VladPolskiy/MariaDB-open-code
1
0
Fork 0
mirror of https://github.com/MariaDB/server.git synced 2026-01-13 06:03:12 +00:00
Code Issues Packages Projects Releases Wiki Activity
Files
bb-main-atomic
MariaDB-open-code/sql/sql_partition_admin.cc
Monty 7b54e04965 MDEV-25292 Atomic CREATE OR REPLACE TABLE 
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.

InnoDB changes
  Added ha_innobase::can_be_renamed_to_backup() to check if
  a table with foreign keys can be renamed.

Aria changes:
- Fixed issue in Aria engine with CREATE + locked tables
  that data was not properly commited in some cases in
  case of crashes.

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).

Known issues:
- InnoDB tables with foreign key definitions are not fully supported
  with atomic create and replace:
  - ha_innobase::can_be_renamed_to_backup() can detect some cases
    where InnoDB does not support renaming table with foreign key
    constraints.  In this case MariaDB will drop the old table before
    creating the new one.
    The detected cases are:
    - The new and old table is using the same foreign key constraint
      name.
    - The old table has self referencing constraints.
  - If the old and new table uses the same name for a constraint the
    create of the new table will fail. The orignal table will be
    restored in this case.
  - The above issues will be fixed in a future commit.
- 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

InnoDB related changes:
- ha_innodb::rename_table() does not handle foreign key constraint
  when renaming an normal table to internal tempory tables. This
  causes problems for CREATE OR REPLACE as the old constraints causes
  failure when creating a new table with the same constraints.
  This is fixed inside InnoDB by not threating tempfiles (#sql-create-..),
  created as part of CREATE OR REPLACE, as temporary files.
- In ha_innobase::delete_table(), ignore checking of constraints when
  dropping a #sql-create temporary table.
- In tablename_to_filename() and filename_to_tablename(), don't do
  filename conversion for internal temporary tables (#sql-...)

Other things:
- 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.

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.
2026-01-06 17:50:43 +02:00

1053 lines
36 KiB
C++
Raw Permalink Blame History

/* Copyright (c) 2010, 2013, Oracle and/or its affiliates. All rights reserved.
Copyright (c) 2014, SkySQL Ab.
Copyright (c) 2016, 2018, MariaDB Corporation.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; version 2 of the License.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program; if not, write to the Free Software
Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1335 USA */
#include "mariadb.h"
#include "sql_parse.h" // check_one_table_access
// check_merge_table_access
// check_one_table_access
#include "sql_table.h" // mysql_alter_table, etc.
#include "sql_cmd.h" // Sql_cmd
#include "sql_alter.h" // Sql_cmd_alter_table
#include "sql_partition.h" // struct partition_info, etc.
#include "debug_sync.h" // DEBUG_SYNC
#include "sql_truncate.h" // mysql_truncate_table,
// Sql_cmd_truncate_table
#include "sql_admin.h" // Analyze/Check/.._table_statement
#include "sql_partition_admin.h" // Alter_table_*_partition
#ifdef WITH_PARTITION_STORAGE_ENGINE
#include "ha_partition.h" // ha_partition
#endif
#include "sql_base.h" // open_and_lock_tables
#include "ddl_log.h"
#include "wsrep_mysqld.h"
#ifndef WITH_PARTITION_STORAGE_ENGINE
bool Sql_cmd_partition_unsupported::execute(THD *)
{
DBUG_ENTER("Sql_cmd_partition_unsupported::execute");
/* error, partitioning support not compiled in... */
my_error(ER_FEATURE_DISABLED, MYF(0), "partitioning",
"--with-plugin-partition");
DBUG_RETURN(TRUE);
}
#else
static bool return_with_logging(THD *thd)
{
if (thd->slave_thread &&
write_bin_log_with_if_exists(thd, true, false, true) > 0)
return(true);
my_ok(thd);
return(false);
}
bool Sql_cmd_alter_table_exchange_partition::execute(THD *thd)
{
/* Moved from mysql_execute_command */
LEX *lex= thd->lex;
/* first SELECT_LEX (have special meaning for many of non-SELECTcommands) */
SELECT_LEX *select_lex= lex->first_select_lex();
/* first table of first SELECT_LEX */
TABLE_LIST *first_table= (TABLE_LIST*) select_lex->table_list.first;
/*
Code in mysql_alter_table() may modify its HA_CREATE_INFO argument,
so we have to use a copy of this structure to make execution
prepared statement- safe. A shallow copy is enough as no memory
referenced from this structure will be modified.
@todo move these into constructor...
