanation for inc == 1 for why we use std::min. } while (!auto_incr.compare_exchange_weak(last_val, std::min(new_val + 1, max_val))); } *first_value = new_val; *nb_reserved_values = 1; } #ifndef DBUG_OFF /* Debugger help function */ static char dbug_item_print_buf[512]; const char *dbug_print_item(Item *const item) { char *const buf = dbug_item_print_buf; String str(buf, sizeof(dbug_item_print_buf), &my_charset_bin); str.length(0); if (!item) return "(Item*)nullptr"; item->print(&str, QT_ORDINARY); if (str.c_ptr() == buf) return buf; else return "Couldn't fit into buffer"; } #endif /*DBUG_OFF*/ /** SQL layer calls this function to push an index condition. @details The condition is for index keyno (only one condition can be pushed at a time). The caller guarantees that condition refers only to index fields; besides that, fields must have $field->part_of_key.set_bit(keyno) which means that (handler->index_flags(keyno, $keypart, 0) & HA_KEYREAD_ONLY) == 1 which means that field value can be restored from the index tuple. @return Part of condition we couldn't check (always nullptr). */ class Item *ha_rocksdb::idx_cond_push(uint keyno, class Item *const idx_cond) { DBUG_ENTER_FUNC(); DBUG_ASSERT(keyno != MAX_KEY); DBUG_ASSERT(idx_cond != nullptr); pushed_idx_cond = idx_cond; pushed_idx_cond_keyno = keyno; in_range_check_pushed_down = TRUE; /* We will check the whole condition */ DBUG_RETURN(nullptr); } /* Checks if inplace alter is supported for a given operation. */ my_core::enum_alter_inplace_result ha_rocksdb::check_if_supported_inplace_alter( TABLE *altered_table, my_core::Alter_inplace_info *const ha_alter_info) { DBUG_ENTER_FUNC(); DBUG_ASSERT(ha_alter_info != nullptr); if (ha_alter_info->handler_flags & ~(ALTER_DROP_NON_UNIQUE_NON_PRIM_INDEX | ALTER_DROP_UNIQUE_INDEX | ALTER_ADD_NON_UNIQUE_NON_PRIM_INDEX | ALTER_PARTITIONED | ALTER_ADD_UNIQUE_INDEX | ALTER_CHANGE_CREATE_OPTION)) { DBUG_RETURN(my_core::HA_ALTER_INPLACE_NOT_SUPPORTED); } /* We don't support unique keys on table w/ no primary keys */ if ((ha_alter_info->handler_flags & ALTER_ADD_UNIQUE_INDEX) && has_hidden_pk(altered_table)) { DBUG_RETURN(my_core::HA_ALTER_INPLACE_NOT_SUPPORTED); } /* We only support changing auto_increment for table options. */ if ((ha_alter_info->handler_flags & ALTER_CHANGE_CREATE_OPTION) && !(ha_alter_info->create_info->used_fields & HA_CREATE_USED_AUTO)) { DBUG_RETURN(my_core::HA_ALTER_INPLACE_NOT_SUPPORTED); } /* FIXME: MDEV-16099 Use alter algorithm=nocopy or algorithm=instant for non-InnoDB engine */ DBUG_RETURN(my_core::HA_ALTER_INPLACE_COPY_LOCK); } /** Allows the storage engine to update internal structures with concurrent writes blocked. If check_if_supported_inplace_alter() returns HA_ALTER_INPLACE_COPY_NO_LOCK or HA_ALTER_INPLACE_COPY_LOCK. this function is called with exclusive lock otherwise the same level of locking as for inplace_alter_table() will be used. @note Storage engines are responsible for reporting any errors by calling my_error()/print_error() @note If this function reports error, commit_inplace_alter_table() will be called with commit= false. @note For partitioning, failing to prepare one partition, means that commit_inplace_alter_table() will be called to roll back changes for all partitions. This means that commit_inplace_alter_table() might be called without prepare_inplace_alter_table() having been called first for a given partition. @param altered_table TABLE object for new version of table. @param ha_alter_info Structure describing changes to be done by ALTER TABLE and holding data used during in-place alter. @retval true Error @retval false Success */ bool ha_rocksdb::prepare_inplace_alter_table( TABLE *const altered_table, my_core::Alter_inplace_info *const ha_alter_info) { DBUG_ENTER_FUNC(); DBUG_ASSERT(altered_table != nullptr); DBUG_ASSERT(ha_alter_info != nullptr); Rdb_tbl_def *new_tdef = nullptr; std::shared_ptr *old_key_descr = nullptr; std::shared_ptr *new_key_descr = nullptr; uint old_n_keys = m_tbl_def->m_key_count; uint new_n_keys = altered_table->s->keys; std::unordered_set> added_indexes; std::unordered_set dropped_index_ids; uint n_dropped_keys = 0; uint n_added_keys = 0; ulonglong max_auto_incr = 0; if (ha_alter_info->handler_flags & (ALTER_DROP_NON_UNIQUE_NON_PRIM_INDEX | ALTER_DROP_UNIQUE_INDEX | ALTER_ADD_NON_UNIQUE_NON_PRIM_INDEX | ALTER_ADD_UNIQUE_INDEX)) { if (has_hidden_pk(altered_table)) { new_n_keys += 1; } const TABLE *const old_table = table; old_key_descr = m_tbl_def->m_key_descr_arr; new_key_descr = new std::shared_ptr[new_n_keys]; new_tdef = new Rdb_tbl_def(m_tbl_def->full_tablename()); new_tdef->m_key_descr_arr = new_key_descr; new_tdef->m_key_count = new_n_keys; new_tdef->m_auto_incr_val = m_tbl_def->m_auto_incr_val.load(std::memory_order_relaxed); new_tdef->m_hidden_pk_val = m_tbl_def->m_hidden_pk_val.load(std::memory_order_relaxed); if (create_key_defs(altered_table, new_tdef, table, m_tbl_def)) { /* Delete the new key descriptors */ delete[] new_key_descr; /* Explicitly mark as nullptr so we don't accidentally remove entries from data dictionary on cleanup (or cause double delete[]). */ new_tdef->m_key_descr_arr = nullptr; delete new_tdef; my_error(ER_KEY_CREATE_DURING_ALTER, MYF(0)); DBUG_RETURN(HA_EXIT_FAILURE); } uint i; uint j; /* Determine which(if any) key definition(s) need to be dropped */ for (i = 0; i < ha_alter_info->index_drop_count; i++) { const KEY *const dropped_key = ha_alter_info->index_drop_buffer[i]; for (j = 0; j < old_n_keys; j++) { const KEY *const old_key = &old_table->key_info[old_key_descr[j]->get_keyno()]; if (!compare_keys(old_key, dropped_key)) { dropped_index_ids.insert(old_key_descr[j]->get_gl_index_id()); break; } } } /* Determine which(if any) key definitions(s) need to be added */ int identical_indexes_found = 0; for (i = 0; i < ha_alter_info->index_add_count; i++) { const KEY *const added_key = &ha_alter_info->key_info_buffer[ha_alter_info->index_add_buffer[i]]; for (j = 0; j < new_n_keys; j++) { const KEY *const new_key = &altered_table->key_info[new_key_descr[j]->get_keyno()]; if (!compare_keys(new_key, added_key)) { /* Check for cases where an 'identical' index is being dropped and re-added in a single ALTER statement. Turn this into a no-op as the index has not changed. E.G. Unique index -> non-unique index requires no change Note that cases where the index name remains the same but the key-parts are changed is already handled in create_inplace_key_defs. In these cases the index needs to be rebuilt. */ if (dropped_index_ids.count(new_key_descr[j]->get_gl_index_id())) { dropped_index_ids.erase(new_key_descr[j]->get_gl_index_id()); identical_indexes_found++; } else { added_indexes.insert(new_key_descr[j]); } break; } } } n_dropped_keys = ha_alter_info->index_drop_count - identical_indexes_found; n_added_keys = ha_alter_info->index_add_count - identical_indexes_found; DBUG_ASSERT(dropped_index_ids.size() == n_dropped_keys); DBUG_ASSERT(added_indexes.size() == n_added_keys); DBUG_ASSERT(new_n_keys == (old_n_keys - n_dropped_keys + n_added_keys)); } if (ha_alter_info->handler_flags & ALTER_CHANGE_CREATE_OPTION) { if (!new_tdef) { new_tdef = m_tbl_def; } if (table->found_next_number_field) { max_auto_incr = load_auto_incr_value_from_index(); } } ha_alter_info->handler_ctx = new Rdb_inplace_alter_ctx( new_tdef, old_key_descr, new_key_descr, old_n_keys, new_n_keys, added_indexes, dropped_index_ids, n_added_keys, n_dropped_keys, max_auto_incr); DBUG_RETURN(HA_EXIT_SUCCESS); } /** Alter the table structure in-place with operations specified using HA_ALTER_FLAGS and Alter_inplace_info. The level of concurrency allowed during this operation depends on the return value from check_if_supported_inplace_alter(). @note Storage engines are responsible for reporting any errors by calling my_error()/print_error() @note If this function reports error, commit_inplace_alter_table() will be called with commit= false. @param altered_table TABLE object for new version of table. @param ha_alter_info Structure describing changes to be done by ALTER TABLE and holding data used during in-place alter. @retval true Error @retval false Success */ bool ha_rocksdb::inplace_alter_table( TABLE *const altered_table, my_core::Alter_inplace_info *const ha_alter_info) { DBUG_ENTER_FUNC(); DBUG_ASSERT(altered_table != nullptr); DBUG_ASSERT(ha_alter_info != nullptr); DBUG_ASSERT(ha_alter_info->handler_ctx != nullptr); Rdb_inplace_alter_ctx *const ctx = static_cast(ha_alter_info->handler_ctx); if (ha_alter_info->handler_flags & (ALTER_ADD_NON_UNIQUE_NON_PRIM_INDEX | ALTER_ADD_UNIQUE_INDEX)) { /* Buffers need