MariaDB 11.2.2

What's new in this version:

InnoDB:
- DROP INDEX followed by CREATE INDEX may corrupt data
- ROW_FORMAT=COMPRESSED table: InnoDB: 2048 bytes should have been read. Only 0 bytes read.
- Server aborts during alter operation when table doesn't have foreign index
- row_merge_fts_doc_tokenize() handles FTS plugin parser inconsistently
- InnoDB: tried to purge non-delete-marked record of an index on a virtual column prefix
- lock_row_lock_current_waits counter in information_schema.innodb_metrics may become negative
- SET GLOBAL innodb_max_purge_lag_wait=… hangs if innodb_read_only=ON
- Auto-increment no longer works for explicit FTS_DOC_ID
- Assertion `pos < table->n_def' failed in dict_table_get_nth_col
- innochecksum man pages seem to be inconsistent with the binary (10.2.25)
- innodb_compression_algorithm=0 (none) increments Innodb_num_pages_page_compression_error
- wrong table name in innodb's "row too big" errors
- Optimize is_file_on_ssd() to speedup opening tablespaces on Windows
- Race condition between page write completion and log checkpoint
- After crash recovery, Checksum mismatch + Failing assertion: !i || prev_id + 1 == space_id,
- Deadlock due to log_free_check(), involving trx_purge_truncate_rseg_history() and trx_undo_assign_low()
- Write-ahead logging is broken for freed pages
- X-lock on supremum for prepared transaction for RR
- Assertion failures in log_sort_flush_list upon crash recovery
- Assertion `purge_sys.tail.trx_no <= purge_sys.rseg->last_trx_no()'
- Assertion `index->is_btree() || index->is_ibuf()' failed in btr_search_guess_on_hash
- InnoDB hang in buf_flush_wait_LRU_batch_end()
- InnoDB may hang when running out of buffer pool
- Corrupt index(es) on busy table when using FOREIGN KEY
- InnoDB Recovery doesn't display encryption message when no encryption configuration passed
- Wrong information about innodb_checksum_algorithm in the information_schema.SYSTEM_VARIABLES
- InnoDB may fail to recover after being killed in fil_delete_tablespace()
- Create separate tpool thread for async aio
- UNDO logs still growing for write-intensive workloads
- Increase the default for innodb_purge_batch_size from 300 to 1000
- Deprecate innodb_purge_rseg_truncate_frequency.
- The motivation for introducing this in MySQL seems to have been to avoid stalls due to freeing undo log pages or truncating undo log tablespaces. In MariaDB, innodb_undo_log_truncate=ON should be a much lighter operation because it will not involve any log checkpoint.
- Slow full index scan in 10.6 vs 10.5 for the (slow) I/O-bound case
- LOAD DATA into InnoDB w/partitions: huge performance loss, affected 10.6+
- Disable read-ahead for temporary tablespace
- Wrong result of: WHERE inet6_column IN ('','::1')

Optimizer:
- Crash when HAVING in a correlated subquery references columns in the outer query
- Server crashes at TABLE::add_tmp_key
- Server crashes inside filesort at my_decimal::to_binary
- Assertion `bitmap_is_set(&m_part_info->read_partitions, m_part_spec.start_part)' failed in ha_partition::handle_ordered_index_scan
- Crash when searching for the best split of derived table
- Test case from opt_tvc.test fails with statement memory protection
- Significant slowdown for query with many outer joins
- test_if_skip_sort_order() should catch the join types JT_EQ_REF, JT_CONST and JT_SYSTEM and skip sort order for these
- jointable materialization subquery optimization ignoring errors, then failing ASSERT.
- Server crashes in JOIN::cleanup after erroneous query with view
- Prepared statement return wrong result (missing row)
- Assertion `range->rows >= s->found_records' failed in best_access_path
- Raise notes when an index cannot be used on data type mismatch
- Replication:
- rpl.rpl_parallel_temptable failure due to incorrect commit optimization of temptables
- Lock wait timeout with INSERT-SELECT, autoinc, and statement-based replication
- strings/ctype-ucs2.c:2336: my_vsnprintf_utf32: Assertion `(n % 4) == 0' failed in my_vsnprintf_utf32 on INSERT
- Assertion fails in MDL_context::acquire_lock upon parallel replication of CREATE SEQUENCE
- SHOW SLAVE STATUS Last_SQL_Errno Race Condition on Errored Slave Restart
- seconds_behind_master is inaccurate for Delayed replication
- detailize the semisync replication magic number error
- Parallel replication deadlock victim preference code errorneously removed
- Parallel replication lags because innobase_kill_query() may fail to interrupt a lock wait
- Missed kill when the SQL driver thread goes to wait for parallel slave worker queues to drain

Galera:
- Assertion `state() == s_executing || state() == s_prepared || state() == s_committing || state() == s_must_abort || state() == s_replaying' failed.
- Assertion `state() == s_executing || state() == s_preparing || state() == s_prepared || state() == s_must_abort || state() == s_aborting || state() == s_cert_failed || state() == s_must_replay' failed
- wsrep_sst_mariabackup not working on FreeBSD
- Galera library 26.4.16 fails with every server version
- Galera node remains paused after interleaving FTWRLs
- Failed to insert streaming client
- When set at runtime, wsrep_sst_method accepts any value
- galera needs packaging script changes to successfully build
- replication breaks when using optimistic replication and replica is a galera node
- McAfee database vulnerability scan caused MariaDB crash with signal 6 (system abort)
- Assertion `mode_ == m_local || transaction_.is_streaming()' failed in int wsrep::client_state::bf_abort(wsrep::seqno)
- Data Definition:
- MariaDB crash on calling function
- ASAN errors in grn_obj_unlink / ha_mroonga::clear_indexes upon index operations
- vcol circular references lead to stack overflow
- OPTIMIZE TABLE crash
- Server crashes in Alter_info::add_stat_drop_index upon CREATE TABLE
- Scripts and Clients:
- mariadb-binlog -T/--table (mysqlbinlog) option
- mariadb-admin (mysqladmin) wrong error with simple_password_check
- mariadb-install-db shows warning on missing directory $pamtooldir/auth_pam_tool_dir
- main.mysql_client_test, main.mysql_client_test_comp failed on ASAN build with error: 5888, status: 23, errno: 2
- mariadb-install-db (mysql_install_db) doesn't properly grant proxy privileges to all default root user accounts