*/
IF_DBUG(HA_CREATE_INFO create_info(lex->create_info);,)
Alter_info alter_info(lex->alter_info, thd->mem_root);
privilege_t priv_needed(ALTER_ACL | DROP_ACL | INSERT_ACL | CREATE_ACL);
DBUG_ENTER("Sql_cmd_alter_table_exchange_partition::execute");
if (unlikely(thd->is_fatal_error))
{
/* out of memory creating a copy of alter_info */
DBUG_RETURN(TRUE);
}
/* Must be set in the parser */
DBUG_ASSERT(select_lex->db.str);
/* also check the table to be exchanged with the partition */
DBUG_ASSERT(alter_info.partition_flags & ALTER_PARTITION_EXCHANGE);
if (unlikely(check_access(thd, priv_needed, first_table->db.str,
&first_table->grant.privilege,
&first_table->grant.m_internal,
0, 0)) ||
unlikely(check_access(thd, priv_needed, first_table->next_local->db.str,
&first_table->next_local->grant.privilege,
&first_table->next_local->grant.m_internal,
0, 0)))
DBUG_RETURN(TRUE);
if (unlikely(check_grant(thd, priv_needed, first_table, FALSE, UINT_MAX,
FALSE)))
DBUG_RETURN(TRUE);
/* Not allowed with EXCHANGE PARTITION */
DBUG_ASSERT(!create_info.data_file_name && !create_info.index_file_name);
WSREP_TO_ISOLATION_BEGIN_WRTCHK(NULL, NULL, first_table);
DBUG_RETURN(exchange_partition(thd, first_table, &alter_info));
#ifdef WITH_WSREP
wsrep_error_label:
/* handle errors in TO_ISOLATION here */
DBUG_RETURN(true);
#endif /* WITH_WSREP */
}
/**
@brief Checks that the tables will be able to be used for EXCHANGE PARTITION.
@param table Non partitioned table.
@param part_table Partitioned table.
@retval FALSE if OK, otherwise error is reported and TRUE is returned.
*/
static bool check_exchange_partition(TABLE *table, TABLE *part_table)
{
DBUG_ENTER("check_exchange_partition");
/* Both tables must exist */
if (unlikely(!part_table || !table))
{
my_error(ER_CHECK_NO_SUCH_TABLE, MYF(0));
DBUG_RETURN(TRUE);
}
/* The first table must be partitioned, and the second must not */
if (unlikely(!part_table->part_info))
{
my_error(ER_PARTITION_MGMT_ON_NONPARTITIONED, MYF(0));
DBUG_RETURN(TRUE);
}
if (unlikely(table->part_info))
{
my_error(ER_PARTITION_EXCHANGE_PART_TABLE, MYF(0),
table->s->table_name.str);
DBUG_RETURN(TRUE);
}
if (unlikely(part_table->file->ht != partition_hton))
{
/*
Only allowed on partitioned tables throughout the generic ha_partition
handler, i.e not yet for native partitioning.
*/
my_error(ER_PARTITION_MGMT_ON_NONPARTITIONED, MYF(0));
DBUG_RETURN(TRUE);
}
if (unlikely(table->file->ht != part_table->part_info->default_engine_type))
{
my_error(ER_MIX_HANDLER_ERROR, MYF(0));
DBUG_RETURN(TRUE);
}
/* Verify that table is not tmp table, partitioned tables cannot be tmp. */
if (unlikely(table->s->tmp_table != NO_TMP_TABLE))
{
my_error(ER_PARTITION_EXCHANGE_TEMP_TABLE, MYF(0),
table->s->table_name.str);
DBUG_RETURN(TRUE);
}
/* The table cannot have foreign keys constraints or be referenced */
if (unlikely(!table->file->can_switch_engines()))
{
my_error(ER_PARTITION_EXCHANGE_FOREIGN_KEY, MYF(0),
table->s->table_name.str);
DBUG_RETURN(TRUE);
}
DBUG_RETURN(FALSE);
}
/**
@brief Compare table structure/options between a non partitioned table
and a specific partition of a partitioned table.
@param thd Thread object.
@param table Non partitioned table.
@param part_table Partitioned table.
@param part_elem Partition element to use for partition specific compare.
*/
bool compare_table_with_partition(THD *thd, TABLE *table, TABLE *part_table,
partition_element *part_elem, uint part_id)
{
HA_CREATE_INFO table_create_info;
Table_specification_st part_create_info;
Alter_info part_alter_info;
Alter_table_ctx part_alter_ctx; // Not used
DBUG_ENTER("compare_table_with_partition");
bool metadata_equal= false;
part_create_info.init();
table_create_info.init();
update_create_info_from_table(&table_create_info, table);
/* get the current auto_increment value */
table->file->update_create_info(&table_create_info);
/* mark all columns used, since they are used when preparing the new table */
part_table->use_all_columns();
table->use_all_columns();
if (unlikely(mysql_prepare_alter_table(thd, part_table, &part_create_info,
&part_alter_info, &part_alter_ctx)))
{
my_error(ER_TABLES_DIFFERENT_METADATA, MYF(0));
DBUG_RETURN(TRUE);
}
/* db_type is not set in prepare_alter_table */
part_create_info.db_type= part_table->part_info->default_engine_type;
((ha_partition*)(part_table->file))->update_part_create_info(&part_create_info, part_id);
/*
Since we exchange the partition with the table, allow exchanging
auto_increment value as well.
