Also, apply the MDEV-17957 changes to encrypted page checksums,
and remove error message output from the checksum function,
because these messages would be useless noise when mariabackup
is retrying reads of corrupted-looking pages, and not that
useful during normal server operation either.
The error messages in fil_space_verify_crypt_checksum()
should be refactored separately.
ha_innobase::truncate(): Because CREATE TEMPORARY TABLE
allows invalid table options when innodb_file_per_table=1,
do allow them also in TRUNCATE for temporary tables.
btr_node_ptr_max_size(): Treat CHAR(0) from SQL as a special case.
The InnoDB internal SQL parser maps the type "CHAR" to DATA_VARCHAR,
but MariaDB does allow CHAR(0) with an empty value, and does enforce
the length limitation.
ha_innobase::prepare_inplace_alter_table(): check max column length for every
index in a table, not just added in this particular ALTER TABLE with ADD INDEX ones.
Alter statement changed the THD structure by setting the value to FIELD_CHECK_WARN
and then not resetting it back. This led ANALYZE to throw a warning which previously
it didn't.
The problem here is EITS statistics does not calculate statistics for the partitions of the table.
So a temporary solution would be to not read EITS statistics for partitioned tables.
Also disabling reading of EITS for columns that participate in the partition list of a table.
merge_role_db_privileges() was remembering pointers into Dynamic_array
acl_dbs, and later was using them, while pushing more elements into the
array. But pushing can cause realloc, and it can invalidate all pointers.
Fix: remember and use indexes of elements, not pointers.
- This is a regression of commit b26e603aeb. While dropping
the incompletely created table, InnoDB shouldn't consider that operation as non-atomic one.
When the with clause of a query contains a recursive CTE that is not used
then processing of EXPLAIN for this query does not require optimization
of the unit specifying this CTE. In this case if 'derived' is the
TABLE_LIST object created for this CTE then derived->derived_result is NULL
and any assignment to derived->derived_result->table causes a crash.
After fixing this problem in the code of st_select_lex_unit::prepare()
EXPLAIN for such a query worked without crashes. Yet an execution
plan for the recursive CTE appeared there. The cause of this problem was
an incorrect condition used in JOIN::save_explain_data_intern() that
determined whether CTE was to be optimized or not. A similar condition was
used in select_describe() and this patch has corrected it as well.
Also, related to MDEV-15522, MDEV-17304, MDEV-17835,
remove the Galera xtrabackup tests, because xtrabackup never worked
with MariaDB Server 10.3 due to InnoDB redo log format changes.
There was an incorrect check for MariaDB and InnoDB
tables fields count. Corruption was reported when there was no corruption.
Also, a warning message had incorrect field numbers for both MariaDB and InnoDB
tables.
ha_innobase::open(): fixed check and message
This is a regression due to MDEV-17816.
When creating a table fails, we must roll back the dictionary
transaction. Because the rollback may rename tables, and because
InnoDB lacks proper undo logging for CREATE operations, we must
drop the incompletely created table before rolling back the
transaction, which could include a RENAME operation.
But, we must not blindly drop the table by name; after all,
the operation could have failed because another table by the
same name already existed.
create_table_info_t::m_drop_before_rollback: A flag that is set
if the table needs to be dropped before transaction rollback.
create_table_info_t::create_table(): Remove some duplicated
error handling.
ha_innobase::create(): On error, only drop the table if it was
actually created.
If a table had a KEY_BLOCK_SIZE attribute, but no ROW_FORMAT,
it would be created as ROW_FORMAT=COMPRESSED in InnoDB.
However, TRUNCATE TABLE would lose the KEY_BLOCK_SIZE attribute
and create the table with the innodb_default_row_format (DYNAMIC).
This is a regression that was introduced by MDEV-13564.
update_create_info_from_table(): Copy also KEY_BLOCK_SIZE.
The error handling in the MDEV-13564 TRUNCATE TABLE was broken
when an error occurred during table creation.
row_create_index_for_mysql(): Do not drop the table on error.
fts_create_one_common_table(), fts_create_one_index_table():
Do drop the table on error.
create_index(), create_table_info_t::create_table():
Let the caller handle the index creation errors.
ha_innobase::create(): If create_table_info_t::create_table()
fails, drop the incomplete table, roll back the transaction,
and finally return an error to the caller.
