Sync

What's new in this version:

Upgrading from MySQL:
- MariaDB now detects and converts previously incompatible MySQL partition schemes

General:
- As per the MariaDB Deprecation Policy, this will be the last release of MariaDB 11.0 for Ubuntu 18.04 LTS "Bionic" and Ubuntu 22.10 "Kinetic"
- In this release repositories for Debian 12 "Bookworm" have been added
- mariadb-dump --force doesn't ignore error as it should
- 280 Bytes lost in mysys/array.c, mysys/hash.c, sql/sp.cc, sql/sp.cc, sql/item_create.cc, sql/item_create.cc, sql/sql_yacc.yy:10748 when using oracle sql_mode
- SQL/PL package body does not appear in I_S.ROUTINES.ROUTINE_DEFINITION
- Unexpected result when combining DISTINCT, subselect and LIMIT
- ROW variables do not get assigned from subselects
- Character Sets, Data Types:
- UBSAN: null pointer passed as argument 1, which is declared to never be null in my_strnncoll_binary on SELECT ... COUNT or GROUP_CONCAT
- Possibly wrong result or Assertion `0' failed in Item_func_round::native_op
- Assertion `(length % 4) == 0' failed in my_lengthsp_utf32 on SELECT
- UBSAN: negation of -X cannot be represented in type 'long long int'; cast to an unsigned type to negate this value to itself in Item_func_mul::int_op and Item_func_round::int_op

InnoDB:
- Crashing on I/O error is unhelpful
- SIGSEGV in log_sort_flush_list() in InnoDB crash recovery
- InnoDB tables are being flagged as corrupted on an I/O bound server
- Duplicate entry allowed into a UNIQUE column
- Server Status Innodb_row_lock_time% is reported in seconds
- innochecksum dies with Floating point exception
- Add InnoDB engine information to the slow query log
- Deadlock with 3 concurrent DELETEs by unique key
- innodb protection against dual processes accessing data insufficient
- ER_DUP_KEY in mysql.innodb_table_stats upon RENAME on sequence
- Assertion `!strcmp(index->table->name.m_name, "SYS_FOREIGN") || !strcmp(index->table->name.m_name, "SYS_FOREIGN_COLS")' failed in btr_node_ptr_max_size
- InnoDB: Failing assertion: page_type == i_s_page_type[page_type].type_value
- btr_estimate_n_rows_in_range() accesses unfixed, unlatched page
- MODIFY COLUMN can break FK constraints, and lead to unrestorable dumps
- Recovery or backup failure after innodb_undo_log_truncate=ON
- Assertion 'n & PENDING' failed in fil_space_t::set_needs_flush()
- fil_node_open_file() releases fil_system.mutex allowing other thread to open its file node
- ASAN errors in dict_v_col_t::detach upon adding key to virtual column
- Purge trying to access freed secondary index page
- Freed data pages are not always being scrubbed
- InnoDB recovery hangs after reporting corruption
- !cursor->index->is_committed() in row0ins.cc after update to 10.4.13 from 10.3.21
- innodb_undo_log_truncate=ON fails to wait for purge of enough transaction history
- SET GLOBAL innodb_undo_log_truncate=ON does not free space when no undo logs exist
- innodb_read_ahead_threshold (linear read-ahead) does not work
- InnoDB recovery and mariadb-backup --prepare fail to report detailed progress
- fil_ibd_create() may hijack the file handle of an old file
- innodb_undo_log_truncate=ON recovery results in a corrupted undo log
- Server freeze due to innodb_change_buffering and innodb_file_per_table=0
- Change buffer entries are left behind when freeing a page, causing secondary index corruption when the page is later reused

Aria:
- Various crashes upon INSERT/UPDATE after changing Aria settings
- Various crashes/asserts/corruptions when Aria encryption is enabled/used, but the encryption plugin is not loaded

Spider:
- SIGSEGV in spider_db_open_item_field and SIGSEGV in spider_db_print_item_type, on SELECT
- Spider variables that double as table params overriding mechanism is buggy

Optimizer:
- Assertion `last_key_entry >= end_pos' failed in virtual bool JOIN_CACHE_HASHED::put_record()
- Problem with open ranges on prefix blobs keys
- Equal on two RANK window functions create wrong result
- Recursive CTE execution is interrupted without errors or warnings
- Assertion `s->table->opt_range_condition_rows <= s->found_records' failed in apply_selectivity_for_table
- Inconsistency between MRR and SQL layer costs can cause poor query plan
- MAX_SEL_ARG memory exhaustion is not visible in the optimizer trace
- SHOW TABLES not working properly with lower_case_table_names=2
- Segfault on select query using index for group-by and filesort

Replication:
- Parallel Slave SQL Thread Can Update Seconds_Behind_Master with Active Workers
- ALTER SEQUENCE ends up in optimistic parallel slave binlog out-of-order
- STOP SLAVE takes very long time on a busy system
- On slave XA COMMIT/XA ROLLBACK fail to return an error in read-only mode
- Calling a function from a different database in a slave side trigger crashes

Galera:
- Node never returns from Donor/Desynced to Synced when wsrep_mode = BF_ABORT_MARIABACKUP
- Node has been dropped from the cluster on Startup / Shutdown with async replica
- KILL QUERY maintains nodes data consistency but breaks GTID sequence
- Assertion failure !lock.was_chosen_as_deadlock_victim in trx0trx.h:1065
- MariaDB stuck on starting commit state (waiting on commit order critical section)
- Assertion state() == s_aborting || state() == s_must_replay failed in int wsrep::transaction::after_rollback()
- Assertion !wsrep_has_changes(thd) || (thd->lex->sql_command == SQLCOM_CREATE_TABLE && !thd->is_current_stmt_binlog_format_row()) || thd->wsrep_cs().transaction().state() == wsrep::transaction::s_aborted failed
- Server crashes when wsrep_sst_donor and wsrep_cluster_address set to NULL
- Create temporary sequence can cause inconsistency
- Galera 4 unable to query cluster state if not primary component