*/
part_create_info.auto_increment_value=
table_create_info.auto_increment_value;
/* Check compatible row_types and set create_info accordingly. */
{
enum row_type part_row_type= part_table->file->get_row_type();
enum row_type table_row_type= table->file->get_row_type();
if (part_row_type != table_row_type)
{
my_error(ER_PARTITION_EXCHANGE_DIFFERENT_OPTION, MYF(0),
"ROW_FORMAT");
DBUG_RETURN(true);
}
part_create_info.row_type= table->s->row_type;
}
/*
NOTE: ha_blackhole does not support check_if_compatible_data,
so this always fail for blackhole tables.
ha_myisam compares pointers to verify that DATA/INDEX DIRECTORY is
the same, so any table using data/index_file_name will fail.
*/
if (mysql_compare_tables(table, &part_alter_info, &part_create_info,
&metadata_equal))
{
my_error(ER_TABLES_DIFFERENT_METADATA, MYF(0));
DBUG_RETURN(TRUE);
}
DEBUG_SYNC(thd, "swap_partition_after_compare_tables");
if (!metadata_equal)
{
my_error(ER_TABLES_DIFFERENT_METADATA, MYF(0));
DBUG_RETURN(TRUE);
}
if (table->s->db_create_options != part_table->s->db_create_options)
{
my_error(ER_TABLES_DIFFERENT_METADATA, MYF(0));
DBUG_RETURN(TRUE);
}
DBUG_ASSERT(table->s->db_options_in_use ==
part_table->s->db_options_in_use);
if (table_create_info.avg_row_length != part_create_info.avg_row_length)
{
my_error(ER_PARTITION_EXCHANGE_DIFFERENT_OPTION, MYF(0),
"AVG_ROW_LENGTH");
DBUG_RETURN(TRUE);
}
if (table_create_info.table_options != part_create_info.table_options)
{
my_error(ER_PARTITION_EXCHANGE_DIFFERENT_OPTION, MYF(0),
"TABLE OPTION");
DBUG_RETURN(TRUE);
}
if (table->s->table_charset != part_table->s->table_charset)
{
my_error(ER_PARTITION_EXCHANGE_DIFFERENT_OPTION, MYF(0),
"CHARACTER SET");
DBUG_RETURN(TRUE);
}
/*
NOTE: We do not support update of frm-file, i.e. change
max/min_rows, data/index_file_name etc.
The workaround is to use REORGANIZE PARTITION to rewrite
the frm file and then use EXCHANGE PARTITION when they are the same.
*/
if (part_elem && compare_partition_options(&table_create_info, part_elem))
DBUG_RETURN(TRUE);
DBUG_RETURN(FALSE);
}
/**
@brief Exchange partition/table with ddl log.
@details How to handle a crash in the middle of the rename (break on error):
1) register in ddl_log that we are going to exchange swap_table with part.
2) do the first rename (swap_table -> tmp-name) and sync the ddl_log.
3) do the second rename (part -> swap_table) and sync the ddl_log.
4) do the last rename (tmp-name -> part).
5) mark the entry done.
Recover by:
5) is done, All completed. Nothing to recover.
4) is done see 3). (No mark or sync in the ddl_log...)
3) is done -> try rename part -> tmp-name (ignore failure) goto 2).
2) is done -> try rename swap_table -> part (ignore failure) goto 1).
1) is done -> try rename tmp-name -> swap_table (ignore failure).
before 1) Nothing to recover...
@param thd Thread handle
@param name name of table/partition 1 (to be exchanged with 2)
@param from_name name of table/partition 2 (to be exchanged with 1)
@param tmp_name temporary name to use while exchanging
@param ht handlerton of the table/partitions
@return Operation status
@retval TRUE Error
@retval FALSE Success
@note ha_heap always succeeds in rename (since it is created upon usage).
This is OK when to recover from a crash since all heap are empty and the
recover is done early in the startup of the server (right before
read_init_file which can populate the tables).
And if no crash we can trust the syncs in the ddl_log.
What about if the rename is put into a background thread? That will cause
corruption and is avoided by the exlusive metadata lock.
