Commit eab2be0a authored by Mattias Jonsson's avatar Mattias Jonsson

Bug#47343: InnoDB fails to clean-up after lock wait timeout on

           REORGANIZE PARTITION

There were several problems which lead to this this,
all related to bad error handling.

1) There was several bugs preventing the ddl-log to be used for
   cleaning up created files on error.

2) The error handling after the copy partition rows did not close
   and unlock the tables, resulting in deletion of partitions
   which were in use, which lead InnoDB to put the partition to
   drop in a background queue.

sql/ha_partition.cc:
  Bug#47343: InnoDB fails to clean-up after lock wait timeout on
             REORGANIZE PARTITION
  
  Better error handling, if partition has been created/opened/locked
  then make sure it is unlocked and closed before returning error.
  The delete of the newly created partition is handled by the ddl-log.
sql/sql_parse.cc:
  Bug#47343: InnoDB fails to clean-up after lock wait timeout on
             REORGANIZE PARTITION
  
  Fix a bug found when experimenting, thd could really be NULL here,
  as mentioned in the function header.
sql/sql_partition.cc:
  Bug#47343: InnoDB fails to clean-up after lock wait timeout on
             REORGANIZE PARTITION
  
  Used the correct .frm shadow name to put into the ddl-log.
  Really use the ddl-log to handle errors.
sql/sql_table.cc:
  Bug#47343: InnoDB fails to clean-up after lock wait timeout on
             REORGANIZE PARTITION
  
