Atomic CREATE OR REPLACE allows to keep an old table intact if the
command fails or during the crash. That is done by renaming the
original table to temporary name, as a backup and restoring it if the
CREATE fails. When the command is complete and logged the backup
table is deleted.
Atomic replace algorithm
Two DDL chains are used for CREATE OR REPLACE:
ddl_log_state_create (C) and ddl_log_state_rm (D).
1. (C) Log rename of ORIG to TMP table (Rename TMP to original).
2. Rename orignal to TMP.
3. (C) Log CREATE_TABLE_ACTION of ORIG (drops ORIG);
4. Do everything with ORIG (like insert data)
5. (D) Log drop of TMP
6. Write query to binlog (this marks (C) to be closed in
case of failure)
7. Execute drop of TMP through (D)
8. Close (C) and (D)
If there is a failure before 6) we revert the changes in (C)
Chain (D) is only executed if 6) succeded (C is closed on
crash recovery).
Foreign key errors will be found at the 1) stage.
Additional notes
- CREATE TABLE without REPLACE and temporary tables is not affected
by this commit.
set @@drop_before_create_or_replace=1 can be used to
get old behaviour where existing tables are dropped
in CREATE OR REPLACE.
- CREATE TABLE is reverted if binlogging the query fails.
- Engines having HTON_EXPENSIVE_RENAME flag set are not affected by
this commit. Conflicting tables marked with this flag will be
deleted with CREATE OR REPLACE.
- Replication execution is not affected by this commit.
- Replication will first drop the conflicting table and then
creating the new one.
- CREATE TABLE .. SELECT XID usage is fixed and now there is no need
to log DROP TABLE via DDL_CREATE_TABLE_PHASE_LOG (see comments in
do_postlock()). XID is now correctly updated so it disables
DDL_LOG_DROP_TABLE_ACTION. Note that binary log is flushed at the
final stage when the table is ready. So if we have XID in the
binary log we don't need to drop the table.
- Three variations of CREATE OR REPLACE handled:
1. CREATE OR REPLACE TABLE t1 (..);
2. CREATE OR REPLACE TABLE t1 LIKE t2;
3. CREATE OR REPLACE TABLE t1 SELECT ..;
- Test case uses 6 combinations for engines (aria, aria_notrans,
myisam, ib, lock_tables, expensive_rename) and 2 combinations for
binlog types (row, stmt). Combinations help to check differences
between the results. Error failures are tested for the above three
variations.
- expensive_rename tests CREATE OR REPLACE without atomic
replace. The effect should be the same as with the old behaviour
before this commit.
- Triggers mechanism is unaffected by this change. This is tested in
create_replace.test.
- LOCK TABLES is affected. Lock restoration must be done after new
table is created or TMP is renamed back to ORIG
- Moved ddl_log_complete() from send_eof() to finalize_ddl(). This
checkpoint was not executed before for normal CREATE TABLE but is
executed now.
- CREATE TABLE will now rollback also if writing to the binary
logging failed. See rpl_gtid_strict.test
backup ddl log changes
- In case of a successfull CREATE OR REPLACE we only log
the CREATE event, not the DROP TABLE event of the old table.
ddl_log.cc changes
ddl_log_execute_action() now properly return error conditions.
ddl_log_disable_entry() added to allow one to disable one entry.
The entry on disk is still reserved until ddl_log_complete() is
executed.
On XID usage
Like with all other atomic DDL operations XID is used to avoid
inconsistency between master and slave in the case of a crash after
binary log is written and before ddl_log_state_create is closed. On
recovery XIDs are taken from binary log and corresponding DDL log
events get disabled. That is done by
ddl_log_close_binlogged_events().
On linking two chains together
Chains are executed in the ascending order of entry_pos of execute
entries. But entry_pos assignment order is undefined: it may assign
bigger number for the first chain and then smaller number for the
second chain. So the execution order in that case will be reverse:
second chain will be executed first.
To avoid that we link one chain to another. While the base chain
(ddl_log_state_create) is active the secondary chain
(ddl_log_state_rm) is not executed. That is: only one chain can be
executed in two linked chains.
The interface ddl_log_link_chains() was defined in "MDEV-22166
ddl_log_write_execute_entry() extension".
Atomic info parameters in HA_CREATE_INFO
Many functions in CREATE TABLE pass the same parameters. These
parameters are part of table creation info and should be in
HA_CREATE_INFO (or whatever). Passing parameters via single
structure is much easier for adding new data and
refactoring.
