After table->file->add_index() in mysql_alter_table() the table in the engine
has the intermediate temporary structure, it's neither the original nor the
final table structure (it'll be final after successful table->file->drop_index()
call). So, when add_index() fails with a unique key violation, we cannot simply
get the failed key number and easily map it to the key name and key structure via
table->key_info[key_no].
For now we'll create this "intermediate temporary structure", emulating InnoDB
internal rules.
This bug and the fix will go away in 10.0 that uses completely different online
alter table code.
mysql-test/t/alter_table_trans.test:
mdev:5406
merged from 5.6:
Bug#14521864: MYSQL 5.1 TO 5.5 BUGS PARTITIONING
Bug#16589511: MYSQL_UPGRADE FAILS TO WRITE OUT ENTIRE ALTER TABLE ... ALGORITHM= ... STATEMENT
Bug#16274455: CAN NOT ACESS PARTITIONED TABLES WHEN DOWNGRADED FROM 5.6.11 TO 5.6.10
plus minor changes from 5.6, mainly comments
This is now otherwise on level wsrep-25.9, but storage/innobase has not been fully merged
wsrep-5.5 is not good source for that, so we probably have to cherry pick innodb changes from wsrep-5.6
- Character set code & tests from Alexander Barkov
- Integration with ALTER TABLE, REPAIR and open_table from Monty
The problem was that MySQL 5.6 added some croatian and vitanamese character set collations that are incompatible with MariaDB.
The fix is to move the MariaDB conflicting collation numbers out of the region that MySQL is likely to use.
mysql_upgrade, REPAIR TABLE or ALTER TABLE will fix the collations.
If one tries to access and old incompatible table, one will get the error "Table upgrade required...."
After this patch, MariaDB supports all the MySQL character set collations and the old MariaDB croatian collations, which are closer to the latest standard than the MySQL versions.
New character sets:
ucs2_croatian_mysql561_uca_ci
utf8_croatian_mysql561_uca_ci
utf16_croatian_mysql561_uca_ci
utf32_croatian_mysql561_uca_ci
utf8mb4_croatian_mysql561_uca_ci
Other things:
- Fixed some compiler warnings
- mysql_upgrade prints information about repaired tables.
- Increased version number
VERSION:
Increased VERSION number
client/mysqlcheck.c:
Print repaired table name when using --verbose
include/m_ctype.h:
Add new MariaDB collation regions that are not likely to conflict with MySQL
include/my_base.h:
Added flag to detect if table was opened for ALTER TABLE
mysql-test/r/ctype_ldml.result:
Updated result
mysql-test/r/ctype_uca.result:
Updated result
mysql-test/r/ctype_upgrade.result:
Updated result
mysql-test/r/ctype_utf16_uca.result:
Updated result
mysql-test/r/ctype_utf32_uca.result:
Updated result
mysql-test/r/ctype_utf8mb4_uca.result:
Updated result
mysql-test/std_data/ctype_upgrade:
Test files for testing upgrading of conflicting collations
mysql-test/suite/engines/funcs/r/db_alter_collate_ascii.result:
New collations added
mysql-test/suite/engines/funcs/r/db_alter_collate_utf8.result:
New collations added
mysql-test/suite/innodb/r/innodb_ctype_ldml.result:
Updated test result
mysql-test/suite/innodb/t/innodb_ctype_ldml.test:
Updated test result
mysql-test/suite/plugins/r/show_all_plugins.result:
Updated version number
mysql-test/suite/roles/create_and_drop_role_invalid_user_table.result:
Updated version number
mysql-test/t/ctype_ldml.test:
Updated test
mysql-test/t/ctype_uca.test:
Testing of new collations
mysql-test/t/ctype_upgrade.test:
Testing of upgrading tables with old collations
The test ensures that:
- We will get an error if we try to open a table with old collations.
- CHECK TABLE will detect that the table needs to be upgraded.
- ALTER TABLE and REPAIR will fix the table.
- mysql_upgrade works as expected
mysql-test/t/ctype_utf16_uca.test:
Testing of new collations
mysql-test/t/ctype_utf32_uca.test:
Testing of new collations
mysql-test/t/ctype_utf8mb4_uca.test:
Testing of new collations
mysys/charset-def.c:
Added new character sets
mysys/charset.c:
Always give an error, if requested, if a character set didn't exist
sql/handler.cc:
- Added upgrade_collation() to check if collation is compatible with old version
- check_collation_compatibility() checks if we are using an old collation from MariaDB 5.5 or MySQL 5.6
- ha_check_for_upgrade() returns HA_ADMIN_NEEDS_ALTER if we have an incompatible collation
sql/handler.h:
Added new prototypes
sql/sql_table.cc:
- Mark that tables are opened for ALTER TABLE
- If table needs to be upgraded, ensure we are not using online alter table.
sql/table.cc:
- If we are using an old incompatible collation, change to use the new one and mark table as incompatible.