to be set up again to account for new, possibly longer secondary keys. */ free_key_buffers(); DBUG_ASSERT(ctx != nullptr); /* If adding unique index, allocate special buffers for duplicate checking. */ int err; if ((err = alloc_key_buffers( altered_table, ctx->m_new_tdef, ha_alter_info->handler_flags & ALTER_ADD_UNIQUE_INDEX))) { my_error(ER_OUT_OF_RESOURCES, MYF(0)); DBUG_RETURN(err); } /* Populate all new secondary keys by scanning the primary key. */ if ((err = inplace_populate_sk(altered_table, ctx->m_added_indexes))) { my_error(ER_SK_POPULATE_DURING_ALTER, MYF(0)); DBUG_RETURN(HA_EXIT_FAILURE); } } DBUG_EXECUTE_IF("myrocks_simulate_index_create_rollback", { dbug_create_err_inplace_alter(); DBUG_RETURN(HA_EXIT_FAILURE); };); DBUG_RETURN(HA_EXIT_SUCCESS); } /** Scan the Primary Key index entries and populate the new secondary keys. */ int ha_rocksdb::inplace_populate_sk( TABLE *const new_table_arg, const std::unordered_set> &indexes) { DBUG_ENTER_FUNC(); int res = HA_EXIT_SUCCESS; const std::unique_ptr wb = dict_manager.begin(); rocksdb::WriteBatch *const batch = wb.get(); /* Update the data dictionary */ std::unordered_set create_index_ids; for (const auto &index : indexes) { create_index_ids.insert(index->get_gl_index_id()); } dict_manager.add_create_index(create_index_ids, batch); res = dict_manager.commit(batch); if (res != HA_EXIT_SUCCESS) { return res; } /* Add uncommitted key definitons to ddl_manager. We need to do this so that the property collector can find this keydef when it needs to update stats. The property collector looks for the keydef in the data dictionary, but it won't be there yet since this key definition is still in the creation process. */ ddl_manager.add_uncommitted_keydefs(indexes); const bool hidden_pk_exists = has_hidden_pk(table); Rdb_transaction *tx = get_or_create_tx(table->in_use); /* There is one specific scenario where m_sst_info may not be nullptr. This happens if the handler we're using happens to be the handler where the PK bulk load was done on. The sequence of events that lead to this is as follows (T1 is PK bulk load, T2 is SK alter table): T1: Execute last INSERT statement T1: Return TABLE and handler object back to Table_cache_manager T1: Close connection T2: Execute ALTER statement T2: Take same TABLE/handler from Table_cache_manager T2: Call closefrm which will call finalize_bulk_load on every other open table/handler *except* the one it's on. T2: Acquire stale snapshot of PK T1: Call finalize_bulk_load This is rare because usually, closefrm will call the destructor (and thus finalize_bulk_load) on the handler where PK bulk load is done. However, if the thread ids of the bulk load thread and the alter thread differ by a multiple of table_cache_instances (8 by default), then they hash to the same bucket in Table_cache_manager and the alter thread will not not call the destructor on the handler it is holding. Thus, its m_sst_info will not be nullptr. At this point, it is safe to refresh the snapshot because we know all other open handlers have been closed at this point, and the one we're on is the only one left. */ if (m_sst_info) { if ((res = finalize_bulk_load())) { DBUG_RETURN(res); } tx->commit(); } const ulonglong rdb_merge_buf_size = THDVAR(ha_thd(), merge_buf_size); const ulonglong rdb_merge_combine_read_size = THDVAR(ha_thd(), merge_combine_read_size); const ulonglong rdb_merge_tmp_file_removal_delay = THDVAR(ha_thd(), merge_tmp_file_removal_delay_ms); for (const auto &index : indexes) { bool is_unique_index = new_table_arg->key_info[index->get_keyno()].flags & HA_NOSAME; Rdb_index_merge rdb_merge(tx->get_rocksdb_tmpdir(), rdb_merge_buf_size, rdb_merge_combine_read_size, rdb_merge_tmp_file_removal_delay, index->get_cf()); if ((res = rdb_merge.init())) { DBUG_RETURN(res); } /* Note: We pass in the currently existing table + tbl_def object here, as the pk index position may have changed in the case of hidden primary keys. */ const uint pk = pk_index(table, m_tbl_def); ha_index_init(pk, true); /* Scan each record in the primary key in order */ for (res = index_first(table->record[0]); res == 0; res = index_next(table->record[0])) { longlong hidden_pk_id = 0; if (hidden_pk_exists && (res = read_hidden_pk_id_from_rowkey(&hidden_pk_id))) { // NO_LINT_DEBUG sql_print_error("Error retrieving hidden pk id."); ha_index_end(); DBUG_RETURN(res); } /* Create new secondary index entry */ const int new_packed_size = index->pack_record( new_table_arg, m_pack_buffer, table->record[0], m_sk_packed_tuple, &m_sk_tails, should_store_row_debug_checksums(), hidden_pk_id, 0, nullptr, nullptr, m_ttl_bytes); const rocksdb::Slice key = rocksdb::Slice( reinterpret_cast(m_sk_packed_tuple), new_packed_size); const rocksdb::Slice val = rocksdb::Slice(reinterpret_cast(m_sk_tails.ptr()), m_sk_tails.get_current_pos()); /* Add record to offset tree in preparation for writing out to disk in sorted chunks. */ if ((res = rdb_merge.add(key, val))) { ha_index_end(); DBUG_RETURN(res); } } if (res != HA_ERR_END_OF_FILE) { // NO_LINT_DEBUG sql_print_error("Error retrieving index entry from primary key."); ha_index_end(); DBUG_RETURN(res); } ha_index_end(); /* Perform an n-way merge of n sorted buffers on disk, then writes all results to RocksDB via SSTFileWriter API. */ rocksdb::Slice merge_key; rocksdb::Slice merge_val; struct unique_sk_buf_info sk_info; sk_info.dup_sk_buf = m_dup_sk_packed_tuple; sk_info.dup_sk_buf_old = m_dup_sk_packed_tuple_old; while ((res = rdb_merge.next(&merge_key, &merge_val)) == 0) { /* Perform uniqueness check if needed */ if (is_unique_index) { if (check_duplicate_sk(new_table_arg, *index, &merge_key, &sk_info)) { /* Duplicate entry found when trying to create unique secondary key. We need to unpack the record into new_table_arg->record[0] as it is used inside print_keydup_error so that the error message shows the duplicate record. */ if (index->unpack_record(new_table_arg, new_table_arg->record[0], &merge_key, nullptr, m_verify_row_debug_checksums)) { /* Should never reach here */ DBUG_ASSERT(0); } print_keydup_error(new_table_arg, &new_table_arg->key_info[index->get_keyno()], MYF(0)); DBUG_RETURN(ER_DUP_ENTRY); } } /* Insert key and slice to SST via SSTFileWriter API. */ if ((res = bulk_load_key(tx, *index, merge_key, merge_val, false))) { break; } } /* Here, res == -1 means that we are finished, while > 0 means an error occurred. */ if (res > 0) { // NO_LINT_DEBUG sql_print_error("Error while bulk loading keys in external merge sort."); DBUG_RETURN(res); } if ((res = tx->finish_bulk_load())) { // NO_LINT_DEBUG sql_print_error("Error finishing bulk load."); DBUG_RETURN(res); } } /* Explicitly tell jemalloc to clean up any unused dirty pages at this point. See https://reviews.facebook.net/D63723 for more details. */ purge_all_jemalloc_arenas(); DBUG_EXECUTE_IF("crash_during_online_index_creation", DBUG_SUICIDE();); DBUG_RETURN(res); } /** Commit or rollback the changes made during prepare_inplace_alter_table() and inplace_alter_table() inside the storage engine. Note that in case of rollback the allowed level of concurrency during this operation will be the same as for inplace_alter_table() and thus might be higher than during prepare_inplace_alter_table(). (For example, concurrent writes were blocked during prepare, but might not be during rollback). @note Storage engines are responsible for reporting any errors by calling my_error()/print_error() @note If this function with commit= true reports error, it will be called again with commit= false. @note In case of partitioning, this function might be called for rollback without prepare_inplace_alter_table() having been called first. Also partitioned tables sets ha_alter_info->group_commit_ctx to a NULL terminated array of the partitions handlers and if all of them are committed as one, then group_commit_ctx should be set to NULL to indicate to the partitioning handler that all partitions handlers are committed. @see prepare_inplace_alter_table(). @param altered_table TABLE object for new version of table. @param ha_alter_info Structure describing changes to be done by ALTER TABLE and holding data used during in-place alter. @param commit True => Commit, False => Rollback. @retval true Error @retval false Success */ bool ha_rocksdb::commit_inplace_alter_table( my_core::TABLE *const altered_table, my_core::Alter_inplace_info *const ha_alter_info, bool commit) { DBUG_ENTER_FUNC(); DBUG_ASSERT(altered_table != nullptr); DBUG_ASSERT(ha_alter_info != nullptr); Rdb_inplace_alter_ctx *const ctx0 = static_cast(ha_alter_info->handler_ctx); DEBUG_SYNC(ha_thd(), "rocksdb.commit_in_place_alter_table"); /* IMPORTANT: When rollback is requested, mysql will abort with an assertion failure. That means every failed commit during inplace alter table will result in a fatal error on the server. Indexes ongoing creation will be detected when the server restarts, and dropped. For partitioned tables, a rollback call to this function (commit == false) is done for each partition. A successful commit call only executes once for all partitions. */ if (!commit) { /* If ctx has not been created yet, nothing to do here */ if (!ctx0) { DBUG_RETURN(HA_EXIT_SUCCESS); } /* Cannot call destructor for Rdb_tbl_def directly because we don't want to erase the mappings inside the ddl_manager, as the old_key_descr is still using them. */ if (ctx0->m_new_key_descr) { /* Delete the new key descriptors */ for (uint i = 0; i < ctx0->m_new_tdef->m_key_count; i++) { ctx0->m_new_key_descr[i] = nullptr; } delete[] ctx0->m_new_key_descr; ctx0->m_new_key_descr = nullptr; ctx0->m_new_tdef->m_key_descr_arr = nullptr; delete ctx0->m_new_tdef; } /* Remove uncommitted key definitons from ddl_manager */ ddl_manager.remove_uncommitted_keydefs(ctx0->m_added_indexes); /* Rollback any partially created indexes */ dict_manager.rollback_ongoing_index_creation(); DBUG_RETURN(HA_EXIT_SUCCESS); } DBUG_ASSERT(ctx0); /* For partitioned tables, we need to commit all changes to all tables at once, unlike in the other inplace alter API methods. */ inplace_alter_handler_ctx **ctx_array; inplace_alter_handler_ctx *ctx_single[2]; if (ha_alter_info->group_commit_ctx) { DBUG_EXECUTE_IF("crash_during_index_creation_partition", DBUG_SUICIDE();); ctx_array = ha_alter_info->group_commit_ctx; } else { ctx_single[0] = ctx0; ctx_single[1] = nullptr; ctx_array = ctx_single; } DBUG_ASSERT(ctx0 == ctx_array[0]); ha_alter_info->group_commit_ctx = nullptr; if (ha_alter_info->handler_flags & (ALTER_DROP_NON_UNIQUE_NON_PRIM_INDEX | ALTER_DROP_UNIQUE_INDEX | ALTER_ADD_NON_UNIQUE_NON_PRIM_INDEX | ALTER_ADD_UNIQUE_INDEX)) { const std::unique_ptr wb = dict_manager.begin(); rocksdb::WriteBatch *const batch = wb.get(); std::unordered_set create_index_ids; m_tbl_def = ctx0->m_new_tdef; m_key_descr_arr = m_tbl_def->m_key_descr_arr; m_pk_descr = m_key_descr_arr[pk_index(altered_table, m_tbl_def)]; dict_manager.lock(); for (inplace_alter_handler_ctx **pctx = ctx_array; *pctx; pctx++) { Rdb_inplace_alter_ctx *const ctx = static_cast(*pctx); /* Mark indexes to be dropped */ dict_manager.add_drop_index(ctx->m_dropped_index_ids, batch); for (const auto &index : ctx->m_added_indexes) { create_index_ids.insert(index->get_gl_index_id()); } if (ddl_manager.put_and_write(ctx->m_new_tdef, batch)) { /* Failed to write new entry into data dictionary, this should never happen. */ DBUG_ASSERT(0); } /* Remove uncommitted key definitons from ddl_manager, as they are now committed into the data dictionary. */ ddl_manager.remove_uncommitted_keydefs(ctx->m_added_indexes); } if (dict_manager.commit(batch)) { /* Should never reach here. We assume MyRocks will abort if commit fails. */ DBUG_ASSERT(0); } dict_manager.unlock(); /* Mark ongoing create indexes as finished/remove from data dictionary */ dict_manager.finish_indexes_operation( create_index_ids, Rdb_key_def::DDL_CREATE_INDEX_ONGOING); rdb_drop_idx_thread.signal(); } if (ha_alter_info->handler_flags & ALTER_CHANGE_CREATE_OPTION) { const std::unique_ptr wb = dict_manager.begin(); rocksdb::WriteBatch *const batch = wb.get(); std::unordered_set create_index_ids; ulonglong auto_incr_val = ha_alter_info->create_info->auto_increment_value; for (inplace_alter_handler_ctx **pctx = ctx_array; *pctx; pctx++) { Rdb_inplace_alter_ctx *const ctx = static_cast(*pctx); auto_incr_val = std::max(auto_incr_val, ctx->m_max_auto_incr); dict_manager.put_auto_incr_val( batch, ctx->m_new_tdef->get_autoincr_gl_index_id(), auto_incr_val, true /* overwrite */); ctx->m_new_tdef->m_auto_incr_val = auto_incr_val; } if (dict_manager.commit(batch)) { DBUG_ASSERT(0); } } DBUG_RETURN(HA_EXIT_SUCCESS); } #define SHOW_FNAME(name) rocksdb_show_##name #define DEF_SHOW_FUNC(name, key) \ static int SHOW_FNAME(name)(MYSQL_THD thd, SHOW_VAR * var, char *buff) { \ rocksdb_status_counters.name = \ rocksdb_stats->getTickerCount(rocksdb::key); \ var->type = SHOW_LONGLONG; \ var->value = (char *)&rocksdb_status_counters.name; \ return HA_EXIT_SUCCESS; \ } #define DEF_STATUS_VAR(name) \ { "rocksdb_" #name, (char *)&SHOW_FNAME(name), SHOW_FUNC } #define DEF_STATUS_VAR_PTR(name, ptr, option) \ { "rocksdb_" name, (char *)ptr, option } #define DEF_STATUS_VAR_FUNC(name, ptr, option) \ { name, reinterpret_cast(ptr), option } struct rocksdb_status_counters_t { uint64_t block_cache_miss; uint64_t block_cache_hit; uint64_t block_cache_add; uint64_t block_cache_add_failures; uint64_t block_cache_index_miss; uint64_t block_cache_index_hit; uint64_t block_cache_index_add; uint64_t block_cache_index_bytes_insert; uint64_t block_cache_index_bytes_evict; uint64_t block_cache_filter_miss; uint64_t block_cache_filter_hit; uint64_t block_cache_filter_add; uint64_t block_cache_filter_bytes_insert; uint64_t block_cache_filter_bytes_evict; uint64_t block_cache_bytes_read; uint64_t block_cache_bytes_write; uint64_t block_cache_data_bytes_insert; uint64_t block_cache_data_miss; uint64_t block_cache_data_hit; uint64_t block_cache_data_add; uint64_t bloom_filter_useful; uint64_t bloom_filter_full_positive; uint64_t bloom_filter_full_true_positive; uint64_t memtable_hit; uint64_t memtable_miss; uint64_t get_hit_l0; uint64_t get_hit_l1; uint64_t get_hit_l2_and_up; uint64_t compaction_key_drop_new; uint64_t compaction_key_drop_obsolete; uint64_t compaction_key_drop_user; uint64_t number_keys_written; uint64_t number_keys_read; uint64_t number_keys_updated; uint64_t bytes_written; uint64_t bytes_read; uint64_t number_db_seek; uint64_t number_db_seek_found; uint64_t number_db_next; uint64_t number_db_next_found; uint64_t number_db_prev; uint64_t number_db_prev_found; uint64_t iter_bytes_read; uint64_t no_file_closes; uint64_t no_file_opens; uint64_t no_file_errors; uint64_t stall_micros; uint64_t num_iterators; uint64_t number_multiget_get; uint64_t number_multiget_keys_read; uint64_t number_multiget_bytes_read; uint64_t number_deletes_filtered; uint64_t number_merge_failures; uint64_t bloom_filter_prefix_checked; uint64_t bloom_filter_prefix_useful; uint64_t number_reseeks_iteration; uint64_t getupdatessince_calls; uint64_t block_cachecompressed_miss; uint64_t block_cachecompressed_hit; uint64_t wal_synced; uint64_t wal_bytes; uint64_t write_self; uint64_t write_other; uint64_t write_timedout; uint64_t write_wal; uint64_t flush_write_bytes; uint64_t compact_read_bytes; uint64_t compact_write_bytes; uint64_t number_superversion_acquires; uint64_t number_superversion_releases; uint64_t number_superversion_cleanups; uint64_t number_block_not_compressed; }; static rocksdb_status_counters_t rocksdb_status_counters; DEF_SHOW_FUNC(block_cache_miss, BLOCK_CACHE_MISS) DEF_SHOW_FUNC(block_cache_hit, BLOCK_CACHE_HIT) DEF_SHOW_FUNC(block_cache_add, BLOCK_CACHE_ADD) DEF_SHOW_FUNC(block_cache_add_failures, BLOCK_CACHE_ADD_FAILURES) DEF_SHOW_FUNC(block_cache_index_miss, BLOCK_CACHE_INDEX_MISS) DEF_SHOW_FUNC(block_cache_index_hit, BLOCK_CACHE_INDEX_HIT) DEF_SHOW_FUNC(block_cache_index_add, BLOCK_CACHE_INDEX_ADD) DEF_SHOW_FUNC(block_cache_index_bytes_insert, BLOCK_CACHE_INDEX_BYTES_INSERT) DEF_SHOW_FUNC(block_cache_index_bytes_evict, BLOCK_CACHE_INDEX_BYTES_EVICT) DEF_SHOW_FUNC(block_cache_filter_miss, BLOCK_CACHE_FILTER_MISS) DEF_SHOW_FUNC(block_cache_filter_hit, BLOCK_CACHE_FILTER_HIT) DEF_SHOW_FUNC(block_cache_filter_add, BLOCK_CACHE_FILTER_ADD) DEF_SHOW_FUNC(block_cache_filter_bytes_insert, BLOCK_CACHE_FILTER_BYTES_INSERT) DEF_SHOW_FUNC(block_cache_filter_bytes_evict, BLOCK_CACHE_FILTER_BYTES_EVICT) DEF_SHOW_FUNC(block_cache_bytes_read, BLOCK_CACHE_BYTES_READ) DEF_SHOW_FUNC(block_cache_bytes_write, BLOCK_CACHE_BYTES_WRITE) DEF_SHOW_FUNC(block_cache_data_bytes_insert, BLOCK_CACHE_DATA_BYTES_INSERT) DEF_SHOW_FUNC(block_cache_data_miss, BLOCK_CACHE_DATA_MISS) DEF_SHOW_FUNC(block_cache_data_hit, BLOCK_CACHE_DATA_HIT) DEF_SHOW_FUNC(block_cache_data_add, BLOCK_CACHE_DATA_ADD) DEF_SHOW_FUNC(bloom_filter_useful, BLOOM_FILTER_USEFUL) DEF_SHOW_FUNC(bloom_filter_full_positive, BLOOM_FILTER_FULL_POSITIVE) DEF_SHOW_FUNC(bloom_filter_full_true_positive, BLOOM_FILTER_FULL_TRUE_POSITIVE) DEF_SHOW_FUNC(memtable_hit, MEMTABLE_HIT) DEF_SHOW_FUNC(memtable_miss, MEMTABLE_MISS) DEF_SHOW_FUNC(get_hit_l0, GET_HIT_L0) DEF_SHOW_FUNC(get_hit_l1, GET_HIT_L1) DEF_SHOW_FUNC(get_hit_l2_and_up, GET_HIT_L2_AND_UP) DEF_SHOW_FUNC(compaction_key_drop_new, COMPACTION_KEY_DROP_NEWER_ENTRY) DEF_SHOW_FUNC(compaction_key_drop_obsolete, COMPACTION_KEY_DROP_OBSOLETE) DEF_SHOW_FUNC(compaction_key_drop_user, COMPACTION_KEY_DROP_USER) DEF_SHOW_FUNC(number_keys_written, NUMBER_KEYS_WRITTEN) DEF_SHOW_FUNC(number_keys_read, NUMBER_KEYS_READ) DEF_SHOW_FUNC(number_keys_updated, NUMBER_KEYS_UPDATED) DEF_SHOW_FUNC(bytes_written, BYTES_WRITTEN) DEF_SHOW_FUNC(bytes_read, BYTES_READ) DEF_SHOW_FUNC(number_db_seek, NUMBER_DB_SEEK) DEF_SHOW_FUNC(number_db_seek_found, NUMBER_DB_SEEK_FOUND) DEF_SHOW_FUNC(number_db_next, NUMBER_DB_NEXT) DEF_SHOW_FUNC(number_db_next_found, NUMBER_DB_NEXT_FOUND) DEF_SHOW_FUNC(number_db_prev, NUMBER_DB_PREV) DEF_SHOW_FUNC(number_db_prev_found, NUMBER_DB_PREV_FOUND) DEF_SHOW_FUNC(iter_bytes_read, ITER_BYTES_READ) DEF_SHOW_FUNC(no_file_closes, NO_FILE_CLOSES) DEF_SHOW_FUNC(no_file_opens, NO_FILE_OPENS) DEF_SHOW_FUNC(no_file_errors, NO_FILE_ERRORS) DEF_SHOW_FUNC(stall_micros, STALL_MICROS) DEF_SHOW_FUNC(num_iterators, NO_ITERATORS) DEF_SHOW_FUNC(number_multiget_get, NUMBER_MULTIGET_CALLS) DEF_SHOW_FUNC(number_multiget_keys_read, NUMBER_MULTIGET_KEYS_READ) DEF_SHOW_FUNC(number_multiget_bytes_read, NUMBER_MULTIGET_BYTES_READ) DEF_SHOW_FUNC(number_deletes_filtered, NUMBER_FILTERED_DELETES) DEF_SHOW_FUNC(number_merge_failures, NUMBER_MERGE_FAILURES) DEF_SHOW_FUNC(bloom_filter_prefix_checked, BLOOM_FILTER_PREFIX_CHECKED) DEF_SHOW_FUNC(bloom_filter_prefix_useful, BLOOM_FILTER_PREFIX_USEFUL) DEF_SHOW_FUNC(number_reseeks_iteration, NUMBER_OF_RESEEKS_IN_ITERATION) DEF_SHOW_FUNC(getupdatessince_calls, GET_UPDATES_SINCE_CALLS) DEF_SHOW_FUNC(block_cachecompressed_miss, BLOCK_CACHE_COMPRESSED_MISS) DEF_SHOW_FUNC(block_cachecompressed_hit, BLOCK_CACHE_COMPRESSED_HIT) DEF_SHOW_FUNC(wal_synced, WAL_FILE_SYNCED) DEF_SHOW_FUNC(wal_bytes, WAL_FILE_BYTES) DEF_SHOW_FUNC(write_self, WRITE_DONE_BY_SELF) DEF_SHOW_FUNC(write_other, WRITE_DONE_BY_OTHER) DEF_SHOW_FUNC(write_timedout, WRITE_TIMEDOUT) DEF_SHOW_FUNC(write_wal, WRITE_WITH_WAL) DEF_SHOW_FUNC(flush_write_bytes, FLUSH_WRITE_BYTES) DEF_SHOW_FUNC(compact_read_bytes, COMPACT_READ_BYTES) DEF_SHOW_FUNC(compact_write_bytes, COMPACT_WRITE_BYTES) DEF_SHOW_FUNC(number_superversion_acquires, NUMBER_SUPERVERSION_ACQUIRES) DEF_SHOW_FUNC(number_superversion_releases, NUMBER_SUPERVERSION_RELEASES) DEF_SHOW_FUNC(number_superversion_cleanups, NUMBER_SUPERVERSION_CLEANUPS) DEF_SHOW_FUNC(number_block_not_compressed, NUMBER_BLOCK_NOT_COMPRESSED) static void myrocks_update_status() { export_stats.rows_deleted = global_stats.rows[ROWS_DELETED]; export_stats.rows_inserted = global_stats.rows[ROWS_INSERTED]; export_stats.rows_read = global_stats.rows[ROWS_READ]; export_stats.rows_updated = global_stats.rows[ROWS_UPDATED]; export_stats.rows_deleted_blind = global_stats.rows[ROWS_DELETED_BLIND]; export_stats.rows_expired = global_stats.rows[ROWS_EXPIRED]; export_stats.rows_filtered = global_stats.rows[ROWS_FILTERED]; export_stats.system_rows_deleted = global_stats.system_rows[ROWS_DELETED]; export_stats.system_rows_inserted = global_stats.system_rows[ROWS_INSERTED]; export_stats.system_rows_read = global_stats.system_rows[ROWS_READ]; export_stats.system_rows_updated = global_stats.system_rows[ROWS_UPDATED]; export_stats.queries_point = global_stats.queries[QUERIES_POINT]; export_stats.queries_range = global_stats.queries[QUERIES_RANGE]; export_stats.covered_secondary_key_lookups = global_stats.covered_secondary_key_lookups; } static void myrocks_update_memory_status() { std::vector dbs; std::unordered_set cache_set; dbs.push_back(rdb); std::map temp_usage_by_type; rocksdb::MemoryUtil::GetApproximateMemoryUsageByType(dbs, cache_set, &temp_usage_by_type); memory_stats.memtable_total = temp_usage_by_type[rocksdb::MemoryUtil::kMemTableTotal]; memory_stats.memtable_unflushed = temp_usage_by_type[rocksdb::MemoryUtil::kMemTableUnFlushed]; } static SHOW_VAR myrocks_status_variables[] = { DEF_STATUS_VAR_FUNC("rows_deleted", &export_stats.rows_deleted, SHOW_LONGLONG), DEF_STATUS_VAR_FUNC("rows_inserted", &export_stats.rows_inserted, SHOW_LONGLONG), DEF_STATUS_VAR_FUNC("rows_read", &export_stats.rows_read, SHOW_LONGLONG), DEF_STATUS_VAR_FUNC("rows_updated", &export_stats.rows_updated, SHOW_LONGLONG), DEF_STATUS_VAR_FUNC("rows_deleted_blind", &export_stats.rows_deleted_blind, SHOW_LONGLONG), DEF_STATUS_VAR_FUNC("rows_expired", &export_stats.rows_expired, SHOW_LONGLONG), DEF_STATUS_VAR_FUNC("rows_filtered", &export_stats.rows_filtered, SHOW_LONGLONG), DEF_STATUS_VAR_FUNC("system_rows_deleted", &export_stats.system_rows_deleted, SHOW_LONGLONG), DEF_STATUS_VAR_FUNC("system_rows_inserted", &export_stats.system_rows_inserted, SHOW_LONGLONG), DEF_STATUS_VAR_FUNC("system_rows_read", &export_stats.system_rows_read, SHOW_LONGLONG), DEF_STATUS_VAR_FUNC("system_rows_updated", &export_stats.system_rows_updated, SHOW_LONGLONG), DEF_STATUS_VAR_FUNC("memtable_total", &memory_stats.memtable_total, SHOW_LONGLONG), DEF_STATUS_VAR_FUNC("memtable_unflushed", &memory_stats.memtable_unflushed, SHOW_LONGLONG), DEF_STATUS_VAR_FUNC("queries_point", &export_stats.queries_point, SHOW_LONGLONG), DEF_STATUS_VAR_FUNC("queries_range", &export_stats.queries_range, SHOW_LONGLONG), DEF_STATUS_VAR_FUNC("covered_secondary_key_lookups", &export_stats.covered_secondary_key_lookups, SHOW_LONGLONG), {NullS, NullS, SHOW_LONG}}; static void show_myrocks_vars(THD *thd, SHOW_VAR *var, char *buff) { myrocks_update_status(); myrocks_update_memory_status(); var->type = SHOW_ARRAY; var->value = reinterpret_cast(&myrocks_status_variables); } static ulonglong io_stall_prop_value(const std::map &props, const std::string &key) { std::map::const_iterator iter = props.find("io_stalls." + key); if (iter != props.end()) { return std::stoull(iter->second); } else { DBUG_PRINT("warning", ("RocksDB GetMapPropery hasn't returned key=%s", key.c_str())); DBUG_ASSERT(0); return 0; } } static void update_rocksdb_stall_status() { st_io_stall_stats local_io_stall_stats; for (const auto &cf_name : cf_manager.get_cf_names()) { rocksdb::ColumnFamilyHandle *cfh = cf_manager.get_cf(cf_name); if (cfh == nullptr) { continue; } std::map props; if (!rdb->GetMapProperty(cfh, "rocksdb.cfstats", &props)) { continue; } local_io_stall_stats.level0_slowdown += io_stall_prop_value(props, "level0_slowdown"); local_io_stall_stats.level0_slowdown_with_compaction += io_stall_prop_value(props, "level0_slowdown_with_compaction"); local_io_stall_stats.level0_numfiles += io_stall_prop_value(props, "level0_numfiles"); local_io_stall_stats.level0_numfiles_with_compaction += io_stall_prop_value(props, "level0_numfiles_with_compaction"); local_io_stall_stats.stop_for_pending_compaction_bytes += io_stall_prop_value(props, "stop_for_pending_compaction_bytes"); local_io_stall_stats.slowdown_for_pending_compaction_bytes += io_stall_prop_value(props, "slowdown_for_pending_compaction_bytes"); local_io_stall_stats.memtable_compaction += io_stall_prop_value(props, "memtable_compaction"); local_io_stall_stats.memtable_slowdown += io_stall_prop_value(props, "memtable_slowdown"); local_io_stall_stats.total_stop += io_stall_prop_value(props, "total_stop"); local_io_stall_stats.total_slowdown += io_stall_prop_value(props, "total_slowdown"); } io_stall_stats = local_io_stall_stats; } static SHOW_VAR rocksdb_stall_status_variables[] = { DEF_STATUS_VAR_FUNC("l0_file_count_limit_slowdowns", &io_stall_stats.level0_slowdown, SHOW_LONGLONG), DEF_STATUS_VAR_FUNC("locked_l0_file_count_limit_slowdowns", &io_stall_stats.level0_slowdown_with_compaction, SHOW_LONGLONG), DEF_STATUS_VAR_FUNC("l0_file_count_limit_stops", &io_stall_stats.level0_numfiles, SHOW_LONGLONG), DEF_STATUS_VAR_FUNC("locked_l0_file_count_limit_stops", &io_stall_stats.level0_numfiles_with_compaction, SHOW_LONGLONG), DEF_STATUS_VAR_FUNC("pending_compaction_limit_stops", &io_stall_stats.stop_for_pending_compaction_bytes, SHOW_LONGLONG), DEF_STATUS_VAR_FUNC("pending_compaction_limit_slowdowns", &io_stall_stats.slowdown_for_pending_compaction_bytes, SHOW_LONGLONG), DEF_STATUS_VAR_FUNC("memtable_limit_stops", &io_stall_stats.memtable_compaction, SHOW_LONGLONG), DEF_STATUS_VAR_FUNC("memtable_limit_slowdowns", &io_stall_stats.memtable_slowdown, SHOW_LONGLONG), DEF_STATUS_VAR_FUNC("total_stops", &io_stall_stats.total_stop, SHOW_LONGLONG), DEF_STATUS_VAR_FUNC("total_slowdowns", &io_stall_stats.total_slowdown, SHOW_LONGLONG), // end of the array marker {NullS, NullS, SHOW_LONG}}; static void show_rocksdb_stall_vars(THD *thd, SHOW_VAR *var, char *buff) { update_rocksdb_stall_status(); var->type = SHOW_ARRAY; var->value = reinterpret_cast(&rocksdb_stall_status_variables); } static SHOW_VAR rocksdb_status_vars[] = { DEF_STATUS_VAR(block_cache_miss), DEF_STATUS_VAR(block_cache_hit), DEF_STATUS_VAR(block_cache_add), DEF_STATUS_VAR(block_cache_add_failures), DEF_STATUS_VAR(block_cache_index_miss), DEF_STATUS_VAR(block_cache_index_hit), DEF_STATUS_VAR(block_cache_index_add), DEF_STATUS_VAR(block_cache_index_bytes_insert), DEF_STATUS_VAR(block_cache_index_bytes_evict), DEF_STATUS_VAR(block_cache_filter_miss), DEF_STATUS_VAR(block_cache_filter_hit), DEF_STATUS_VAR(block_cache_filter_add), DEF_STATUS_VAR(block_cache_filter_bytes_insert), DEF_STATUS_VAR(block_cache_filter_bytes_evict), DEF_STATUS_VAR(block_cache_bytes_read), DEF_STATUS_VAR(block_cache_bytes_write), DEF_STATUS_VAR(block_cache_data_bytes_insert), DEF_STATUS_VAR(block_cache_data_miss), DEF_STATUS_VAR(block_cache_data_hit), DEF_STATUS_VAR(block_cache_data_add), DEF_STATUS_VAR(bloom_filter_useful), DEF_STATUS_VAR(bloom_filter_full_positive), DEF_STATUS_VAR(bloom_filter_full_true_positive), DEF_STATUS_VAR(memtable_hit), DEF_STATUS_VAR(memtable_miss), DEF_STATUS_VAR(get_hit_l0), DEF_STATUS_VAR(get_hit_l1), DEF_STATUS_VAR(get_hit_l2_and_up), DEF_STATUS_VAR(compaction_key_drop_new), DEF_STATUS_VAR(compaction_key_drop_obsolete), DEF_STATUS_VAR(compaction_key_drop_user), DEF_STATUS_VAR(number_keys_written), DEF_STATUS_VAR(number_keys_read), DEF_STATUS_VAR(number_keys_updated), DEF_STATUS_VAR(bytes_written), DEF_STATUS_VAR(bytes_read), DEF_STATUS_VAR(number_db_seek), DEF_STATUS_VAR(number_db_seek_found), DEF_STATUS_VAR(number_db_next), DEF_STATUS_VAR(number_db_next_found), DEF_STATUS_VAR(number_db_prev), DEF_STATUS_VAR(number_db_prev_found), DEF_STATUS_VAR(iter_bytes_read), DEF_STATUS_VAR(no_file_closes), DEF_STATUS_VAR(no_file_opens), DEF_STATUS_VAR(no_file_errors), DEF_STATUS_VAR(stall_micros), DEF_STATUS_VAR(num_iterators), DEF_STATUS_VAR(number_multiget_get), DEF_STATUS_VAR(number_multiget_keys_read), DEF_STATUS_VAR(number_multiget_bytes_read), DEF_STATUS_VAR(number_deletes_filtered), DEF_STATUS_VAR(number_merge_failures), DEF_STATUS_VAR(bloom_filter_prefix_checked), DEF_STATUS_VAR(bloom_filter_prefix_useful), DEF_STATUS_VAR(number_reseeks_iteration), DEF_STATUS_VAR(getupdatessince_calls), DEF_STATUS_VAR(block_cachecompressed_miss), DEF_STATUS_VAR(block_cachecompressed_hit), DEF_STATUS_VAR(wal_synced), DEF_STATUS_VAR(wal_bytes), DEF_STATUS_VAR(write_self), DEF_STATUS_VAR(write_other), DEF_STATUS_VAR(write_timedout), DEF_STATUS_VAR(write_wal), DEF_STATUS_VAR(flush_write_bytes), DEF_STATUS_VAR(compact_read_bytes), DEF_STATUS_VAR(compact_write_bytes), DEF_STATUS_VAR(number_superversion_acquires), DEF_STATUS_VAR(number_superversion_releases), DEF_STATUS_VAR(number_superversion_cleanups), DEF_STATUS_VAR(number_block_not_compressed), DEF_STATUS_VAR_PTR("row_lock_deadlocks", &rocksdb_row_lock_deadlocks, SHOW_LONGLONG), DEF_STATUS_VAR_PTR("row_lock_wait_timeouts", &rocksdb_row_lock_wait_timeouts, SHOW_LONGLONG), DEF_STATUS_VAR_PTR("snapshot_conflict_errors", &rocksdb_snapshot_conflict_errors, SHOW_LONGLONG), DEF_STATUS_VAR_PTR("wal_group_syncs", &rocksdb_wal_group_syncs, SHOW_LONGLONG), DEF_STATUS_VAR_PTR("manual_compactions_processed", &rocksdb_manual_compactions_processed, SHOW_LONGLONG), DEF_STATUS_VAR_PTR("manual_compactions_running", &rocksdb_manual_compactions_running, SHOW_LONGLONG), DEF_STATUS_VAR_PTR("number_sst_entry_put", &rocksdb_num_sst_entry_put, SHOW_LONGLONG), DEF_STATUS_VAR_PTR("number_sst_entry_delete", &rocksdb_num_sst_entry_delete, SHOW_LONGLONG), DEF_STATUS_VAR_PTR("number_sst_entry_singledelete", &rocksdb_num_sst_entry_singledelete, SHOW_LONGLONG), DEF_STATUS_VAR_PTR("number_sst_entry_merge", &rocksdb_num_sst_entry_merge, SHOW_LONGLONG), DEF_STATUS_VAR_PTR("number_sst_entry_other", &rocksdb_num_sst_entry_other, SHOW_LONGLONG), // the variables generated by SHOW_FUNC are sorted only by prefix (first // arg in the tuple below), so make sure it is unique to make sorting // deterministic as quick sort is not stable {"rocksdb", reinterpret_cast(&show_myrocks_vars), SHOW_FUNC}, {"rocksdb_stall", reinterpret_cast(&show_rocksdb_stall_vars), SHOW_FUNC}, {NullS, NullS, SHOW_LONG}}; /* Background thread's main logic */ void Rdb_background_thread::run() { // How many seconds to wait till flushing the WAL next time. const int WAKE_UP_INTERVAL = 1; timespec ts_next_sync; set_timespec(ts_next_sync, WAKE_UP_INTERVAL); for (;;) { // Wait until the next timeout or until we receive a signal to stop the // thread. Request to stop the thread should only be triggered when the // storage engine is being unloaded. RDB_MUTEX_LOCK_CHECK(m_signal_mutex); const auto ret MY_ATTRIBUTE((__unused__)) = mysql_cond_timedwait(&m_signal_cond, &m_signal_mutex, &ts_next_sync); // Check that we receive only the expected error codes. DBUG_ASSERT(ret == 0 || ret == ETIMEDOUT); const bool local_stop = m_stop; const bool local_save_stats = m_save_stats; reset(); RDB_MUTEX_UNLOCK_CHECK(m_signal_mutex); if (local_stop) { // If we're here then that's because condition variable was signaled by // another thread and we're shutting down. Break out the loop to make // sure that shutdown thread can proceed. break; } // This path should be taken only when the timer expired. DBUG_ASSERT(ret == ETIMEDOUT); if (local_save_stats) { ddl_manager.persist_stats(); } // Set the next timestamp for mysql_cond_timedwait() (which ends up calling // pthread_cond_timedwait()) to wait on. set_timespec(ts_next_sync, WAKE_UP_INTERVAL); // Flush the WAL. Sync it for both background and never modes to copy // InnoDB's behavior. For mode never, the wal file isn't even written, // whereas background writes to the wal file, but issues the syncs in a // background thread. if (rdb && (rocksdb_flush_log_at_trx_commit != FLUSH_LOG_SYNC) && !rocksdb_db_options->allow_mmap_writes) { const rocksdb::Status s = rdb->FlushWAL(true); if (!s.ok()) { rdb_handle_io_error(s, RDB_IO_ERROR_BG_THREAD); } } // Recalculate statistics for indexes. if (rocksdb_stats_recalc_rate) { std::unordered_map> to_recalc; if (rdb_indexes_to_recalc.empty()) { struct Rdb_index_collector : public Rdb_tables_scanner { int add_table(Rdb_tbl_def *tdef) override { for (uint i = 0; i < tdef->m_key_count; i++) { rdb_indexes_to_recalc.push_back( tdef->m_key_descr_arr[i]->get_gl_index_id()); } return HA_EXIT_SUCCESS; } } collector; ddl_manager.scan_for_tables(&collector); } while (to_recalc.size() < rocksdb_stats_recalc_rate && !rdb_indexes_to_recalc.empty()) { const auto index_id = rdb_indexes_to_recalc.back(); rdb_indexes_to_recalc.pop_back(); std::shared_ptr keydef = ddl_manager.safe_find(index_id); if (keydef) { to_recalc.insert(std::make_pair(keydef->get_gl_index_id(), keydef)); } } if (!to_recalc.empty()) { calculate_stats(to_recalc, false); } } } // save remaining stats which might've left unsaved ddl_manager.persist_stats(); } /* A background thread to handle manual compactions, except for dropping indexes/tables. Every second, it checks pending manual compactions, and it calls CompactRange if there is. */ void Rdb_manual_compaction_thread::run() { mysql_mutex_init(0, &m_mc_mutex, MY_MUTEX_INIT_FAST); RDB_MUTEX_LOCK_CHECK(m_signal_mutex); for (;;) { if (m_stop) { break; } timespec ts; set_timespec(ts, 1); const auto ret MY_ATTRIBUTE((__unused__)) = mysql_cond_timedwait(&m_signal_cond, &m_signal_mutex, &ts); if (m_stop) { break; } // make sure, no program error is returned DBUG_ASSERT(ret == 0 || ret == ETIMEDOUT); RDB_MUTEX_UNLOCK_CHECK(m_signal_mutex); RDB_MUTEX_LOCK_CHECK(m_mc_mutex); // Grab the first item and proceed, if not empty. if (m_requests.empty()) { RDB_MUTEX_UNLOCK_CHECK(m_mc_mutex); RDB_MUTEX_LOCK_CHECK(m_signal_mutex); continue; } Manual_compaction_request &mcr = m_requests.begin()->second; DBUG_ASSERT(mcr.cf != nullptr); DBUG_ASSERT(mcr.state == Manual_compaction_request::INITED); mcr.state = Manual_compaction_request::RUNNING; RDB_MUTEX_UNLOCK_CHECK(m_mc_mutex); DBUG_ASSERT(mcr.state == Manual_compaction_request::RUNNING); // NO_LINT_DEBUG sql_print_information("Manual Compaction id %d cf %s started.", mcr.mc_id, mcr.cf->GetName().c_str()); rocksdb_manual_compactions_running++; if (rocksdb_debug_manual_compaction_delay > 0) { my_sleep(rocksdb_debug_manual_compaction_delay * 1000000); } // CompactRange may take a very long time. On clean shutdown, // it is cancelled by CancelAllBackgroundWork, then status is // set to shutdownInProgress. const rocksdb::Status s = rdb->CompactRange( getCompactRangeOptions(mcr.concurrency), mcr.cf, mcr.start, mcr.limit); rocksdb_manual_compactions_running--; if (s.ok()) { // NO_LINT_DEBUG sql_print_information("Manual Compaction id %d cf %s ended.", mcr.mc_id, mcr.cf->GetName().c_str()); } else { // NO_LINT_DEBUG sql_print_information("Manual Compaction id %d cf %s aborted. %s", mcr.mc_id, mcr.cf->GetName().c_str(), s.getState()); if (!s.IsShutdownInProgress()) { rdb_handle_io_error(s, RDB_IO_ERROR_BG_THREAD); } else { DBUG_ASSERT(m_requests.size() == 1); } } rocksdb_manual_compactions_processed++; clear_manual_compaction_request(mcr.mc_id, false); RDB_MUTEX_LOCK_CHECK(m_signal_mutex); } clear_all_manual_compaction_requests(); DBUG_ASSERT(m_requests.empty()); RDB_MUTEX_UNLOCK_CHECK(m_signal_mutex); mysql_mutex_destroy(&m_mc_mutex); } void Rdb_manual_compaction_thread::clear_all_manual_compaction_requests() { RDB_MUTEX_LOCK_CHECK(m_mc_mutex); m_requests.clear(); RDB_MUTEX_UNLOCK_CHECK(m_mc_mutex); } void Rdb_manual_compaction_thread::clear_manual_compaction_request( int mc_id, bool init_only) { bool erase = true; RDB_MUTEX_LOCK_CHECK(m_mc_mutex); auto it = m_requests.find(mc_id); if (it != m_requests.end()) { if (init_only) { Manual_compaction_request mcr = it->second; if (mcr.state != Manual_compaction_request::INITED) { erase = false; } } if (erase) { m_requests.erase(it); } } else { // Current code path guarantees that erasing by the same mc_id happens // at most once. INITED state may be erased by a thread that requested // the compaction. RUNNING state is erased by mc thread only. DBUG_ASSERT(0); } RDB_MUTEX_UNLOCK_CHECK(m_mc_mutex); } int Rdb_manual_compaction_thread::request_manual_compaction( rocksdb::ColumnFamilyHandle *cf, rocksdb::Slice *start, rocksdb::Slice *limit, int concurrency) { int mc_id = -1; RDB_MUTEX_LOCK_CHECK(m_mc_mutex); if (m_requests.size() >= rocksdb_max_manual_compactions) { RDB_MUTEX_UNLOCK_CHECK(m_mc_mutex); return mc_id; } Manual_compaction_request mcr; mc_id = mcr.mc_id = ++m_latest_mc_id; mcr.state = Manual_compaction_request::INITED; mcr.cf = cf; mcr.start = start; mcr.limit = limit; mcr.concurrency = concurrency; m_requests.insert(std::make_pair(mcr.mc_id, mcr)); RDB_MUTEX_UNLOCK_CHECK(m_mc_mutex); return mc_id; } bool Rdb_manual_compaction_thread::is_manual_compaction_finished(int mc_id) { bool finished = false; RDB_MUTEX_LOCK_CHECK(m_mc_mutex); if (m_requests.count(mc_id) == 0) { finished = true; } RDB_MUTEX_UNLOCK_CHECK(m_mc_mutex); return finished; } bool ha_rocksdb::check_bloom_and_set_bounds( THD *thd, const Rdb_key_def &kd, const rocksdb::Slice &eq_cond, const bool use_all_keys, size_t bound_len, uchar *const lower_bound, uchar *const upper_bound, rocksdb::Slice *lower_bound_slice, rocksdb::Slice *upper_bound_slice) { bool can_use_bloom = can_use_bloom_filter(thd, kd, eq_cond, use_all_keys); if (!can_use_bloom) { setup_iterator_bounds(kd, eq_cond, bound_len, lower_bound, upper_bound, lower_bound_slice, upper_bound_slice); } return can_use_bloom; } /** Deciding if it is possible to use bloom filter or not. @detail Even if bloom filter exists, it is not always possible to use bloom filter. If using bloom filter when you shouldn't, false negative may happen -- fewer rows than expected may be returned. It is users' responsibility to use bloom filter correctly. If bloom filter does not exist, return value does not matter because RocksDB does not use bloom filter internally. @param kd @param eq_cond Equal condition part of the key. This always includes system index id (4 bytes). @param use_all_keys True if all key parts are set with equal conditions. This is aware of extended keys. */ bool ha_rocksdb::can_use_bloom_filter(THD *thd, const Rdb_key_def &kd, const rocksdb::Slice &eq_cond, const bool use_all_keys) { bool can_use = false; if (THDVAR(thd, skip_bloom_filter_on_read)) { return can_use; } const rocksdb::SliceTransform *prefix_extractor = kd.get_extractor(); if (prefix_extractor) { /* This is an optimized use case for CappedPrefixTransform. If eq_cond length >= prefix extractor length and if all keys are used for equal lookup, it is always possible to use bloom filter. Prefix bloom filter can't be used on descending scan with prefix lookup (i.e. WHERE id1=1 ORDER BY id2 DESC), because of RocksDB's limitation. On ascending (or not sorting) scan, keys longer than the capped prefix length will be truncated down to the capped length and the resulting key is added to the bloom filter. Keys shorter than the capped prefix length will be added to the bloom filter. When keys are looked up, key conditionals longer than the capped length can be used; key conditionals shorter require all parts of the key to be available for the short key match. */ if ((use_all_keys && prefix_extractor->InRange(eq_cond)) || prefix_extractor->SameResultWhenAppended(eq_cond)) can_use = true; else can_use = false; } else { /* if prefix extractor is not defined, all key parts have to be used by eq_cond. */ if (use_all_keys) can_use = true; else can_use = false; } return can_use; } /* For modules that need access to the global data structures */ rocksdb::TransactionDB *rdb_get_rocksdb_db() { return rdb; } Rdb_cf_manager &rdb_get_cf_manager() { return cf_manager; } const rocksdb::BlockBasedTableOptions &rdb_get_table_options() { return *rocksdb_tbl_options; } bool rdb_is_ttl_enabled() { return rocksdb_enable_ttl; } bool rdb_is_ttl_read_filtering_enabled() { return rocksdb_enable_ttl_read_filtering; } #ifndef NDEBUG int rdb_dbug_set_ttl_rec_ts() { return rocksdb_debug_ttl_rec_ts; } int rdb_dbug_set_ttl_snapshot_ts() { return rocksdb_debug_ttl_snapshot_ts; } int rdb_dbug_set_ttl_read_filter_ts() { return rocksdb_debug_ttl_read_filter_ts; } bool rdb_dbug_set_ttl_ignore_pk() { return rocksdb_debug_ttl_ignore_pk; } #endif void rdb_update_global_stats(const operation_type &type, uint count, bool is_system_table) { DBUG_ASSERT(type < ROWS_MAX); if (count == 0) { return; } if (is_system_table) { global_stats.system_rows[type].add(count); } else { global_stats.rows[type].add(count); } } int rdb_get_table_perf_counters(const char *const tablename, Rdb_perf_counters *const counters) { DBUG_ASSERT(tablename != nullptr); Rdb_table_handler *table_handler; table_handler = rdb_open_tables.get_table_handler(tablename); if (table_handler == nullptr) { return HA_ERR_ROCKSDB_INVALID_TABLE; } counters->load(table_handler->m_table_perf_context); rdb_open_tables.release_table_handler(table_handler); return HA_EXIT_SUCCESS; } const char *get_rdb_io_error_string(const RDB_IO_ERROR_TYPE err_type) { // If this assertion fails then this means that a member has been either added // to or removed from RDB_IO_ERROR_TYPE enum and this function needs to be // changed to return the appropriate value. static_assert(RDB_IO_ERROR_LAST == 4, "Please handle all the error types."); switch (err_type) { case RDB_IO_ERROR_TYPE::RDB_IO_ERROR_TX_COMMIT: return "RDB_IO_ERROR_TX_COMMIT"; case RDB_IO_ERROR_TYPE::RDB_IO_ERROR_DICT_COMMIT: return "RDB_IO_ERROR_DICT_COMMIT"; case RDB_IO_ERROR_TYPE::RDB_IO_ERROR_BG_THREAD: return "RDB_IO_ERROR_BG_THREAD"; case RDB_IO_ERROR_TYPE::RDB_IO_ERROR_GENERAL: return "RDB_IO_ERROR_GENERAL"; default: DBUG_ASSERT(false); return "(unknown)"; } } // In case of core dump generation we want this function NOT to be optimized // so that we can capture as much data as possible to debug the root cause // more efficiently. #ifdef __GNUC__ #endif void rdb_handle_io_error(const rocksdb::Status status, const RDB_IO_ERROR_TYPE err_type) { if (status.IsIOError()) { switch (err_type) { case RDB_IO_ERROR_TX_COMMIT: case RDB_IO_ERROR_DICT_COMMIT: { rdb_log_status_error(status, "failed to write to WAL"); /* NO_LINT_DEBUG */ sql_print_error("MyRocks: aborting on WAL write error."); abort(); break; } case RDB_IO_ERROR_BG_THREAD: { rdb_log_status_error(status, "BG thread failed to write to RocksDB"); /* NO_LINT_DEBUG */ sql_print_error("MyRocks: aborting on BG write error."); abort(); break; } case RDB_IO_ERROR_GENERAL: { rdb_log_status_error(status, "failed on I/O"); /* NO_LINT_DEBUG */ sql_print_error("MyRocks: aborting on I/O error."); abort(); break; } default: DBUG_ASSERT(0); break; } } else if (status.IsCorruption()) { rdb_log_status_error(status, "data corruption detected!"); rdb_persist_corruption_marker(); /* NO_LINT_DEBUG */ sql_print_error("MyRocks: aborting because of data corruption."); abort(); } else if (!status.ok()) { switch (err_type) { case RDB_IO_ERROR_DICT_COMMIT: { rdb_log_status_error(status, "Failed to write to WAL (dictionary)"); /* NO_LINT_DEBUG */ sql_print_error("MyRocks: aborting on WAL write error."); abort(); break; } default: rdb_log_status_error(status, "Failed to read/write in RocksDB"); break; } } } #ifdef __GNUC__ #endif Rdb_dict_manager *rdb_get_dict_manager(void) { return &dict_manager; } Rdb_ddl_manager *rdb_get_ddl_manager(void) { return &ddl_manager; } Rdb_binlog_manager *rdb_get_binlog_manager(void) { return &binlog_manager; } void rocksdb_set_compaction_options( my_core::THD *const thd MY_ATTRIBUTE((__unused__)), my_core::st_mysql_sys_var *const var MY_ATTRIBUTE((__unused__)), void *const var_ptr, const void *const save) { if (var_ptr && save) { *(uint64_t *)var_ptr = *(const uint64_t *)save; } const Rdb_compact_params params = { (uint64_t)rocksdb_compaction_sequential_deletes, (uint64_t)rocksdb_compaction_sequential_deletes_window, (uint64_t)rocksdb_compaction_sequential_deletes_file_size}; if (properties_collector_factory) { properties_collector_factory->SetCompactionParams(params); } } void rocksdb_set_table_stats_sampling_pct( my_core::THD *const thd MY_ATTRIBUTE((__unused__)), my_core::st_mysql_sys_var *const var MY_ATTRIBUTE((__unused__)), void *const var_ptr MY_ATTRIBUTE((__unused__)), const void *const save) { RDB_MUTEX_LOCK_CHECK(rdb_sysvars_mutex); const uint32_t new_val = *static_cast(save); if (new_val != rocksdb_table_stats_sampling_pct) { rocksdb_table_stats_sampling_pct = new_val; if (properties_collector_factory) { properties_collector_factory->SetTableStatsSamplingPct( rocksdb_table_stats_sampling_pct); } } RDB_MUTEX_UNLOCK_CHECK(rdb_sysvars_mutex); } /* This function allows setting the rate limiter's bytes per second value but only if the rate limiter is turned on which has to be done at startup. If the rate is already 0 (turned off) or we are changing it to 0 (trying to turn it off) this function will push a warning to the client and do nothing. This is similar to the code in innodb_doublewrite_update (found in storage/innobase/handler/ha_innodb.cc). */ void rocksdb_set_rate_limiter_bytes_per_sec( my_core::THD *const thd, my_core::st_mysql_sys_var *const var MY_ATTRIBUTE((__unused__)), void *const var_ptr MY_ATTRIBUTE((__unused__)), const void *const save) { const uint64_t new_val = *static_cast(save); if (new_val == 0 || rocksdb_rate_limiter_bytes_per_sec == 0) { /* If a rate_limiter was not enabled at startup we can't change it nor can we disable it if one was created at startup */ push_warning_printf(thd, Sql_condition::WARN_LEVEL_WARN, ER_WRONG_ARGUMENTS, "RocksDB: rocksdb_rate_limiter_bytes_per_sec cannot " "be dynamically changed to or from 0. Do a clean " "shutdown if you want to change it from or to 0."); } else if (new_val != rocksdb_rate_limiter_bytes_per_sec) { /* Apply the new value to the rate limiter and store it locally */ DBUG_ASSERT(rocksdb_rate_limiter != nullptr); rocksdb_rate_limiter_bytes_per_sec = new_val; rocksdb_rate_limiter->SetBytesPerSecond(new_val); } } void rocksdb_set_sst_mgr_rate_bytes_per_sec( my_core::THD *const thd, my_core::st_mysql_sys_var *const var MY_ATTRIBUTE((__unused__)), void *const var_ptr MY_ATTRIBUTE((__unused__)), const void *const save) { RDB_MUTEX_LOCK_CHECK(rdb_sysvars_mutex); const uint64_t new_val = *static_cast(save); if (new_val != rocksdb_sst_mgr_rate_bytes_per_sec) { rocksdb_sst_mgr_rate_bytes_per_sec = new_val; rocksdb_db_options->sst_file_manager->SetDeleteRateBytesPerSecond( rocksdb_sst_mgr_rate_bytes_per_sec); } RDB_MUTEX_UNLOCK_CHECK(rdb_sysvars_mutex); } void rocksdb_set_delayed_write_rate(THD *thd, struct st_mysql_sys_var *var, void *var_ptr, const void *save) { RDB_MUTEX_LOCK_CHECK(rdb_sysvars_mutex); const uint64_t new_val = *static_cast(save); if (rocksdb_delayed_write_rate != new_val) { rocksdb_delayed_write_rate = new_val; rocksdb::Status s = rdb->SetDBOptions({{"delayed_write_rate", std::to_string(new_val)}}); if (!s.ok()) { /* NO_LINT_DEBUG */ sql_print_warning("MyRocks: failed to update delayed_write_rate. " "status code = %d, status = %s", s.code(), s.ToString().c_str()); } } RDB_MUTEX_UNLOCK_CHECK(rdb_sysvars_mutex); } void rocksdb_set_max_latest_deadlocks(THD *thd, struct st_mysql_sys_var *var, void *var_ptr, const void *save) { RDB_MUTEX_LOCK_CHECK(rdb_sysvars_mutex); const uint32_t new_val = *static_cast(save); if (rocksdb_max_latest_deadlocks != new_val) { rocksdb_max_latest_deadlocks = new_val; rdb->SetDeadlockInfoBufferSize(rocksdb_max_latest_deadlocks); } RDB_MUTEX_UNLOCK_CHECK(rdb_sysvars_mutex); } void rdb_set_collation_exception_list(const char *const exception_list) { DBUG_ASSERT(rdb_collation_exceptions != nullptr); if (!rdb_collation_exceptions->set_patterns(exception_list)) { my_core::warn_about_bad_patterns(rdb_collation_exceptions, "strict_collation_exceptions"); } } void rocksdb_set_collation_exception_list(THD *const thd, struct st_mysql_sys_var *const var, void *const var_ptr, const void *const save) { const char *const val = *static_cast(save); rdb_set_collation_exception_list(val == nullptr ? "" : val); //psergey-todo: what is the purpose of the below?? const char *val_copy= val? my_strdup(val, MYF(0)): nullptr; my_free(*static_cast(var_ptr)); *static_cast(var_ptr) = val_copy; } int mysql_value_to_bool(struct st_mysql_value *value, my_bool *return_value) { int new_value_type = value->value_type(value); if (new_value_type == MYSQL_VALUE_TYPE_STRING) { char buf[16]; int len = sizeof(buf); const char *str = value->val_str(value, buf, &len); if (str && (my_strcasecmp(system_charset_info, "true", str) == 0 || my_strcasecmp(system_charset_info, "on", str) == 0)) { *return_value = TRUE; } else if (str && (my_strcasecmp(system_charset_info, "false", str) == 0 || my_strcasecmp(system_charset_info, "off", str) == 0)) { *return_value = FALSE; } else { return 1; } } else if (new_value_type == MYSQL_VALUE_TYPE_INT) { long long intbuf; value->val_int(value, &intbuf); if (intbuf > 1) return 1; *return_value = intbuf > 0 ? TRUE : FALSE; } else { return 1; } return 0; } int rocksdb_check_bulk_load( THD *const thd, struct st_mysql_sys_var *var MY_ATTRIBUTE((__unused__)), void *save, struct st_mysql_value *value) { my_bool new_value; if (mysql_value_to_bool(value, &new_value) != 0) { return 1; } Rdb_transaction *&tx = get_tx_from_thd(thd); if (tx != nullptr) { const int rc = tx->finish_bulk_load(); if (rc != 0) { // NO_LINT_DEBUG sql_print_error("RocksDB: Error %d finalizing last SST file while " "setting bulk loading variable", rc); THDVAR(thd, bulk_load) = 0; return 1; } } *static_cast(save) = new_value; return 0; } int rocksdb_check_bulk_load_allow_unsorted( THD *const thd, struct st_mysql_sys_var *var MY_ATTRIBUTE((__unused__)), void *save, struct st_mysql_value *value) { my_bool new_value; if (mysql_value_to_bool(value, &new_value) != 0) { return 1; } if (THDVAR(thd, bulk_load)) { my_error(ER_ERROR_WHEN_EXECUTING_COMMAND, MYF(0), "SET", "Cannot change this setting while bulk load is enabled"); return 1; } *static_cast(save) = new_value; return 0; } static void rocksdb_set_max_background_jobs(THD *thd, struct st_mysql_sys_var *const var, void *const var_ptr, const void *const save) { DBUG_ASSERT(save != nullptr); DBUG_ASSERT(rocksdb_db_options != nullptr); DBUG_ASSERT(rocksdb_db_options->env != nullptr); RDB_MUTEX_LOCK_CHECK(rdb_sysvars_mutex); const int new_val = *static_cast(save); if (rocksdb_db_options->max_background_jobs != new_val) { rocksdb_db_options->max_background_jobs = new_val; rocksdb::Status s = rdb->SetDBOptions({{"max_background_jobs", std::to_string(new_val)}}); if (!s.ok()) { /* NO_LINT_DEBUG */ sql_print_warning("MyRocks: failed to update max_background_jobs. " "Status code = %d, status = %s.", s.code(), s.ToString().c_str()); } } RDB_MUTEX_UNLOCK_CHECK(rdb_sysvars_mutex); } static void rocksdb_set_bytes_per_sync( THD *thd MY_ATTRIBUTE((__unused__)), struct st_mysql_sys_var *const var MY_ATTRIBUTE((__unused__)), void *const var_ptr MY_ATTRIBUTE((__unused__)), const void *const save) { DBUG_ASSERT(save != nullptr); DBUG_ASSERT(rocksdb_db_options != nullptr); DBUG_ASSERT(rocksdb_db_options->env != nullptr); RDB_MUTEX_LOCK_CHECK(rdb_sysvars_mutex); const ulonglong new_val = *static_cast(save); if (rocksdb_db_options->bytes_per_sync != new_val) { rocksdb_db_options->bytes_per_sync = new_val; rocksdb::Status s = rdb->SetDBOptions({{"bytes_per_sync", std::to_string(new_val)}}); if (!s.ok()) { /* NO_LINT_DEBUG */ sql_print_warning("MyRocks: failed to update max_background_jobs. " "Status code = %d, status = %s.", s.code(), s.ToString().c_str()); } } RDB_MUTEX_UNLOCK_CHECK(rdb_sysvars_mutex); } static void rocksdb_set_wal_bytes_per_sync( THD *thd MY_ATTRIBUTE((__unused__)), struct st_mysql_sys_var *const var MY_ATTRIBUTE((__unused__)), void *const var_ptr MY_ATTRIBUTE((__unused__)), const void *const save) { DBUG_ASSERT(save != nullptr); DBUG_ASSERT(rocksdb_db_options != nullptr); DBUG_ASSERT(rocksdb_db_options->env != nullptr); RDB_MUTEX_LOCK_CHECK(rdb_sysvars_mutex); const ulonglong new_val = *static_cast(save); if (rocksdb_db_options->wal_bytes_per_sync != new_val) { rocksdb_db_options->wal_bytes_per_sync = new_val; rocksdb::Status s = rdb->SetDBOptions({{"wal_bytes_per_sync", std::to_string(new_val)}}); if (!s.ok()) { /* NO_LINT_DEBUG */ sql_print_warning("MyRocks: failed to update max_background_jobs. " "Status code = %d, status = %s.", s.code(), s.ToString().c_str()); } } RDB_MUTEX_UNLOCK_CHECK(rdb_sysvars_mutex); } /* Validating and updating block cache size via sys_var::check path. SetCapacity may take seconds when reducing block cache, and sys_var::update holds LOCK_global_system_variables mutex, so updating block cache size is done at check path instead. */ static int rocksdb_validate_set_block_cache_size( THD *thd MY_ATTRIBUTE((__unused__)), struct st_mysql_sys_var *const var MY_ATTRIBUTE((__unused__)), void *var_ptr, struct st_mysql_value *value) { DBUG_ASSERT(value != nullptr); long long new_value; /* value is NULL */ if (value->val_int(value, &new_value)) { return HA_EXIT_FAILURE; } if (new_value < RDB_MIN_BLOCK_CACHE_SIZE || (uint64_t)new_value > (uint64_t)LONGLONG_MAX) { return HA_EXIT_FAILURE; } RDB_MUTEX_LOCK_CHECK(rdb_block_cache_resize_mutex); const rocksdb::BlockBasedTableOptions &table_options = rdb_get_table_options(); if (rocksdb_block_cache_size != new_value && table_options.block_cache) { table_options.block_cache->SetCapacity(new_value); } *static_cast(var_ptr) = static_cast(new_value); RDB_MUTEX_UNLOCK_CHECK(rdb_block_cache_resize_mutex); return HA_EXIT_SUCCESS; } static int rocksdb_validate_update_cf_options(THD * /* unused */, struct st_mysql_sys_var * /*unused*/, void *save, struct st_mysql_value *value) { char buff[STRING_BUFFER_USUAL_SIZE]; const char *str; int length; length = sizeof(buff); str = value->val_str(value, buff, &length); *(const char **)save = str; if (str == nullptr) { return HA_EXIT_SUCCESS; } Rdb_cf_options::Name_to_config_t option_map; // Basic sanity checking and parsing the options into a map. If this fails // then there's no point to proceed. if (!Rdb_cf_options::parse_cf_options(str, &option_map)) { my_error(ER_WRONG_VALUE_FOR_VAR, MYF(0), "rocksdb_update_cf_options", str); return HA_EXIT_FAILURE; } return HA_EXIT_SUCCESS; } static void rocksdb_set_update_cf_options(THD *const /* unused */, struct st_mysql_sys_var *const /* unused */, void *const var_ptr, const void *const save) { const char *const val = *static_cast(save); RDB_MUTEX_LOCK_CHECK(rdb_sysvars_mutex); my_free(*reinterpret_cast(var_ptr)); if (!val) { *reinterpret_cast(var_ptr) = nullptr; RDB_MUTEX_UNLOCK_CHECK(rdb_sysvars_mutex); return; } DBUG_ASSERT(val != nullptr); // Reset the pointers regardless of how much success we had with updating // the CF options. This will results in consistent behavior and avoids // dealing with cases when only a subset of CF-s was successfully updated. *reinterpret_cast(var_ptr) = my_strdup(val, MYF(0)); // Do the real work of applying the changes. Rdb_cf_options::Name_to_config_t option_map; // This should never fail, because of rocksdb_validate_update_cf_options if (!Rdb_cf_options::parse_cf_options(val, &option_map)) { my_free(*reinterpret_cast(var_ptr)); RDB_MUTEX_UNLOCK_CHECK(rdb_sysvars_mutex); return; } // For each CF we have, see if we need to update any settings. for (const auto &cf_name : cf_manager.get_cf_names()) { DBUG_ASSERT(!cf_name.empty()); rocksdb::ColumnFamilyHandle *cfh = cf_manager.get_cf(cf_name); DBUG_ASSERT(cfh != nullptr); const auto it = option_map.find(cf_name); std::string per_cf_options = (it != option_map.end()) ? it->second : ""; if (!per_cf_options.empty()) { Rdb_cf_options::Name_to_config_t opt_map; rocksdb::Status s = rocksdb::StringToMap(per_cf_options, &opt_map); if (s != rocksdb::Status::OK()) { // NO_LINT_DEBUG sql_print_warning("MyRocks: failed to convert the options for column " "family '%s' to a map. %s", cf_name.c_str(), s.ToString().c_str()); } else { DBUG_ASSERT(rdb != nullptr); // Finally we can apply the options. s = rdb->SetOptions(cfh, opt_map); if (s != rocksdb::Status::OK()) { // NO_LINT_DEBUG sql_print_warning("MyRocks: failed to apply the options for column " "family '%s'. %s", cf_name.c_str(), s.ToString().c_str()); } else { // NO_LINT_DEBUG sql_print_information("MyRocks: options for column family '%s' " "have been successfully updated.", cf_name.c_str()); // Make sure that data is internally consistent as well and update // the CF options. This is necessary also to make sure that the CF // options will be correctly reflected in the relevant table: // ROCKSDB_CF_OPTIONS in INFORMATION_SCHEMA. rocksdb::ColumnFamilyOptions cf_options = rdb->GetOptions(cfh); std::string updated_options; s = rocksdb::GetStringFromColumnFamilyOptions(&updated_options, cf_options); DBUG_ASSERT(s == rocksdb::Status::OK()); DBUG_ASSERT(!updated_options.empty()); cf_manager.update_options_map(cf_name, updated_options); } } } } // Our caller (`plugin_var_memalloc_global_update`) will call `my_free` to // free up resources used before. RDB_MUTEX_UNLOCK_CHECK(rdb_sysvars_mutex); } void rdb_queue_save_stats_request() { rdb_bg_thread.request_save_stats(); } #ifdef MARIAROCKS_NOT_YET // MDEV-10976 void ha_rocksdb::rpl_before_delete_rows() { DBUG_ENTER_FUNC(); m_in_rpl_delete_rows = true; DBUG_VOID_RETURN; } void ha_rocksdb::rpl_after_delete_rows() { DBUG_ENTER_FUNC(); m_in_rpl_delete_rows = false; DBUG_VOID_RETURN; } void ha_rocksdb::rpl_before_update_rows() { DBUG_ENTER_FUNC(); m_in_rpl_update_rows = true; DBUG_VOID_RETURN; } void ha_rocksdb::rpl_after_update_rows() { DBUG_ENTER_FUNC(); m_in_rpl_update_rows = false; DBUG_VOID_RETURN; } /** @brief Read Free Replication can be used or not. Returning False means Read Free Replication can be used. Read Free Replication can be used on UPDATE or DELETE row events, and table must have user defined primary key. */ bool ha_rocksdb::use_read_free_rpl() { DBUG_ENTER_FUNC(); DBUG_RETURN((m_in_rpl_delete_rows || m_in_rpl_update_rows) && !has_hidden_pk(table) && m_use_read_free_rpl); } #endif // MARIAROCKS_NOT_YET double ha_rocksdb::read_time(uint index, uint ranges, ha_rows rows) { DBUG_ENTER_FUNC(); if (index != table->s->primary_key) { /* Non covering index range scan */ DBUG_RETURN(handler::read_time(index, ranges, rows)); } DBUG_RETURN((rows / 20.0) + 1); } void ha_rocksdb::print_error(int error, myf errflag) { if (error == HA_ERR_ROCKSDB_STATUS_BUSY) { error = HA_ERR_LOCK_DEADLOCK; } handler::print_error(error, errflag); } std::string rdb_corruption_marker_file_name() { std::string ret(rocksdb_datadir); ret.append("/ROCKSDB_CORRUPTED"); return ret; } void sql_print_verbose_info(const char *format, ...) { va_list args; if (global_system_variables.log_warnings > 2) { va_start(args, format); sql_print_information_v(format, args); va_end(args); } } } // namespace myrocks /** Construct and emit duplicate key error message using information from table's record buffer. @sa print_keydup_error(table, key, msg, errflag, thd, org_table_name). */ void print_keydup_error(TABLE *table, KEY *key, myf errflag, const THD *thd, const char *org_table_name) { print_keydup_error(table, key, ER(ER_DUP_ENTRY_WITH_KEY_NAME), errflag); } /* Register the storage engine plugin outside of myrocks namespace so that mysql_declare_plugin does not get confused when it does its name generation. */ struct st_mysql_storage_engine rocksdb_storage_engine = { MYSQL_HANDLERTON_INTERFACE_VERSION}; maria_declare_plugin(rocksdb_se){ MYSQL_STORAGE_ENGINE_PLUGIN, /* Plugin Type */ &rocksdb_storage_engine, /* Plugin Descriptor */ "ROCKSDB", /* Plugin Name */ "Monty Program Ab", /* Plugin Author */ "RocksDB storage engine", /* Plugin Description */ PLUGIN_LICENSE_GPL, /* Plugin Licence */ myrocks::rocksdb_init_func, /* Plugin Entry Point */ myrocks::rocksdb_done_func, /* Plugin Deinitializer */ 0x0001, /* version number (0.1) */ myrocks::rocksdb_status_vars, /* status variables */ myrocks::rocksdb_system_variables, /* system variables */ "1.0", /* string version */ myrocks::MYROCKS_MARIADB_PLUGIN_MATURITY_LEVEL }, myrocks::rdb_i_s_cfstats, myrocks::rdb_i_s_dbstats, myrocks::rdb_i_s_perf_context, myrocks::rdb_i_s_perf_context_global, myrocks::rdb_i_s_cfoptions, myrocks::rdb_i_s_compact_stats, myrocks::rdb_i_s_global_info, myrocks::rdb_i_s_ddl, myrocks::rdb_i_s_sst_props, myrocks::rdb_i_s_index_file_map, myrocks::rdb_i_s_lock_info, myrocks::rdb_i_s_trx_info, myrocks::rdb_i_s_deadlock_info maria_declare_plugin_end;