Aria changes:
- Fixed issue in Aria engine with CREATE + locked tables
that data was not properly commited in some cases in
case of crashes.
InnoDB related changes (by Marko Mäkelä):
- table_name_t::is_create_or_replace(): A new predicate to check for
CREATE OR REPLACE TABLE will rename an old table to
and eventually drop after creating the replacement.
- dict_table_t::parse_name(): Do acquire MDL on #sql-create- names
for partitioned tables.
- dict_table_rename_in_cache(): On CREATE OR REPLACE TABLE ... SELECT,
forget the original dict_table_t::mdl_name so that purge will
acquire MDL on the #sql-create- name instead. In this way, the
MDL_EXCLUSIVE that the CREATE OR REPLACE TABLE holds on the
user-visible name will not unnecessarily block any purge of old history
until the very end when the #sql-create- table will be dropped.
- ha_innobase::delete_table(): Do not check FOREIGN KEY consistency
when dropping an #sql-create- table.
- row_rename_table_for_mysql(): Update SYS_FOREIGN.ID also
when renaming to #sql-create- in order to avoid any
duplicate key error when CREATE OR REPLACE TABLE is
creating some FOREIGN KEY constraints by names
that existed in the old table.
Other changes:
- Removed some auto variables in log.cc for better code readability.
- Fixed old bug that CREATE ... SELECT would not be able to auto repair
a table that is part of the SELECT.
- Marked MyISAM that it does not support ROLLBACK (not required but
done for better consistency with other engines).
- maria_create_trn_for_mysql() does not register a new transaction
handler for commits. This was needed to ensure create or replace
will not end with an active transaction.
- We do not get anymore warnings about "Engine not supporting atomic
create" when doing a legal CREATE OR REPLACE on a table with
foreign key constraints.
- Updated VIDEX engine flags to disable CREATE SEQUENCE.
- Removed mysql_mutex_unlock(&LOCK_gdl) / mysql_mutex_lock(&LOCK_gdl)
around calls to binlog as these are unsafe. The binlog code uses
global variables that needs protection from other caller.
- EITS data is preserved if create or replace fails if
drop_before_create_or_replace=OFF. If ON, then create or replace
will drop EITS before the drop of the original table (as before).
- Using CREATE OR REPLACE on a encrypted table that the user cannot
decrypt will fail instead of replacing the encrypted table.
The encrypted table will unchanged.
Known issues:
- One cannot use create or replace on an InnoDB tables that has foreign
key references point to it
- CREATE OR REPLACE TEMPORARY table is not full atomic. Any conflicting
table will always be dropped before creating a new one. (Old behaviour).
Bug fixes related to this MDEV:
MDEV-36435 Assertion failure in finalize_locked_tables()
MDEV-36439 Assertion `thd_arg->lex->sql_command != SQLCOM_CREATE_SEQUENCE...
MDEV-36498 Failed CoR in non-atomic mode no longer generates DROP in RBR...
MDEV-36508 Temporary files #sql-create-....frm occasionally stay after
crash recovery
MDEV-38479 Crash in CREATE OR REPLACE SEQUENCE when new sequence cannot
be created
MDEV-36497 Assertion failure after atomic CoR with Aria under lock in
transactional context
MDEV-36501 EITS data is lost after failed attempt to CREATE OR REPLACE
table
MDEV-36493 Atomic CREATE OR REPLACE ... SELECT blocks InnoDB purge
MDEV-39367 MSAN/valgrind errors in temp_file_size_cb_func,
main.tmp_space_usage fails
MDEV-39446 Atomic CREATE OR REPLACE fails if a table cannot be decrypted
MDEV-40776 Atomic CREATE OR REPLACE silently breaks the foreign key
MDEV-40765 Assertion `"unexpected references" == 0' failed upon failing
CREATE OR REPLACE
MDEV-40994 Atomic CoR: Transaction not registered for MariaDB 2PC, but
transaction is active
Reverted commits:
- MDEV-36685 "CREATE-SELECT may lose in binlog side-effects of
stored-routine" as it did not take into account that it safe to
clear binlogs if the created table is non transactional and there
are no other non transactional tables used.
This was done because it caused extra logging when it is not
needed (not using any non transactional tables) and it also did
not solve side effects when using statement based loggging.
In MDEV-25644 vers_check_update() sets vers_write to false in case it
returns false. It is ok for UPDATE but is not correct for ODKU is bulk
insert requires vers_write on next tuple.