- Give an error if we try to open an incompatible table.
sql/table.h:
Added error that table needs to be rebuild
storage/connect/ha_connect.cc:
Fixed compiler warning
strings/ctype-uca.c:
New character sets
sql/create_options.cc:
an utility function to compare two filled-in engine_option structures
sql/sql_table.cc:
* two keys are different if their option_struct's differ
(for ALTER TABLE DROP key, ADD key)
* engines doing inplace alter must see the new frm image
STATUS OF ROLLBACKED TRANSACTION" and bug #17054007 - "TRANSACTION
IS NOT FULLY ROLLED BACK IN CASE OF INNODB DEADLOCK".
The problem in the first bug report was that although deadlock involving
metadata locks was reported using the same error code and message as InnoDB
deadlock it didn't rollback transaction like the latter. This caused
confusion to users as in some cases after ER_LOCK_DEADLOCK transaction
could have been restarted immediately and in some cases rollback was
required.
The problem in the second bug report was that although InnoDB deadlock
caused transaction rollback in all storage engines it didn't cause release
of metadata locks. So concurrent DDL on the tables used in transaction was
blocked until implicit or explicit COMMIT or ROLLBACK was issued in the
connection which got InnoDB deadlock.
The former issue has stemmed from the fact that when support for detection
and reporting metadata locks deadlocks was added we erroneously assumed
that InnoDB doesn't rollback transaction on deadlock but only last statement
(while this is what happens on InnoDB lock timeout actually) and so didn't
implement rollback of transactions on MDL deadlocks.
The latter issue was caused by the fact that rollback of transaction due
to deadlock is carried out by setting THD::transaction_rollback_request
flag at the point where deadlock is detected and performing rollback
inside of trans_rollback_stmt() call when this flag is set. And
trans_rollback_stmt() is not aware of MDL locks, so no MDL locks are
released.
This patch solves these two problems in the following way:
- In case when MDL deadlock is detect transaction rollback is requested
by setting THD::transaction_rollback_request flag.
- Code performing rollback of transaction if THD::transaction_rollback_request
is moved out from trans_rollback_stmt(). Now we handle rollback request
on the same level as we call trans_rollback_stmt() and release statement/
transaction MDL locks.
STATUS OF ROLLBACKED TRANSACTION" and bug #17054007 - "TRANSACTION
IS NOT FULLY ROLLED BACK IN CASE OF INNODB DEADLOCK".
The problem in the first bug report was that although deadlock involving
metadata locks was reported using the same error code and message as InnoDB
deadlock it didn't rollback transaction like the latter. This caused
confusion to users as in some cases after ER_LOCK_DEADLOCK transaction
could have been restarted immediately and in some cases rollback was
required.
The problem in the second bug report was that although InnoDB deadlock
caused transaction rollback in all storage engines it didn't cause release
of metadata locks. So concurrent DDL on the tables used in transaction was
blocked until implicit or explicit COMMIT or ROLLBACK was issued in the
connection which got InnoDB deadlock.
The former issue has stemmed from the fact that when support for detection
and reporting metadata locks deadlocks was added we erroneously assumed
that InnoDB doesn't rollback transaction on deadlock but only last statement
(while this is what happens on InnoDB lock timeout actually) and so didn't
implement rollback of transactions on MDL deadlocks.
The latter issue was caused by the fact that rollback of transaction due
to deadlock is carried out by setting THD::transaction_rollback_request
flag at the point where deadlock is detected and performing rollback
inside of trans_rollback_stmt() call when this flag is set. And
trans_rollback_stmt() is not aware of MDL locks, so no MDL locks are
released.
This patch solves these two problems in the following way:
- In case when MDL deadlock is detect transaction rollback is requested
by setting THD::transaction_rollback_request flag.
- Code performing rollback of transaction if THD::transaction_rollback_request
is moved out from trans_rollback_stmt(). Now we handle rollback request
on the same level as we call trans_rollback_stmt() and release statement/
transaction MDL locks.
Added partition_exchange test.
Do not set HA_OPTION_PACK_RECORD for InnoDB specific row formats
(e.g. COMPACT, REDUNDANT). Adjusted mysql_compare_tables() accordingly.
Following variables do not require LOCK_open protection anymore:
- table_def_cache (renamed to tdc_hash) is protected by rw-lock
LOCK_tdc_hash;
- table_def_shutdown_in_progress doesn't need LOCK_open protection;
- last_table_id use atomics;
- TABLE_SHARE::ref_count (renamed to TABLE_SHARE::tdc.ref_count)
is protected by TABLE_SHARE::tdc.LOCK_table_share;
- TABLE_SHARE::next, ::prev (renamed to tdc.next and tdc.prev),
oldest_unused_share, end_of_unused_share are protected by
LOCK_unused_shares;
- TABLE_SHARE::m_flush_tickets (renamed to tdc.m_flush_tickets)
is protected by TABLE_SHARE::tdc.LOCK_table_share;
- refresh_version (renamed to tdc_version) use atomics.