  Fixes of the ddl-log when used as error recovery (no crash).
  When executing an entry from memory (not read from disk)
  the name_len was not set correctly.
parent 0305aea8
...@@ -274,3 +274,47 @@ CREATE TABLE t1 (a INT) ENGINE=InnoDB ...@@ -274,3 +274,47 @@ CREATE TABLE t1 (a INT) ENGINE=InnoDB
PARTITION BY list(a) (PARTITION p1 VALUES IN (1)); PARTITION BY list(a) (PARTITION p1 VALUES IN (1));
CREATE INDEX i1 ON t1 (a); CREATE INDEX i1 ON t1 (a);
DROP TABLE t1; DROP TABLE t1;
#
# Bug#47343: InnoDB fails to clean-up after lock wait timeout on
# REORGANIZE PARTITION
#
CREATE TABLE t1 (
a INT,
b DATE NOT NULL,
PRIMARY KEY (a, b)
) ENGINE=InnoDB
PARTITION BY RANGE (a) (
PARTITION pMAX VALUES LESS THAN MAXVALUE
) ;
INSERT INTO t1 VALUES (1, '2001-01-01'), (2, '2002-02-02'), (3, '2003-03-03');
START TRANSACTION;
SELECT * FROM t1 FOR UPDATE;
a b
1 2001-01-01
2 2002-02-02
3 2003-03-03
# Connection con1
ALTER TABLE t1 REORGANIZE PARTITION pMAX INTO
(PARTITION p3 VALUES LESS THAN (3),
PARTITION pMAX VALUES LESS THAN MAXVALUE);
ERROR HY000: Lock wait timeout exceeded; try restarting transaction
SHOW WARNINGS;
Level Code Message
Error 1205 Lock wait timeout exceeded; try restarting transaction
ALTER TABLE t1 REORGANIZE PARTITION pMAX INTO
(PARTITION p3 VALUES LESS THAN (3),
PARTITION pMAX VALUES LESS THAN MAXVALUE);
ERROR HY000: Lock wait timeout exceeded; try restarting transaction
SHOW WARNINGS;
Level Code Message
Error 1205 Lock wait timeout exceeded; try restarting transaction
t1.frm
t1.par
# Connection default
SELECT * FROM t1;
a b
1 2001-01-01
2 2002-02-02
3 2003-03-03
COMMIT;
DROP TABLE t1;
--innodb_lock_wait_timeout=1
...@@ -5,6 +5,8 @@ ...@@ -5,6 +5,8 @@
drop table if exists t1; drop table if exists t1;
--enable_warnings --enable_warnings
let $MYSQLD_DATADIR= `SELECT @@datadir`;
# #
# Bug#47029: Crash when reorganize partition with subpartition # Bug#47029: Crash when reorganize partition with subpartition
# #
...@@ -296,6 +298,47 @@ CREATE TABLE t1 (a INT) ENGINE=InnoDB ...@@ -296,6 +298,47 @@ CREATE TABLE t1 (a INT) ENGINE=InnoDB
PARTITION BY list(a) (PARTITION p1 VALUES IN (1)); PARTITION BY list(a) (PARTITION p1 VALUES IN (1));
CREATE INDEX i1 ON t1 (a); CREATE INDEX i1 ON t1 (a);
DROP TABLE t1; DROP TABLE t1;
let $MYSQLD_DATADIR= `SELECT @@datadir`;
# Before the fix it should show extra file like #sql-2405_2.par # Before the fix it should show extra file like #sql-2405_2.par
--list_files $MYSQLD_DATADIR/test/ * --list_files $MYSQLD_DATADIR/test/ *
--echo #
--echo # Bug#47343: InnoDB fails to clean-up after lock wait timeout on
--echo # REORGANIZE PARTITION
--echo #
CREATE TABLE t1 (
a INT,
b DATE NOT NULL,
PRIMARY KEY (a, b)
) ENGINE=InnoDB
PARTITION BY RANGE (a) (
PARTITION pMAX VALUES LESS THAN MAXVALUE
) ;
INSERT INTO t1 VALUES (1, '2001-01-01'), (2, '2002-02-02'), (3, '2003-03-03');
START TRANSACTION;
SELECT * FROM t1 FOR UPDATE;
connect (con1, localhost, root,,);
--echo # Connection con1
--error ER_LOCK_WAIT_TIMEOUT
ALTER TABLE t1 REORGANIZE PARTITION pMAX INTO
(PARTITION p3 VALUES LESS THAN (3),
PARTITION pMAX VALUES LESS THAN MAXVALUE);
SHOW WARNINGS;
--error ER_LOCK_WAIT_TIMEOUT
ALTER TABLE t1 REORGANIZE PARTITION pMAX INTO
(PARTITION p3 VALUES LESS THAN (3),
PARTITION pMAX VALUES LESS THAN MAXVALUE);
SHOW WARNINGS;
#Contents of the 'test' database directory:
--list_files $MYSQLD_DATADIR/test
disconnect con1;
connection default;
--echo # Connection default
SELECT * FROM t1;
COMMIT;
DROP TABLE t1;