*/
static bool exchange_name_with_ddl_log(THD *thd,
const char *name,
const char *from_name,
const char *tmp_name,
handlerton *ht)
{
DDL_LOG_ENTRY exchange_entry;
DDL_LOG_MEMORY_ENTRY *log_entry= NULL;
DDL_LOG_MEMORY_ENTRY *exec_log_entry= NULL;
bool error= TRUE;
bool error_set= FALSE;
handler *file= NULL;
DBUG_ENTER("exchange_name_with_ddl_log");
if (unlikely(!(file= get_new_handler(NULL, thd->mem_root, ht))))
DBUG_RETURN(TRUE);
/* prepare the action entry */
bzero(&exchange_entry, sizeof(exchange_entry));
exchange_entry.entry_type= DDL_LOG_ENTRY_CODE;
exchange_entry.action_type= DDL_LOG_EXCHANGE_ACTION;
lex_string_set(&exchange_entry.name, name);
lex_string_set(&exchange_entry.from_name, from_name);
lex_string_set(&exchange_entry.tmp_name, tmp_name);
lex_string_set(&exchange_entry.handler_name,
ha_resolve_storage_engine_name(ht));
exchange_entry.phase= EXCH_PHASE_NAME_TO_TEMP;
mysql_mutex_lock(&LOCK_gdl);
/*
write to the ddl log what to do by:
1) write the action entry (i.e. which names to be exchanged)
2) write the execution entry with a link to the action entry
*/
DBUG_EXECUTE_IF("exchange_partition_fail_1", goto err_no_action_written;);
DBUG_EXECUTE_IF("exchange_partition_abort_1", DBUG_SUICIDE(););
if (unlikely(ddl_log_write_entry(&exchange_entry, &log_entry)))
goto err_no_action_written;
DBUG_EXECUTE_IF("exchange_partition_fail_2", goto err_no_execute_written;);
DBUG_EXECUTE_IF("exchange_partition_abort_2", DBUG_SUICIDE(););
if (unlikely(ddl_log_write_execute_entry(log_entry->entry_pos, 0,
&exec_log_entry)))
goto err_no_execute_written;
/* ddl_log is written and synced */
mysql_mutex_unlock(&LOCK_gdl);
/*
Execute the name exchange.
Do one rename, increase the phase, update the action entry and sync.
In case of errors in the ddl_log we must fail and let the ddl_log try
to revert the changes, since otherwise it could revert the command after
we sent OK to the client.
*/
/* call rename table from table to tmp-name */
DBUG_EXECUTE_IF("exchange_partition_fail_3",
my_error(ER_ERROR_ON_RENAME, MYF(0), name, tmp_name, 0);
error_set= TRUE;
goto err_rename;);
DBUG_EXECUTE_IF("exchange_partition_abort_3", DBUG_SUICIDE(););
if (unlikely(file->ha_rename_table(name, tmp_name)))
{
my_error(ER_ERROR_ON_RENAME, MYF(0), name, tmp_name, my_errno);
error_set= TRUE;
goto err_rename;
}
DBUG_EXECUTE_IF("exchange_partition_fail_4", goto err_rename;);
DBUG_EXECUTE_IF("exchange_partition_abort_4", DBUG_SUICIDE(););
if (unlikely(ddl_log_increment_phase(log_entry->entry_pos)))
goto err_rename;
/* call rename table from partition to table */
DBUG_EXECUTE_IF("exchange_partition_fail_5",
my_error(ER_ERROR_ON_RENAME, MYF(0), from_name, name, 0);
error_set= TRUE;
goto err_rename;);
DBUG_EXECUTE_IF("exchange_partition_abort_5", DBUG_SUICIDE(););
if (unlikely(file->ha_rename_table(from_name, name)))
{
my_error(ER_ERROR_ON_RENAME, MYF(0), from_name, name, my_errno);
error_set= TRUE;
goto err_rename;
}
DBUG_EXECUTE_IF("exchange_partition_fail_6", goto err_rename;);
DBUG_EXECUTE_IF("exchange_partition_abort_6", DBUG_SUICIDE(););
if (unlikely(ddl_log_increment_phase(log_entry->entry_pos)))
goto err_rename;
/* call rename table from tmp-nam to partition */
DBUG_EXECUTE_IF("exchange_partition_fail_7",
my_error(ER_ERROR_ON_RENAME, MYF(0), tmp_name, from_name, 0);
error_set= TRUE;
goto err_rename;);
DBUG_EXECUTE_IF("exchange_partition_abort_7", DBUG_SUICIDE(););
if (unlikely(file->ha_rename_table(tmp_name, from_name)))
{
my_error(ER_ERROR_ON_RENAME, MYF(0), tmp_name, from_name, my_errno);
error_set= TRUE;
goto err_rename;
}
DBUG_EXECUTE_IF("exchange_partition_fail_8", goto err_rename;);
DBUG_EXECUTE_IF("exchange_partition_abort_8", DBUG_SUICIDE(););
if (unlikely(ddl_log_increment_phase(log_entry->entry_pos)))
goto err_rename;
/* The exchange is complete and ddl_log is deactivated */
DBUG_EXECUTE_IF("exchange_partition_fail_9", goto err_rename;);
DBUG_EXECUTE_IF("exchange_partition_abort_9", DBUG_SUICIDE(););
/* all OK */
error= FALSE;
delete file;
DBUG_RETURN(error);
err_rename:
/*
Nothing to do if any of these commands fails :( the commands itselfs
will log to the error log about the failures...
*/
/* execute the ddl log entry to revert the renames */
(void) ddl_log_execute_entry(current_thd, log_entry->entry_pos);
mysql_mutex_lock(&LOCK_gdl);
/* mark the execute log entry done */
(void) ddl_log_disable_execute_entry(&exec_log_entry);
/* release the execute log entry */
(void) ddl_log_release_memory_entry(exec_log_entry);
err_no_execute_written:
/* release the action log entry */
(void) ddl_log_release_memory_entry(log_entry);
err_no_action_written:
mysql_mutex_unlock(&LOCK_gdl);
delete file;
if (!error_set)
my_error(ER_DDL_LOG_ERROR, MYF(0));
DBUG_RETURN(error);
}
/**
@brief Swap places between a partition and a table.