IN TIME RECOVERY FAILURE ON SLAVES
Problem:
DROP TEMP TABLE IF EXISTS commands can cause point
in time recovery (re-applying binlog) failures.
Analyses:
In RBR, 'DROP TEMPORARY TABLE' commands are
always binlogged by adding 'IF EXISTS' clauses.
Also, the slave SQL thread will not check replicate.* filter
rules for "DROP TEMPORARY TABLE IF EXISTS" queries.
If log-slave-updates is enabled on slave, these queries
will be binlogged in the format of "USE `db`;
DROP TEMPORARY TABLE IF EXISTS `t1`;" irrespective
of filtering rules and irrespective of the `db` existence.
When users try to recover slave from it's own binlog,
use `db` command might fail if `db` is not present on slave.
Fix:
At the time of writing the 'DROP TEMPORARY TABLE
IF EXISTS' query into the binlog, 'use `db`' will not be
present and the table name in the query will be a fully
qualified table name.
Eg:
'USE `db`; DROP TEMPORARY TABLE IF EXISTS `t1`;'
will be logged as
'DROP TEMPORARY TABLE IF EXISTS `db`.`t1`;'.
IN TIME RECOVERY FAILURE ON SLAVES
Problem:
DROP TEMP TABLE IF EXISTS commands can cause point
in time recovery (re-applying binlog) failures.
Analyses:
In RBR, 'DROP TEMPORARY TABLE' commands are
always binlogged by adding 'IF EXISTS' clauses.
Also, the slave SQL thread will not check replicate.* filter
rules for "DROP TEMPORARY TABLE IF EXISTS" queries.
If log-slave-updates is enabled on slave, these queries
will be binlogged in the format of "USE `db`;
DROP TEMPORARY TABLE IF EXISTS `t1`;" irrespective
of filtering rules and irrespective of the `db` existence.
When users try to recover slave from it's own binlog,
use `db` command might fail if `db` is not present on slave.
Fix:
At the time of writing the 'DROP TEMPORARY TABLE
IF EXISTS' query into the binlog, 'use `db`' will not be
present and the table name in the query will be a fully
qualified table name.
Eg:
'USE `db`; DROP TEMPORARY TABLE IF EXISTS `t1`;'
will be logged as
'DROP TEMPORARY TABLE IF EXISTS `db`.`t1`;'.
Fixed maria.maria and funcs_1.is_tables failure.
sql/sql_table.cc:
C_ALTER_TABLE_FRM_ONLY lost it's meaning after merge of inplace alter:
now it is used even if table is to be copied.
Otherwise: how can row format change go without notifying storage engine?
Removed check for create_table_mode for now, to be discussed.
sql/sql_table.cc:
Removed false assertion that HA_NO_COPY_ON_ALTER is merge specific.
This flag is used at least by Spider and Connect.
storage/spider/ha_spider.cc:
Renamed THD::stmt_da -> THD::get_stmt_da().
Renamed KEY::key_parts -> KEY::user_defined_key_parts.
Renamed ALTER_* flags -> Alter_info::ALTER_*.
storage/spider/spd_conn.cc:
Renamed THD::stmt_da -> THD::get_stmt_da().
storage/spider/spd_db_conn.cc:
Renamed MYSQL_ERROR -> Sql_condition.
Renamed KEY::key_parts -> KEY::user_defined_key_parts.
storage/spider/spd_db_include.h:
current_comment_start argument of generate_parition_syntax()
was removed during merge.
storage/spider/spd_db_mysql.cc:
Renamed KEY::key_parts -> KEY::user_defined_key_parts.
storage/spider/spd_ping_table.cc:
Renamed THD::stmt_da -> THD::get_stmt_da().
storage/spider/spd_table.cc:
Include my_getopt.h so that it exports my_defaults_file and
my_defaults_extra_file.
Renamed KEY::key_parts -> KEY::user_defined_key_parts.
* update results
* don't force HA_CREATE_DELAY_KEY_WRITE on all temp tables,
(bad for CREATE ... LIKE) instead imply it in myisam/aria
* restore HA_ERR_TABLE_DEF_CHANGED in archive
* increase the default number of rwlock classes in P_S to fit all our rwlocks
includes:
* remove some remnants of "Bug#14521864: MYSQL 5.1 TO 5.5 BUGS PARTITIONING"
* introduce LOCK_share, now LOCK_ha_data is strictly for engines
* rea_create_table() always creates .par file (even in "frm-only" mode)
* fix a 5.6 bug, temp file leak on dummy ALTER TABLE
Includes 5.6 changesets for:
*****
Fix for BUG#13489996 valgrind:conditional jump or move depends on uninitialised values-field_blob.
blob_ptr_size was not initialized properly: remove this variable.
*****
Bug#14021323 CRASH IN FIELD::SET_NULL WHEN INSERTING ROWS TO NEW TABLE
*****