...@@ -1215,17 +1215,28 @@ int ha_partition::prepare_new_partition(TABLE *tbl, ...@@ -1215,17 +1215,28 @@ int ha_partition::prepare_new_partition(TABLE *tbl,
partition_element *p_elem) partition_element *p_elem)
{ {
int error; int error;
bool create_flag= FALSE;
DBUG_ENTER("prepare_new_partition"); DBUG_ENTER("prepare_new_partition");
if ((error= set_up_table_before_create(tbl, part_name, create_info, if ((error= set_up_table_before_create(tbl, part_name, create_info,
0, p_elem))) 0, p_elem)))
goto error; goto error_create;
if ((error= file->ha_create(part_name, tbl, create_info))) if ((error= file->ha_create(part_name, tbl, create_info)))
goto error; {
create_flag= TRUE; /*
Added for safety, InnoDB reports HA_ERR_FOUND_DUPP_KEY
if the table/partition already exists.
If we return that error code, then print_error would try to
get_dup_key on a non-existing partition.
So return a more reasonable error code.
*/
if (error == HA_ERR_FOUND_DUPP_KEY)
error= HA_ERR_TABLE_EXIST;
goto error_create;
}
DBUG_PRINT("info", ("partition %s created", part_name));
if ((error= file->ha_open(tbl, part_name, m_mode, m_open_test_lock))) if ((error= file->ha_open(tbl, part_name, m_mode, m_open_test_lock)))
goto error; goto error_open;
DBUG_PRINT("info", ("partition %s opened", part_name));
/* /*
Note: if you plan to add another call that may return failure, Note: if you plan to add another call that may return failure,
better to do it before external_lock() as cleanup_new_partition() better to do it before external_lock() as cleanup_new_partition()
...@@ -1233,12 +1244,15 @@ int ha_partition::prepare_new_partition(TABLE *tbl, ...@@ -1233,12 +1244,15 @@ int ha_partition::prepare_new_partition(TABLE *tbl,
Otherwise see description for cleanup_new_partition(). Otherwise see description for cleanup_new_partition().
*/ */
if ((error= file->ha_external_lock(ha_thd(), m_lock_type))) if ((error= file->ha_external_lock(ha_thd(), m_lock_type)))
goto error; goto error_external_lock;
DBUG_PRINT("info", ("partition %s external locked", part_name));
DBUG_RETURN(0); DBUG_RETURN(0);
error: error_external_lock:
if (create_flag) VOID(file->close());
VOID(file->ha_delete_table(part_name)); error_open:
VOID(file->ha_delete_table(part_name));
error_create:
DBUG_RETURN(error); DBUG_RETURN(error);
} }
...@@ -1272,19 +1286,23 @@ error: ...@@ -1272,19 +1286,23 @@ error:
void ha_partition::cleanup_new_partition(uint part_count) void ha_partition::cleanup_new_partition(uint part_count)
{ {
handler **save_m_file= m_file;
DBUG_ENTER("ha_partition::cleanup_new_partition"); DBUG_ENTER("ha_partition::cleanup_new_partition");
if (m_added_file && m_added_file[0]) if (m_added_file)
{ {
m_file= m_added_file; THD *thd= ha_thd();
m_added_file= NULL; handler **file= m_added_file;
while ((part_count > 0) && (*file))
{
(*file)->ha_external_lock(thd, F_UNLCK);
(*file)->close();
external_lock(ha_thd(), F_UNLCK); /* Leave the (*file)->ha_delete_table(part_name) to the ddl-log */
/* delete_table also needed, a bit more complex */
close();
m_file= save_m_file; file++;
part_count--;
}
m_added_file= NULL;
} }
DBUG_VOID_RETURN; DBUG_VOID_RETURN;
} }
...@@ -1590,7 +1608,15 @@ int ha_partition::change_partitions(HA_CREATE_INFO *create_info, ...@@ -1590,7 +1608,15 @@ int ha_partition::change_partitions(HA_CREATE_INFO *create_info,
part_elem->part_state= PART_TO_BE_DROPPED; part_elem->part_state= PART_TO_BE_DROPPED;
} }
m_new_file= new_file_array; m_new_file= new_file_array;
DBUG_RETURN(copy_partitions(copied, deleted)); if ((error= copy_partitions(copied, deleted)))
{
/*
Close and unlock the new temporary partitions.