The fix return vers_write value back when vers_check_update() and
related vers_insert_history_row() are done in ODKU.
Problem:
REPLACE DELAYED into a system-versioned table read an uninitialized
lex->vers_conditions.delete_history in TABLE::delete_row().
Delayed_insert has its own THD, and hence its own LEX, of which the
constructor copies only a few fields from the parser LEX.
vers_conditions was not copied, so it stayed uninitialized. It is read
when a REPLACE resolves a duplicate on a versioned table.
Fix:
Copy vers_conditions from the parser LEX in the Delayed_insert
constructor.
Add new acquire_lock() overloads on MDL_context that combine MDL_request
initialization and lock acquisition into a single call, returning
MDL_ticket* directly. Convert applicable call sites across the codebase
to use the new ACQUIRE_LOCK macro.
All new code of the whole pull request, including one or several files
that are either new files or modified ones, are contributed under the
BSD-new license. I am contributing on behalf of my employer Amazon Web
Services, Inc.
The patch introduces the OLD_VALUE() expression to reference the value
of a column before it was updated. The parser is extended to support
RETURNING and OLD_VALUE(), and RETURNING expressions are stored in a
separate returning_list in SELECT_LEX with independent wildcard tracking.
RETURNING is rejected for multi-table UPDATE.
During setup of RETURNING fields, THD::is_setting_returning is used
when resolving fields, particularly for updates through views.
When resolving view fields, Item_direct_view_ref may point to the
view's item_list, losing the information about whether the value
should be old or new. The original item in returning_list still
contains the correct is_old_value_reference flag, which is copied
back to the resolved item.
OLD_VALUE() is implemented by extending Item_field with a new
Item_old_field class. Item_field::set_field() initializes
Field::ptr_old to the corresponding location in record[1],
which stores the old row during UPDATE execution.
When sending result rows, Item_old_field::send() temporarily switches
the field pointer from record[0] (current row) to ptr_old so
OLD_VALUE() returns the value before the update.
The UPDATE execution path is modified to send a result set when
RETURNING is present instead of an OK packet.
Remove the unscoped #pragma GCC diagnostic ignored
"-Wdeprecated-declarations" from include/violite.h and replace all
sprintf/vsprintf calls with snprintf/vsnprintf across the codebase.
Where possible, pass actual buffer sizes as function parameters
instead of using hardcoded constants or recomputing strlen().
Notable API changes:
- get_date(): added to_len parameter for buffer size
- nice_time(), end_timer() in client/mysql.cc: added buff_size
- generate_new_name(): added name_size parameter
- calc_md5(): added buffer_size parameter (fixes off-by-one)
- MD5_HASH_TO_STRING macro: added _size parameter
- Added MYSQL_UDF_MAX_RESULT_LENGTH constant in mysql_com.h
Also fixed a pre-existing bug in sql/rpl_mi.cc where the ellipsis
"..." was written to the wrong buffer (dbuff instead of buff).
Vendored code is excluded: extra/readline, extra/wolfssl,
wsrep-lib, libmariadb, storage/mroonga/vendor, zlib.
Prior to this change, AFTER UPDATE did not fire, if data was unchanged.
Now this is fixed. Also added a test to verify that trigger behavior for
on-duplicate-update is the same as normal update, in case data does not
change.
When prepare_for_replace() fails, the code returned early via
DBUG_RETURN(1), bypassing the abort label which resets
thd->abort_on_warning to 0. This left the THD in a dirty state,
triggering the dispatch_command assertion at end of command dispatch.
Signed-off-by: Amodh Dhakal <amodhakal@gmail.com>
Problem:
=========
An assertion failure occurs in InnoDB during consecutive
ALTER TABLE operations using the COPY algorithm. The crash
happens during the table rename phase because the
source table (dict_table_t::n_ref_count) is non-zero, despite
the thread holding an exclusive Metadata Lock (MDL).
Reason:
========
When ALTER IGNORE TABLE is executed via the COPY algorithm,
it generates undo logs for every row inserted into the
intermediate table (e.g., #sql-alter-...). The background Purge
Thread, responsible for cleaning up these undo logs, attempts
to take an MDL on the table to prevent the table from
being dropped while in use.
Race condition:
==================
First ALTER: Creates #sql-alter-, copies data, and renames it to t1.
Purge Activation: The Purge thread picks up the undo logs from step 1.
