In certain cases AFTER UPDATE/DELETE triggers on NDB tables that referenced
subject table didn't see the results of operation which caused invocation
of those triggers. In other words AFTER trigger invoked as result of update
(or deletion) of particular row saw version of this row before update (or
deletion).
The problem occured because NDB handler in those cases postponed actual
update/delete operations to be able to perform them later as one batch.
This fix solves the problem by disabling this optimization for particular
operation if subject table has AFTER trigger for this operation defined.
To achieve this we introduce two new flags for handler::extra() method:
HA_EXTRA_DELETE_CANNOT_BATCH and HA_EXTRA_UPDATE_CANNOT_BATCH.
These are called if there exists AFTER DELETE/UPDATE triggers during a
statement that potentially can generate calls to delete_row()/update_row().
This includes multi_delete/multi_update statements as well as insert statements
that do delete/update as part of an ON DUPLICATE statement.
include/my_base.h:
Added HA_EXTRA_DELETE_CANNOT_BATCH and HA_EXTRA_UPDATE_CANNOT_BATCH to inform handler when batching of delete/update is not possible.
mysql-test/r/ndb_trigger.result:
Bug #26242 UPDATE with subquery and triggers failing with cluster tables
---
Added new test cases
mysql-test/t/ndb_trigger.test:
Bug #26242 UPDATE with subquery and triggers failing with cluster tables
---
Added new test cases
sql/ha_ndbcluster.cc:
Bug #26242 UPDATE with subquery and triggers failing with cluster tables: Use HA_EXTRA_DELETE_CANNOT_BATCH and HA_EXTRA_UPDATE_CANNOT_BATCH to inform handler when batching of delete/update is not possible
sql/ha_ndbcluster.h:
Bug #26242 UPDATE with subquery and triggers failing with cluster tables: Added member variables for handling of HA_EXTRA_DELETE_CANNOT_BATCH and HA_EXTRA_UPDATE_CANNOT_BATCH to inform handler when batching of delete/update is not possible
sql/mysql_priv.h:
Added new method prepare_triggers_for_insert_stmt to check if batching of delete/update must be disallowed.
sql/sql_delete.cc:
Use HA_EXTRA_DELETE_CANNOT_BATCH to inform handler when batching of delete is not possible
sql/sql_insert.cc:
Added method prepare_triggers_for_insert_stmt to check if batching of delete/update must be dissallowed.
Use HA_EXTRA_DELETE_CANNOT_BATCH and HA_EXTRA_UPDATE_CANNOT_BATCH to inform handler
when batching of delete/update is not possible
sql/sql_load.cc:
Call prepare_triggers_for_insert_stmt to check if batching of delete/update must be dissallowed and
mark fields used by triggers for the insert statement.
sql/sql_trigger.h:
Added has_triggers to support what particular triggers exist on a table.
sql/sql_update.cc:
Use HA_EXTRA_UPDATE_CANNOT_BATCH to inform handler when batching of update is not possible
NDB table".
SQL-layer was not marking fields which were used in triggers as such. As
result these fields were not always properly retrieved/stored by handler
layer. So one might got wrong values or lost changes in triggers for NDB,
Federated and possibly InnoDB tables.
This fix solves the problem by marking fields used in triggers
appropriately.
Also this patch contains the following cleanup of ha_ndbcluster code:
We no longer rely on reading LEX::sql_command value in handler in order
to determine if we can enable optimization which allows us to handle REPLACE
statement in more efficient way by doing replaces directly in write_row()
method without reporting error to SQL-layer.
Instead we rely on SQL-layer informing us whether this optimization
applicable by calling handler::extra() method with
HA_EXTRA_WRITE_CAN_REPLACE flag.
As result we no longer apply this optimzation in cases when it should not
be used (e.g. if we have on delete triggers on table) and use in some
additional cases when it is applicable (e.g. for LOAD DATA REPLACE).
Finally this patch includes fix for bug#20728 "REPLACE does not work
correctly for NDB table with PK and unique index".
This was yet another problem which was caused by improper field mark-up.