Problem was that controlling connection i.e. connection that
executed the query SET GLOBAL wsrep_reject_queries = ALL_KILL;
was also killed but server would try to send result from that
query to controlling connection resulting a assertion
mysqld: /home/jan/mysql/10.2-sst/include/mysql/psi/mysql_socket.h:738: inline_mysql_socket_send: Assertion `mysql_socket.fd != -1' failed.
as socket was closed when controlling connection was closed.
wsrep_close_client_connections()
Do not close controlling connection and instead of
wsrep_close_thread() we do now soft kill by THD::awake
wsrep_reject_queries_update()
Call wsrep_close_client_connections using current thd.
Problem was that controlling connection i.e. connection that
executed the query SET GLOBAL wsrep_reject_queries = ALL_KILL;
was also killed but server would try to send result from that
query to controlling connection resulting a assertion
mysqld: /home/jan/mysql/10.2-sst/include/mysql/psi/mysql_socket.h:738: inline_mysql_socket_send: Assertion `mysql_socket.fd != -1' failed.
as socket was closed when controlling connection was closed.
wsrep_close_client_connections()
Do not close controlling connection and instead of
wsrep_close_thread() we do now soft kill by THD::awake
wsrep_reject_queries_update()
Call wsrep_close_client_connections using current thd.
For prepare statemtent/stored procedures we rollback the items to original
ones after prepare execution in the function reinit_stmt_before_use.
This rollback is done for group by, order by clauses but is not done for
the window specification containing the order by and partition by clause of the
window function.
in 10.3, AWS encryption key was loaded during recovery, by the innodb purge
thread, therefore no note "loaded key" was written by the foreground
thread.
The fix for "MDEV-17698 MEMORY engine performance regression"
previously fixed this problem.
- Adding the test for MDEV-17724
- Re-recording wrong results for tests:
* engines/iuds/r/insert_number
* engines/iuds/r/update_delete_number
which started to fail since MDEV-17698
sp_instr_cursor_copy_struct::exec_core() created TYPELIBs on a wrong mem_root,
the one which is initialized in sp_head::execute(), this code:
/* init per-instruction memroot */
init_sql_alloc(&execute_mem_root, "per_instruction_memroot",
MEM_ROOT_BLOCK_SIZE, 0, MYF(0));
This memory root cleans up after every sp_instr_xxx executed, so later
sp_instr_cfetch::execute() tried to use already freed and trashed memory.
Changing sp_instr_cursor_copy_struct::exec_core() to call tmp.export_structure()
inside this block (not outside of it):
thd->set_n_backup_active_arena(thd->spcont->callers_arena, ¤t_arena);
...
thd->restore_active_arena(thd->spcont->callers_arena, ¤t_arena);
So now TYPELIBs created by sp_instr_cursor_copy_struct::exec_core() are
still available and valid when sp_instr_cfetch::execute() is called.
They are freed at the end of dispatch_command() corresponding to
the "CALL p1" statement.
- clean up DEFAULT() to work only with default value and correctly print
itself.
- fix of DBUG_ASSERT about fields read/write
- fix of field marking for write based really on the thd->mark_used_columns flag
MDEV-17660 sql_mode=ORACLE: Some keywords do not work as label names: history, system, versioning, without
MDEV-17661 Add sql_mode specific tokens for the keyword DECODE
During the optimize state of a query, we come know that the result set
would atmost contain one row, then for such a query we don't need
to compute GROUP BY, ORDER BY and DISTINCT.
Item_direct_view_ref::derived_field_transformer_for_where
upon updating a view
The condition pushed into a materialized derived / view mast be adjusted
for the new context: its column references must be substituted for
references to the columns of the underlying tables if the condition
is pushed into WHERE. The substitution is performed by the 'transform'
method. If the materialized derived is used in a mergeable view then
the references to the columns of the view are represented by
Item_direct_view_ref objects. The transform method first processes
the item wrapped in such an object and only after this it transforms
the object itself.
The transformation procedure of an Item_direct_view_ref object has
to know whether the item it wraps has been substituted. If so the
procedure does not have to do anything. In the code before this patch
it was not possible for the transformation procedure used by an
Item_direct_view_ref object to find out whether a substitution for
the wrapped item had happened.