It takes an MDL on the temporary name (#sql-alter-) and increments
the table's n_ref_count.
Identity Shift: InnoDB renames the physical table object to t1, but
the Purge thread still holds a reference to this object.
Second ALTER: Starts a new copy process. When it attempts to rename
the "new" t1 to a backup name, it checks if n_ref_count == 0.
Because the Purge thread is still "pinning" the object to
clean up logs from the first ALTER, the count is > 0,
triggering the assertion failure.
Solution:
========
ALTER IGNORE TABLE needs row-level undo logging to easily
roll back the last inserted row in case of duplicate key errors.
By discarding the last undo log record after inserting each row,
purge will not process any log records generated by
ALTER IGNORE TABLE, preventing unexpected access from the purge
subsystem during subsequent DDL operations.
Make skip_alter_undo (1-bit) to (2-bit enum)
to support different ALTER operation modes:
- NO_UNDO (0): Normal mode with standard undo logging
- SKIP_UNDO (1): ALTER mode that skips undo logging
- IGNORE_UNDO (2): ALTER IGNORE mode that rewrites undo blocks
trx_undo_report_row_operation(): Add ALTER IGNORE undo
rewriting logic. Store old undo record info before writing
new records for IGNORE_UNDO mode. Reset undo top_offset
to maintain only latest insert undo
row_ins_clust_index_entry_low(): Prevent ALTER IGNORE from
using bulk insert optimization. since it requires to write
undo log for each row operation that's incompatible with
existing bulk operation.
A new assertion added by MDEV-36290 is correct in trying to identify
default_fields that weren't correctly accounted in parse_vcol_defs.
It caught a case for INSERT DELAYED on tables using TIMESTAMP and
DATETIME types with ON UPDATE default values. That is, the insert
delayed logic opens the table in one thread, and then copies the
field info to another table in the user thread. This copy loses most
flag metadata for these fields though. In this case, the flag
ON_UPDATE_NOW_FLAG was dropped, and the parse_col_defs() logic was
unable to properly account for that default field, and the assertion
fired.
This patch fixes this by restoring the ON_UPDATE_NOW_FLAG during the
field copy. Also, a comment is added to Field::make_new_field() to
explain the state of the returned field.
Reviewed-by: Monty <monty@mariadb.com>
Signed-off-by: Brandon Nesterenko <brandon.nesterenko@mariadb.com>
Reason:
======
During InnoDB DDL, statistics updation fails due to lock wait
timeout and calls push_warning_printf() to generate warnings
but then returns success, causing the SQL layer
to attempt calling set_ok_status() when the diagnostics area
is already set.
Solution:
=========
By temporarily setting abort_on_warning to false around operations
that prevents warning to error escalation and restore the original
setting after calling HA_EXTRA_END_ALTER_COPY for alter operation.
don't copy field default values and check constraints in
CREATE ... SELECT.
CREATE ... SELECT means a table is created from a *result set*
not from some other table.
For backward compatibility, though, let's keep copying constant
default values and the "compressed" attribute.
Caused by optimization done in 2e2b2a0469.
Cannot use lookup_handler in default branch of locate_dup_record() as
InnoDB update depends on positioned record and update is done in table
main handler.
The patch reverts some non-pure changes done by 2e2b2a0469 to
original logic from 72429cad. There was no long_unique_table condition
to init search on table->file, so we get into default branch with long
unique and table->file search uninitialized.
ha_rnd_init_with_error() on demand for HA_DUPLICATE_POS branch was
original logic as well.
More info: 2e2b2a0469 reverts 5e345281e3, but it seems to be OK as
MDEV-3888 test case passes. mysql-5.6.13 has the original code with
HA_WHOLE_KEY as well.
Timestamp-versioned row deletion was exposed to a collisional problem: if
current timestamp wasn't changed, then a sequence of row delete+insert could
get a duplication error. A row delete would find another conflicting history row
and return an error.
This is true both for REPLACE and DELETE statements, however in REPLACE, the
"optimized" path is usually taken, especially in the tests. There, delete+insert
is substituted for a single versioned row update. In the end, both paths end up
as ha_update_row + ha_write_row.
The solution is to handle a history collision somehow.
From the design perspective, the user shouldn't experience history rows loss,
unless there's a technical limitation.
To the contrary, trxid-based changes should never generate history for the same
transaction, see MDEV-15427.
If two operations on the same row happened too quickly, so that they happen at
the same timestamp, the history row shouldn't be lost. We can still write a
history row, though it'll have row_start == row_end.