@details Verify that the tables are compatible (same engine, definition etc),
verify that all rows in the table will fit in the partition,
if all OK, rename table to tmp name, rename partition to table
and finally rename tmp name to partition.
1) Take upgradable mdl, open tables and then lock them (inited in parse)
2) Verify that metadata matches
3) verify data
4) Upgrade to exclusive mdl for both tables
5) Rename table <-> partition
6) Rely on close_thread_tables to release mdl and table locks
@param thd Thread handle
@param table_list Table where the partition exists as first table,
Table to swap with the partition as second table
@param alter_info Contains partition name to swap
@note This is a DDL operation so triggers will not be used.
*/
bool Sql_cmd_alter_table_exchange_partition::
exchange_partition(THD *thd, TABLE_LIST *table_list, Alter_info *alter_info)
{
TABLE *part_table, *swap_table;
TABLE_LIST *swap_table_list;
handlerton *table_hton;
partition_element *part_elem;
char temp_name[FN_REFLEN+1];
char part_file_name[2*FN_REFLEN+1];
char swap_file_name[FN_REFLEN+1];
char temp_file_name[FN_REFLEN+1];
char part_table_name[NAME_LEN + 1];
char part_db[NAME_LEN + 1];
char swap_table_name[NAME_LEN + 1];
char swap_db[NAME_LEN + 1];
uchar part_tabledef_version[MY_UUID_SIZE];
uchar swap_tabledef_version[MY_UUID_SIZE];
backup_log_info ddl_log;
bzero(&ddl_log, sizeof(ddl_log));
uint swap_part_id;
uint part_file_name_len;
Alter_table_prelocking_strategy alter_prelocking_strategy;
MDL_ticket *swap_table_mdl_ticket= NULL;
MDL_ticket *part_table_mdl_ticket= NULL;
uint table_counter;
bool error= TRUE, force_if_exists= 0;
ulonglong save_option_bits= thd->variables.option_bits;
DBUG_ENTER("mysql_exchange_partition");
DBUG_ASSERT(alter_info->partition_flags & ALTER_PARTITION_EXCHANGE);
/* Don't allow to exchange with log table */
swap_table_list= table_list->next_local;
if (check_if_log_table(swap_table_list, FALSE, "ALTER PARTITION"))
DBUG_RETURN(TRUE);
/*
Currently no MDL lock that allows both read and write and is upgradeable
to exclusive, so leave the lock type to TL_WRITE_ALLOW_READ also on the
partitioned table.
TODO: add MDL lock that allows both read and write and is upgradable to
exclusive lock. This would allow to continue using the partitioned table
also with update/insert/delete while the verification of the swap table
is running.
*/
/*
NOTE: It is not possible to exchange a crashed partition/table since
we need some info from the engine, which we can only access after open,
to be able to verify the structure/metadata.
*/
table_list->mdl_request.set_type(MDL_SHARED_NO_WRITE);
if (unlikely(open_tables(thd, &table_list, &table_counter, 0,
&alter_prelocking_strategy)))
{
if (thd->lex->if_exists() &&
thd->get_stmt_da()->sql_errno() == ER_NO_SUCH_TABLE)
{
/*
ALTER TABLE IF EXISTS was used on not existing table
We have to log the query on a slave as the table may be a shared one
from the master and we need to ensure that the next slave can see
the statement as this slave may not have the table shared
*/
thd->clear_error();
if (thd->slave_thread &&
write_bin_log(thd, true, thd->query(), thd->query_length()))
DBUG_RETURN(true);
my_ok(thd);
DBUG_RETURN(false);
}
DBUG_RETURN(true);
}
part_table= table_list->table;
swap_table= swap_table_list->table;
/* Don't allow to exchange with a VIEW */
if (unlikely(swap_table_list->view))
{
my_error(ER_WRONG_OBJECT, MYF(0), table_list->db.str,
swap_table_list->table_name.str, "BASE TABLE");
DBUG_RETURN(TRUE);
}
if (unlikely(check_exchange_partition(swap_table, part_table)))
DBUG_RETURN(TRUE);
if (part_table->file->check_if_updates_are_ignored("ALTER"))
DBUG_RETURN(return_with_logging(thd));
/* Add IF EXISTS to binlog if shared table */
if (part_table->file->partition_ht()->flags &
HTON_TABLE_MAY_NOT_EXIST_ON_SLAVE)
force_if_exists= 1;
ddl_log.org_table.str= part_table_name;
DBUG_ASSERT(part_table->s->table_name.length <= NAME_LEN);
ddl_log.org_table.length= part_table->s->table_name.length;
strmake(part_table_name, part_table->s->table_name.str, NAME_LEN);
ddl_log.org_database.str= part_db;