During row replacement fields which weren't explicity used in REPLACE
statement were not marked as fields to be saved (updated) so they have
retained values from old row version. The fix is to mark all table
fields as set for REPLACE statement. Note that in 5.1 we already solve
this problem by notifying handler that it should save values from all
fields only in case when real replacement happens.
include/my_base.h:
Added HA_EXTRA_WRITE_CAN_REPLACE, HA_EXTRA_WRITE_CANNOT_REPLACE - new
parameters for ha_extra() method. We use them to inform handler that
write_row() which tries to insert new row into the table and encounters
some already existing row with same primary/unique key can replace old
row with new row instead of reporting error.
mysql-test/r/federated.result:
Additional test for bug#18437 "Wrong values inserted with a before update
trigger on NDB table".
mysql-test/r/ndb_replace.result:
Added test for bug #20728 "REPLACE does not work correctly for NDB table
with PK and unique index". Updated wrong results from older test.
mysql-test/t/federated.test:
Additional test for bug#18437 "Wrong values inserted with a before update
trigger on NDB table".
mysql-test/t/ndb_replace.test:
Added test for bug #20728 "REPLACE does not work correctly for NDB table
with PK and unique index".
sql/ha_ndbcluster.cc:
We no longer rely on reading LEX::sql_command value in handler in order
to determine if we can enable optimization which allows us to handle REPLACE
statement in more efficient way by doing replaces directly in write_row()
method without reporting error to SQL-layer.
Instead we rely on SQL-layer informing us whether this optimization
applicable by calling handler::extra() method with
HA_EXTRA_WRITE_CAN_REPLACE flag.
As result we no longer apply this optimization in cases when it should not
be used (e.g. if we have on delete triggers on table) and use in some
additional cases when it is applicable (e.g. for LOAD DATA REPLACE).
sql/item.cc:
Item_trigger_field::setup_field():
Added comment explaining why we don't set Field::query_id in this method.
sql/mysql_priv.h:
mysql_alter_table() function no longer takes handle_duplicates argument.
Added declaration of mark_fields_used_by_triggers_for_insert_stmt() function.
sql/sql_delete.cc:
Mark fields which are used by ON DELETE triggers so handler will retrieve
values for these fields.
sql/sql_insert.cc:
Explicitly inform handler that we are doing REPLACE (using ha_extra() method)
in cases when it can promote insert operation done by write_row() to replace.
Also when we do REPLACE we want to store values for all columns so we should
inform handler about it.
Finally we should mark fields used by ON UPDATE/ON DELETE triggers as such
so handler can properly retrieve/restore values in these fields during
execution of REPLACE and INSERT ... ON DUPLICATE KEY UPDATE statements.
sql/sql_load.cc:
Explicitly inform handler that we are doing LOAD DATA REPLACE (using
ha_extra() method) in cases when it can promote insert operation done by
write_row() to replace.
Also when we do replace we want to save (replace) values for all columns
so we should inform handler about it.
Finally to properly execute LOAD DATA for table with triggers we should
mark fields used by ON INSERT triggers as such so handler can properly
store values for these fields.
sql/sql_parse.cc:
mysql_alter_table() function no longer takes handle_duplicates argument.
sql/sql_table.cc:
Got rid of handle_duplicates argument in mysql_alter_table() and
copy_data_between_tables() functions. These functions were always
called with handle_duplicates == DUP_ERROR and thus contained dead
(and probably incorrect) code.
sql/sql_trigger.cc:
Added Table_triggers_list::mark_fields_used() method which is used to mark
fields read/set by triggers as such so handlers will be able properly
retrieve/store values in these fields.
sql/sql_trigger.h:
Table_triggers_list:
Added mark_fields_used() method which is used to mark fields read/set by
triggers as such so handlers will be able properly retrieve/store values
in these fields. To implement this method added 'trigger_fields' member
which is array of lists linking items for all fields used in triggers
grouped by event and action time.
sql/sql_update.cc:
Mark fields which are used by ON UPDATE triggers so handler will retrieve
and save values for these fields.
mysql-test/r/ndb_trigger.result:
Added test for bug#18437 "Wrong values inserted with a before update trigger
on NDB table".
mysql-test/t/ndb_trigger.test:
Added test for bug#18437 "Wrong values inserted with a before update trigger
on NDB table".