Changing the way how a cursor is opened to fetch its structure only,
e.g. for a cursor FOR loop record variable.
The old methods with setting thd->lex->limit_rows_examined to an Item_uint(0)
was not reliable and could push these messages into diagnostics area:
The query examined at least 1 rows, which exceeds LIMIT ROWS EXAMINED (0)
The new method should be more reliable, as it completely prevents the call
of do_select() in JOIN::exec_inner() during the cursor structure discovery,
so the execution of the cursor SELECT query returns immediately after the
preparation step (when the result row structure becomes known),
without even entering the code that fetches the result rows.
The test innodb.log_file_name_debug failed to ensure that
the bogus redo log record that its debug injection emitted
would be consumed by a redo log checkpoint before running a
subsequent test, which could perform crash recovery.
Add an extra shutdown to ensure that a redo log checkpoint is
generated. In this way, the following will succeed:
./mtr --no-reorder innodb.log_file_name_debug innodb.read_only_recovery
This bug in the code of the function With_element::check_unrestricted_recursive()
could force a recursive CTE to be executed in a non-standard compliant mode
in which recursive UNION ALL could lead to an infinite execution. This
problem could occur only in the case when this CTE was used by another
recursive CTE at least twice.
There was a failure in rpl_delayed_slave after recent MDEV-14528 commit.
The parallel applier should not set its
Relay_log::last_master_timestamp from Format-descriptor log event.
The latter may reflect a deep past so Seconds-behind-master will be
computed through it and displayed all time while the first possibly
"slow" group of events is executed.
The main MDEV-14528 is refined, rpl_delayed_slave now passes also
in the parallel mode.
Remove the separate test innodb.alter_instant, because
it can be easily mistaken for innodb.instant_alter, which
in turn is covering various instant ALTER TABLE operations.
When replicated events are from Master unaware of MariaDB GTID their handling by
the Parallel slave misses Seconds_Behind_Master updating.
In the bug condition the Show-Slave-Status' field remains unchanged.
Because in such case event execution is sequential the bug
is fixed with deploying the same logics as in the explicit single-threaded mode
with is to set
Relay_log_event::last_master_timestamp
member early at the end of event reading from the relay log.
Background: Used encryption key_id is stored to encryption metadata
i.e. crypt_data that is stored on page 0 of the tablespace of the
table. crypt_data is created only if implicit encryption/not encryption
is requested i.e. ENCRYPTED=[YES|NO] table option is used
fil_create_new_single_table_tablespace on fil0fil.cc.
Later if encryption is enabled all tables that use default encryption
mode (i.e. no encryption table option is set) are encrypted with
default encryption key_id that is 1. See fil_crypt_start_encrypting_space on
fil0crypt.cc.
ha_innobase::check_table_options()
If default encryption is used and encryption is disabled, you may
not use nondefault encryption_key_id as it is not stored anywhere.
main.derived_cond_pushdown: Move all 10.3 tests to the end,
trim trailing white space, and add an "End of 10.3 tests" marker.
Add --sorted_result to tests where the ordering is not deterministic.
main.win_percentile: Add --sorted_result to tests where the
ordering is no longer deterministic.
Problem affects INPLACE ALTER rename columns.
innobase_rename_column_try(): some strcmp() was replaced with my_strcasecmp(),
queries to update data dictionary was updated to not match column name case.
Race condition. field->flags were copied from s->field->flags during
field->clone(), early in open_table_from_share(). But s->field->flags
were getting their PART_INDIRECT_KEY_FLAG bit much later in
TABLE::mark_columns_used_by_virtual_fields() and only once per share.
If two threads were executing the code between field->clone()
and mark_columns_used_by_virtual_fields() at the same time, only
one would get PART_INDIRECT_KEY_FLAG bits in field[].
thd_rpl_stmt_based(): A new predicate to check if statement-based
replication is active. (This can also hold when replication is not
in use, but binlog is.)
que_thr_stop(), row_ins_duplicate_error_in_clust(),
row_ins_sec_index_entry_low(), row_ins(): On a duplicate key error,
only lock all index records when statement-based replication is in use.