DBUG_ASSERT(part_table->s->db.length <= NAME_LEN);
ddl_log.org_database.length= part_table->s->db.length;
strmake(part_db, part_table->s->db.str, NAME_LEN);
ddl_log.new_table.str= swap_table_name;
DBUG_ASSERT(swap_table->s->table_name.length <= NAME_LEN);
ddl_log.new_table.length= swap_table->s->table_name.length;
strmake(swap_table_name, swap_table->s->table_name.str, NAME_LEN);
ddl_log.new_database.str= swap_db;
DBUG_ASSERT(swap_table->s->db.length <= NAME_LEN);
ddl_log.new_database.length= swap_table->s->db.length;
strmake(swap_db, swap_table->s->db.str, NAME_LEN);
memcpy(part_tabledef_version, part_table->s->tabledef_version.str,
MY_UUID_SIZE);
ddl_log.org_table_id.str= part_tabledef_version;
ddl_log.org_table_id.length= MY_UUID_SIZE;
memcpy(swap_tabledef_version, swap_table->s->tabledef_version.str,
MY_UUID_SIZE);
ddl_log.new_table_id.str= swap_tabledef_version;
ddl_log.new_table_id.length= MY_UUID_SIZE;
/* set lock pruning on first table */
const Lex_cstring_strlen partition_name=
Lex_cstring_strlen(alter_info->partition_names.head());
if (unlikely(table_list->table->part_info->
set_named_partition_bitmap(partition_name.str,
partition_name.length)))
DBUG_RETURN(true);
if (unlikely(lock_tables(thd, table_list, table_counter, 0)))
DBUG_RETURN(true);
table_hton= swap_table->file->ht;
THD_STAGE_INFO(thd, stage_verifying_table);
/* Will append the partition name later in part_info->get_part_elem() */
part_file_name_len= build_table_filename(part_file_name,
sizeof(part_file_name),
table_list->db.str,
table_list->table_name.str,
"", 0);
build_table_filename(swap_file_name,
sizeof(swap_file_name),
swap_table_list->db.str,
swap_table_list->table_name.str,
"", 0);
LEX_STRING tmp= { temp_name, sizeof(temp_name) };
make_tmp_table_name(thd, &tmp, "exchange");
build_table_filename(temp_file_name, sizeof(temp_file_name),
table_list->next_local->db.str,
temp_name, "", FN_IS_TMP);
if (unlikely(!(part_elem=
part_table->part_info->get_part_elem(partition_name,
part_file_name +
part_file_name_len,
sizeof(part_file_name) -
part_file_name_len,
&swap_part_id))))
{
DBUG_RETURN(TRUE);
}
if (unlikely(swap_part_id == NOT_A_PARTITION_ID))
{
DBUG_ASSERT(part_table->part_info->is_sub_partitioned());
my_error(ER_PARTITION_INSTEAD_OF_SUBPARTITION, MYF(0));
DBUG_RETURN(TRUE);
}
if (unlikely(compare_table_with_partition(thd, swap_table, part_table,
part_elem,
swap_part_id)))
DBUG_RETURN(TRUE);
/* Table and partition has same structure/options, OK to exchange */
thd_proc_info(thd, "Verifying data with partition");
if (unlikely(verify_data_with_partition(swap_table, part_table,
swap_part_id)))
DBUG_RETURN(TRUE);
/*
Get exclusive mdl lock on both tables, alway the non partitioned table
first. Remember the tickets for downgrading locks later.
*/
swap_table_mdl_ticket= swap_table->mdl_ticket;
part_table_mdl_ticket= part_table->mdl_ticket;
/*
No need to set used_partitions to only propagate
HA_EXTRA_PREPARE_FOR_RENAME to one part since no built in engine uses
that flag. And the action would probably be to force close all other
instances which is what we are doing any way.
*/
if (wait_while_table_is_used(thd, swap_table, HA_EXTRA_PREPARE_FOR_RENAME) ||
wait_while_table_is_used(thd, part_table, HA_EXTRA_PREPARE_FOR_RENAME))
goto err;
DEBUG_SYNC(thd, "swap_partition_after_wait");
close_all_tables_for_name(thd, swap_table->s, HA_EXTRA_NOT_USED, NULL);
close_all_tables_for_name(thd, part_table->s, HA_EXTRA_NOT_USED, NULL);
DEBUG_SYNC(thd, "swap_partition_before_rename");
if (unlikely(exchange_name_with_ddl_log(thd, swap_file_name, part_file_name,
temp_file_name, table_hton)))
goto err;
/*
Reopen tables under LOCK TABLES. Ignore the return value for now. It's
better to keep master/slave in consistent state. Alternative would be to
try to revert the exchange operation and issue error.
*/
(void) thd->locked_tables_list.reopen_tables(thd, false);
if (force_if_exists)
thd->variables.option_bits|= OPTION_IF_EXISTS;
if (unlikely((error= write_bin_log(thd, TRUE, thd->query(),
thd->query_length()))))
{
/*
The error is reported in write_bin_log().
We try to revert to make it easier to keep the master/slave in sync.