MariaBackup:
- MariaBackup full backup failed with InnoDB: Failing assertion: success in storage/innobase/fil/fil0fil.cc line 657
- mbstream breaks page compression on XFS
- Character Sets, Data Types, Collations:
- Prefix keys for CHAR work differently for MyISAM vs InnoDB
- Inconsistent results of DISTINCT with NOPAD
- Assertion `(length % 4) == 0' failed in my_lengthsp_utf32 on INSERT
- Compressed varchar values lost on joins when sorting on columns from joined table(s)
- UBSAN shift exponent X is too large for 64-bit type 'long long int' in sql/field.cc
- Wrong bit encoding using COALESCE
- utf8mb3_key_col=utf8mb4_value cannot be used for ref access

Spider:
- Spider UBSAN runtime error: applying non-zero offset x to null pointer in st_spider_param_string_parse::restore_delims
- Segfault when setting spider_delete_all_rows to 0 and delete all rows of a spider table, ASAN heap-use-after-free in spider_db_delete_all_rows
- ASAN errors in spider_fields::free_conn_holder or spider_create_group_by_handler
- ASAN: heap-buffer-overflow & stack-buffer-overflow in spider_db_mbase_row::append_to_str | SIGSEGV's in memmove_avx_unaligned_erms from memcpy in Binary_string::q_append, in Static_binary_string::q_append and my_strntoull10rnd_8bit | Unknown error 12801

General:
- binlog_do_db option breaks importing sql dumps
- Crashes in MDL_key::mdl_key_init with lower-case-table-names=2
- getting error 'Illegal parameter data types row and bigint for operation '+' ' when using ITERATE in a FOR..DO
- Assertion `arena_for_set_stmt== 0' failed in LEX::set_arena_for_set_stmt upon SET STATEMENT
- main.mysqlcheck fails on ARM with ASAN use-after-poison in my_mb_wc_filename
- Assertion failed: !pfs->m_idle || (state == PSI_SOCKET_STATE_ACTIVE)
- MyISAM wrong server status flags
- Server crashes in check_sequence_fields upon CREATE TABLE .. SEQUENCE=1 AS SELECT ..
- slow log Rows_examined out of range
- "rpm --setugids" breaks PAM authentication
- incorrect examined rows in case of stored function usage
- Compilation failing on MacOS (unknown warning option -Wno-unused-but-set-variable)
- Server crash upon inserting into Mroonga table with compressed column
- hash unique corrupts index on virtual blobs
- insert into an empty table fails with hash unique
- Valgrind/MSAN warnings in dynamic_column_update_move_left
- Memory leak showed in MDEV-6146 test suite
- Test from subselect.test fails with statement memory protection
- Memory leak when executing PS for query with IN subquery
- Allow the setting of Auto_increment on FK referenced columns
- mariadb-upgrade fails with sql_safe_updates = on
- Assertion `!(thd->server_status & (1U | 8192U))' failed in MDL_context::release_transactional_locks
- Information schema leaks table names and structure to unauthorized users
- Missing CHACHA20-POLY1305 support in WolfSSL
- incorrect error about cyclic reference about JSON type virtual column
- Disable TLS v1.0 and 1.1 for MariaDB
- Better indication of refusing to start because of ProtectHome
- Database upgrade fails: slow_log table
- myrocks_hotbackup.1 and test suite files installed when engine is disabled
- client_ed25519.dll isn't inluded for HeidiSQL.
- Assertion `!m_null_value' failed in int FixedBinTypeBundle<FbtImpl>::cmp_item_fbt::compare or in cmp_item_inet6::compare
- LeakSanitizer errors in get_quick_select or Assertion `status_var.local_memory_used == 0 || !debug_assert_on_not_freed_memory' failed
- Update signal hander user info more compassion and correct url
- maria-install-db fails on MacOS
- MSAN / Valgrind errors in Item_func_like::get_mm_leaf upon query from partitioned table
- MSAN / Valgrind errors in Item_func_like::get_mm_leaf with temporal field
- ASAN errors in base_list_iterator::next / setup_table_map upon 2nd execution of PS
- safe_mutex: Found wrong usage of mutex 'LOCK_thd_data' and 'wait_mutex'
- Docker Official Images:
- Invert single and double quotes for sql command definitions in healthcheck.sh due to failure under sql_mode=ANSI_QUOTES - contribution by Dominik Häckel
- healthcheck.sh --no-defaults behaviour was corrected - reported by Dominik Häckel
- Added /docker-entrypoint-init.d for tar{,compression} from mariadb-backup - instructions
- Refactor docker_mariadb_init in the entrypoint for extending the MariaDB image
- CIS failure due to world-writable directory /var/run/mysqld, added sticky bit - reported by @ollie1
- Add PROXY privileges for root@MARIADB_ROOT_HOST - reported by Matthieu Gusmini
- healthcheck.sh added --galera_online test, to match what the mariadb-operator does.

Variables:
- Added the note_verbosity system variable to manage notes when an index cannot be used