Users expect window functions to produce a certain ordering of rows in
the final result set. Although the standard does not require this, we
already have the filesort result done for when we computed the window
function. If there is no ORDER BY attached to the query, just keep it
till the SELECT is completely evaluated and use that to print the
result.
Update test cases as many did not take care to guarantee a stable
result.
set both `password` and `authentication_string` columns in `mysql`.`user`
table for now.
Suppress the "password was ignored" warning if the password is
the same as the authentication string
* Fine-grained inplace skipping by INNOBASE_ALTER_VERSIONED_REBUILD;
* Fixed column WITHOUT SYSTEM VERSIONING + ADD COLUMN;
* Fixed instant field change (MDEV-16330);
* Revisited test versioning.online;
* Merged the test versioning.trx_id_versioning_attribute_persistence
to versioning.online;
* Renamed some versioning functions:
** change_fields_versioning_cache() -> vers_change_fields_cache()
** change_field_versioning_try() -> vers_change_field_try()
Skip condition moved out of func.
Closestempesta-tech/mariadb#414Closestempesta-tech/mariadb#540
Related to tempesta-tech/mariadb#281
For index merge union[or sort union], the estimates are not taken into account while calculating the selectivity of
a condition. So instead of showing the estimates of the index merge union[or sort union], it shows estimates equal to
all the records of the table.
The fix for the issue is to include the selectivity of index merge
union[or sort union] while calculating the selectivity of a condition.
--gdb now accepts an argument, it will be passed to gdb as a command.
multiple commands can be separated by a (non-standard and not escapable)
delimiter - semicolon (;).
Old usage with a bare --gdb continues to work too, of course.
Cherry-picked c47c0ca50c5441bbd3b1339b905579
row_drop_table_for_mysql(): Avoid accessing non-existing dictionary tables.
dict_create_or_check_foreign_constraint_tables(): Add debug instrumentation
for creating and dropping a table before the creation of any non-core
dictionary tables.
trx_purge_add_update_undo_to_history(): Adjust a debug assertion, so that
it will not fail due to the test instrumentation.
row_drop_table_for_mysql(): Avoid accessing non-existing dictionary tables.
dict_create_or_check_foreign_constraint_tables(): Add debug instrumentation
for creating and dropping a table before the creation of any non-core
dictionary tables.
trx_purge_add_update_undo_to_history(): Adjust a debug assertion, so that
it will not fail due to the test instrumentation.
Fixed by adding table flag HA_WANTS_PRIMARY_KEY, which is like
HA_REQUIRE_PRIMARY_KEY but tells SQL upper layer that the storage engine
internally can handle tables without primary keys (for example for
sequences or trough user variables)
The ONLY_FULL_GROUP_BY mode states that for SELECT ... GROUP BY queries,
disallow SELECTing columns which are not referred to in the GROUP BY clause,
unless they are passed to an aggregate function like COUNT() or MAX().
This holds only for the GROUP BY clause of the query.
The code also checks this for the partition clause of the window function which is
incorrect.
row_build_index_entry_low(): ext does not contain virtual columns.
row_upd_store_v_row(): Copy virtual column values
This is based on the following fix in MySQL 5.7.24:
commit 4ec2158bec73f1582501c4b3e3de250fed9edc9a
Author: Sachin Agarwal <sachin.z.agarwal@oracle.com>
Date: Fri Aug 24 14:44:13 2018 +0530
Bug #27968952 INNODB CRASH/CORRUPTION WITH TEXT PREFIX INDEXES
Problem:
There are two problems:
1. If there is one secondary index on extenally
stored column and another seconday index on virtual column (whose
base column is not externally stored). then while updating seconday
index on vitrual column, virtual column data is replaced by
externally stoared column.
2. In row update operation, node->row contains
shallow copy of virtual data fields. While building an update vector
containing all the fields to be modified, compute virtual column.
which may causes change in virtual data fields in node->row.
In both the above cases, while updating seconday index on virtual
column, couldn't find the row and hit an explicite assert inside
ROW_NOT_FOUND.
Fix:
1. Added check if column is virtual then its ext flag should be ZERO
and virtual column data will not be replaced by offset column data.
2. Deep copy of virtual data fields for node->row.
RB: #20382
Reviewed by : Jimmy.Yang@oracle.com