*/
(void) exchange_name_with_ddl_log(thd, part_file_name, swap_file_name,
temp_file_name, table_hton);
}
else
{
ddl_log.query= { C_STRING_WITH_LEN("EXCHANGE_PARTITION") };
ddl_log.org_partitioned= true;
ddl_log.new_partitioned= false;
ddl_log.org_storage_engine_name= *hton_name(table_hton);
ddl_log.new_storage_engine_name= *hton_name(table_hton);
backup_log_ddl(&ddl_log);
}
thd->variables.option_bits= save_option_bits;
err:
if (thd->locked_tables_mode)
{
if (swap_table_mdl_ticket)
swap_table_mdl_ticket->downgrade_lock(MDL_SHARED_NO_READ_WRITE);
if (part_table_mdl_ticket)
part_table_mdl_ticket->downgrade_lock(MDL_SHARED_NO_READ_WRITE);
}
if (unlikely(!error))
my_ok(thd);
// For query cache
table_list->table= NULL;
table_list->next_local->table= NULL;
query_cache_invalidate3(thd, table_list, FALSE);
DBUG_RETURN(error);
}
bool Sql_cmd_alter_table_analyze_partition::execute(THD *thd)
{
bool res;
DBUG_ENTER("Sql_cmd_alter_table_analyze_partition::execute");
/*
Flag that it is an ALTER command which administrates partitions, used
by ha_partition
*/
thd->lex->alter_info.partition_flags|= ALTER_PARTITION_ADMIN;
res= Sql_cmd_analyze_table::execute(thd);
DBUG_RETURN(res);
}
bool Sql_cmd_alter_table_check_partition::execute(THD *thd)
{
bool res;
DBUG_ENTER("Sql_cmd_alter_table_check_partition::execute");
/*
Flag that it is an ALTER command which administrates partitions, used
by ha_partition
*/
thd->lex->alter_info.partition_flags|= ALTER_PARTITION_ADMIN;
res= Sql_cmd_check_table::execute(thd);
DBUG_RETURN(res);
}
bool Sql_cmd_alter_table_optimize_partition::execute(THD *thd)
{
bool res;
DBUG_ENTER("Alter_table_optimize_partition_statement::execute");
/*
Flag that it is an ALTER command which administrates partitions, used
by ha_partition
*/
thd->lex->alter_info.partition_flags|= ALTER_PARTITION_ADMIN;
res= Sql_cmd_optimize_table::execute(thd);
DBUG_RETURN(res);
}
bool Sql_cmd_alter_table_repair_partition::execute(THD *thd)
{
bool res;
DBUG_ENTER("Sql_cmd_alter_table_repair_partition::execute");
/*
Flag that it is an ALTER command which administrates partitions, used
by ha_partition
*/
thd->lex->alter_info.partition_flags|= ALTER_PARTITION_ADMIN;
res= Sql_cmd_repair_table::execute(thd);
DBUG_RETURN(res);
}
bool Sql_cmd_alter_table_truncate_partition::execute(THD *thd)
{
int error;
ha_partition *partition;
ulong timeout= thd->variables.lock_wait_timeout;
TABLE_LIST *first_table= thd->lex->first_select_lex()->table_list.first;
Alter_info *alter_info= &thd->lex->alter_info;
uint table_counter, i;
List<String> partition_names_list;
bool binlog_stmt, force_if_exists= 0;
DBUG_ENTER("Sql_cmd_alter_table_truncate_partition::execute");
/*
Flag that it is an ALTER command which administrates partitions, used
by ha_partition.
*/
thd->lex->alter_info.partition_flags|= (ALTER_PARTITION_ADMIN |
ALTER_PARTITION_TRUNCATE);
/* Fix the lock types (not the same as ordinary ALTER TABLE). */
first_table->lock_type= TL_WRITE;
first_table->mdl_request.set_type(MDL_EXCLUSIVE);
/*
Check table permissions and open it with a exclusive lock.
Ensure it is a partitioned table and finally, upcast the
handler and invoke the partition truncate method. Lastly,
write the statement to the binary log if necessary.
*/
if (check_one_table_access(thd, DROP_ACL, first_table))
DBUG_RETURN(TRUE);
#ifdef WITH_WSREP
if (WSREP(thd) &&
(!thd->is_current_stmt_binlog_format_row() ||
!thd->find_temporary_table(first_table)) &&
wsrep_to_isolation_begin(
thd, first_table->db.str, first_table->table_name.str, NULL)
)
{
WSREP_WARN("ALTER TABLE TRUNCATE PARTITION isolation failure");
DBUG_RETURN(TRUE);
}
#endif /* WITH_WSREP */
if (open_tables(thd, &first_table, &table_counter, 0))
{
if (thd->lex->if_exists() &&
thd->get_stmt_da()->sql_errno() == ER_NO_SUCH_TABLE)
{
/*
ALTER TABLE IF EXISTS was used on not existing table
We have to log the query on a slave as the table may be a shared one
from the master and we need to ensure that the next slave can see
the statement as this slave may not have the table shared
*/
thd->clear_error();
DBUG_RETURN(return_with_logging(thd));
}
DBUG_RETURN(TRUE);
}
if (!first_table->table || first_table->view)
{
my_error(ER_PARTITION_MGMT_ON_NONPARTITIONED, MYF(0));
DBUG_RETURN(TRUE);
}
if (first_table->table->file->check_if_updates_are_ignored("ALTER"))
DBUG_RETURN(return_with_logging(thd));
if (first_table->table->s->db_type() != partition_hton)
{
my_error(ER_PARTITION_MGMT_ON_NONPARTITIONED, MYF(0));
DBUG_RETURN(TRUE);
}
if (first_table->table->file->partition_ht()->flags &
HTON_TABLE_MAY_NOT_EXIST_ON_SLAVE)
force_if_exists= 1;
/*
Prune all, but named partitions,
to avoid excessive calls to external_lock().