We cannot store more than one such historical row, as this will violate the
unique constraint on row_end. So we will have to phisically delete the row if
the history row is already available.
In this commit:
1. Improve TABLE::delete_row to handle the history collision: if an update
results with a duplicate error, delete a row for real.
2. use TABLE::delete_row in a non-optimistic path of REPLACE, where the
system-versioned case now belongs entirely.
We had a protection against it, by allowing versioned delete if:
trx->id != table->vers_start_id()
For replace this check fails: replace calls ha_delete_row(record[2]), but
table->vers_start_id() returns the value from record[0], which is irrelevant.
The same problem hits Field::is_max, which may have checked the wrong record.
Fix:
* Refactor Field::is_max to optionally accept a pointer as an argument.
* Refactor vers_start_id and vers_end_id to always accept a pointer to the
record. there is a difference with is_max is that is_max accepts the pointer to
the
field data, rather than to the record.
Method val_int() would be too effortful to refactor to accept the argument, so
instead the value in record is fetched directly, like it is done in
Field_longlong.
See also MDEV-30046.
Idempotent write_row works same as REPLACE: if there is a duplicating
record in the table, then it will be deleted and re-inserted, with the
same update optimization.
The code in Rows:log_event::write_row was basically copy-pasted from
write_record.
What's done:
REPLACE operation was unified across replication and sql. It is now
representred as a Write_record class, that holds the whole state, and allows
re-using some resources in between the row writes.
Replace, IODKU and single insert implementations are split across different
methods, reluting in a much cleaner code.
The entry point is preserved as a single Write_record::write_record() call.
The implementation to call is chosen on the constructor stage.
This allowed several optimizations to be done:
1. The table key list is not iterated for every row. We find last unique key in
the order of checking once and preserve it across the rows. See last_uniq_key().
2. ib_handler::referenced_by_foreign_key acquires a global lock. This call was
done per row as well. Not all the table config that allows optimized replace is
folded into a single boolean field can_optimize. All the fields to check are
even stored in a single register on a 64-bit platform.
3. DUP_REPLACE and DUP_UPDATE cases now have one less level of indirection
4. modified_non_trans_tables is checked and set only when it's really needed.
5. Obsolete bitmap manipulations are removed.
Also:
* Unify replace initialization step across implementations:
add prepare_for_replace and finalize_replace
* alloca is removed in favor of mem_root allocation. This memory is reused
across the rows.
* An rpl-related callback is added to the replace branch, meaning that an extra
check is made per row replace even for the common case. It can be avoided with
templates if considered a problem.
Timestamp-versioned row deletion was exposed to a collisional problem: if
current timestamp wasn't changed, then a sequence of row delete+insert could
get a duplication error. A row delete would find another conflicting history row
and return an error.
This is true both for REPLACE and DELETE statements, however in REPLACE, the
"optimized" path is usually taken, especially in the tests. There, delete+insert
is substituted for a single versioned row update. In the end, both paths end up
as ha_update_row + ha_write_row.
The solution is to handle a history collision somehow.
From the design perspective, the user shouldn't experience history rows loss,
unless there's a technical limitation.
To the contrary, trxid-based changes should never generate history for the same
transaction, see MDEV-15427.
If two operations on the same row happened too quickly, so that they happen at
the same timestamp, the history row shouldn't be lost. We can still write a
history row, though it'll have row_start == row_end.
We cannot store more than one such historical row, as this will violate the
unique constraint on row_end. So we will have to phisically delete the row if
the history row is already available.
In this commit:
1. Improve TABLE::delete_row to handle the history collision: if an update
results with a duplicate error, delete a row for real.
2. use TABLE::delete_row in a non-optimistic path of REPLACE, where the
system-versioned case now belongs entirely.
We had a protection against it, by allowing versioned delete if:
trx->id != table->vers_start_id()
For replace this check fails: replace calls ha_delete_row(record[2]), but
table->vers_start_id() returns the value from record[0], which is irrelevant.
The same problem hits Field::is_max, which may have checked the wrong record.
Fix:
* Refactor Field::is_max to optionally accept a pointer as an argument.
* Refactor vers_start_id and vers_end_id to always accept a pointer to the
record. there is a difference with is_max is that is_max accepts the pointer to
the
field data, rather than to the record.
Method val_int() would be too effortful to refactor to accept the argument, so
instead the value in record is fetched directly, like it is done in
Field_longlong.