*/
List_iterator<const char> partition_names_it(alter_info->partition_names);
uint num_names= alter_info->partition_names.elements;
for (i= 0; i < num_names; i++)
{
const char *partition_name= partition_names_it++;
String *str_partition_name= new (thd->mem_root)
String(partition_name, strlen(partition_name), system_charset_info);
if (!str_partition_name)
DBUG_RETURN(true);
partition_names_list.push_back(str_partition_name, thd->mem_root);
}
if (first_table->table->
part_info->set_partition_bitmaps(&partition_names_list))
DBUG_RETURN(true);
if (lock_tables(thd, first_table, table_counter, 0))
DBUG_RETURN(true);
/*
Under locked table modes this might still not be an exclusive
lock. Hence, upgrade the lock since the handler truncate method
mandates an exclusive metadata lock.
*/
MDL_ticket *ticket= first_table->table->mdl_ticket;
if (thd->mdl_context.upgrade_shared_lock(ticket, MDL_EXCLUSIVE, timeout))
DBUG_RETURN(TRUE);
first_table->table->s->tdc->flush(thd, true);
partition= (ha_partition*) first_table->table->file;
/* Invoke the handler method responsible for truncating the partition. */
if (unlikely(error= partition->truncate_partition(alter_info,
&binlog_stmt)))
partition->print_error(error, MYF(0));
/*
All effects of a truncate operation are committed even if the
operation fails. Thus, the query must be written to the binary
log. The exception is a unimplemented truncate method or failure
before any call to handler::truncate() is done.
Also, it is logged in statement format, regardless of the binlog format.
Since we've changed data within the table, we also have to invalidate
the query cache for it.
*/
if (likely(error != HA_ERR_WRONG_COMMAND))
{
ulonglong save_option_bits= thd->variables.option_bits;
if (force_if_exists)
thd->variables.option_bits|= OPTION_IF_EXISTS;
query_cache_invalidate3(thd, first_table, FALSE);
if (binlog_stmt)
error|= write_bin_log(thd, !error, thd->query(), thd->query_length());
thd->variables.option_bits= save_option_bits;
}
/*
A locked table ticket was upgraded to a exclusive lock. After the
the query has been written to the binary log, downgrade the lock
to a shared one.
*/
if (thd->locked_tables_mode)
ticket->downgrade_lock(MDL_SHARED_NO_READ_WRITE);
if (likely(!error))
my_ok(thd);
// Invalidate query cache
DBUG_ASSERT(!first_table->next_local);
query_cache_invalidate3(thd, first_table, FALSE);
DBUG_RETURN(error);
}
/**
Move a table specified in the CONVERT TABLE <table_name> TO PARTITION ...
to the new partition.
@param lpt A structure containing parameters regarding to the statement
ALTER TABLE ... TO PARTITION ...
@param part_file_name a file name of the partition being added
@return false on success, true on error
*/
bool alter_partition_convert_in(ALTER_PARTITION_PARAM_TYPE *lpt)
{
char part_file_name[2*FN_REFLEN+1];
THD *thd= lpt->thd;
const char *path= lpt->table_list->table->s->path.str;
TABLE_LIST *table_from= lpt->table_list->next_local;
const Lex_ident_partition &partition_name=
thd->lex->part_info->curr_part_elem->partition_name;
if (create_partition_name(part_file_name, sizeof(part_file_name), path,
partition_name, NORMAL_PART_NAME, false))
return true;
char from_file_name[FN_REFLEN+1];
build_table_filename(from_file_name, sizeof(from_file_name),
table_from->db.str, table_from->table_name.str, "", 0);
handler *file= get_new_handler(nullptr, thd->mem_root,
table_from->table->file->ht);
if (unlikely(!file))
return true;
close_all_tables_for_name(thd, table_from->table->s,
HA_EXTRA_PREPARE_FOR_RENAME, nullptr);
bool res= file->ha_rename_table(from_file_name, part_file_name);
if (res)
my_error(ER_ERROR_ON_RENAME, MYF(0), from_file_name,
part_file_name, my_errno);
delete file;
return res;
}
#endif /* WITH_PARTITION_STORAGE_ENGINE */
Reference in New Issue View Git Blame Copy Permalink