They will later be deleted through the ddl-log.
*/
cleanup_new_partition(part_count);
}
DBUG_RETURN(error);
} }
......
...@@ -6954,7 +6954,7 @@ bool reload_acl_and_cache(THD *thd, ulong options, TABLE_LIST *tables, ...@@ -6954,7 +6954,7 @@ bool reload_acl_and_cache(THD *thd, ulong options, TABLE_LIST *tables,
/* /*
If the query was killed then this function must fail. If the query was killed then this function must fail.
*/ */
return result || thd->killed; return result || (thd ? thd->killed : 0);
} }
......
...@@ -5708,8 +5708,7 @@ static bool write_log_drop_partition(ALTER_PARTITION_PARAM_TYPE *lpt) ...@@ -5708,8 +5708,7 @@ static bool write_log_drop_partition(ALTER_PARTITION_PARAM_TYPE *lpt)
part_info->first_log_entry= NULL; part_info->first_log_entry= NULL;
build_table_filename(path, sizeof(path) - 1, lpt->db, build_table_filename(path, sizeof(path) - 1, lpt->db,
lpt->table_name, "", 0); lpt->table_name, "", 0);
build_table_filename(tmp_path, sizeof(tmp_path) - 1, lpt->db, build_table_shadow_filename(tmp_path, sizeof(tmp_path) - 1, lpt);
lpt->table_name, "#", 0);
pthread_mutex_lock(&LOCK_gdl); pthread_mutex_lock(&LOCK_gdl);
if (write_log_dropped_partitions(lpt, &next_entry, (const char*)path, if (write_log_dropped_partitions(lpt, &next_entry, (const char*)path,
FALSE)) FALSE))
...@@ -5765,8 +5764,7 @@ static bool write_log_add_change_partition(ALTER_PARTITION_PARAM_TYPE *lpt) ...@@ -5765,8 +5764,7 @@ static bool write_log_add_change_partition(ALTER_PARTITION_PARAM_TYPE *lpt)
build_table_filename(path, sizeof(path) - 1, lpt->db, build_table_filename(path, sizeof(path) - 1, lpt->db,
lpt->table_name, "", 0); lpt->table_name, "", 0);
build_table_filename(tmp_path, sizeof(tmp_path) - 1, lpt->db, build_table_shadow_filename(tmp_path, sizeof(tmp_path) - 1, lpt);
lpt->table_name, "#", 0);
pthread_mutex_lock(&LOCK_gdl); pthread_mutex_lock(&LOCK_gdl);
if (write_log_dropped_partitions(lpt, &next_entry, (const char*)path, if (write_log_dropped_partitions(lpt, &next_entry, (const char*)path,
FALSE)) FALSE))
...@@ -5991,7 +5989,7 @@ void handle_alter_part_error(ALTER_PARTITION_PARAM_TYPE *lpt, ...@@ -5991,7 +5989,7 @@ void handle_alter_part_error(ALTER_PARTITION_PARAM_TYPE *lpt,
partition_info *part_info= lpt->part_info; partition_info *part_info= lpt->part_info;
DBUG_ENTER("handle_alter_part_error"); DBUG_ENTER("handle_alter_part_error");
if (!part_info->first_log_entry && if (part_info->first_log_entry &&
execute_ddl_log_entry(current_thd, execute_ddl_log_entry(current_thd,
part_info->first_log_entry->entry_pos)) part_info->first_log_entry->entry_pos))
{ {
......
...@@ -647,7 +647,7 @@ static bool read_ddl_log_file_entry(uint entry_no) ...@@ -647,7 +647,7 @@ static bool read_ddl_log_file_entry(uint entry_no)
Write one entry from ddl log file Write one entry from ddl log file
SYNOPSIS SYNOPSIS
write_ddl_log_file_entry() write_ddl_log_file_entry()
entry_no Entry number to read entry_no Entry number to write
RETURN VALUES RETURN VALUES
TRUE Error TRUE Error
FALSE Success FALSE Success
...@@ -748,10 +748,10 @@ static uint read_ddl_log_header() ...@@ -748,10 +748,10 @@ static uint read_ddl_log_header()
else else
successful_open= TRUE; successful_open= TRUE;
} }
entry_no= uint4korr(&file_entry_buf[DDL_LOG_NUM_ENTRY_POS]);
global_ddl_log.name_len= uint4korr(&file_entry_buf[DDL_LOG_NAME_LEN_POS]);
if (successful_open) if (successful_open)
{ {
entry_no= uint4korr(&file_entry_buf[DDL_LOG_NUM_ENTRY_POS]);
global_ddl_log.name_len= uint4korr(&file_entry_buf[DDL_LOG_NAME_LEN_POS]);
global_ddl_log.io_size= uint4korr(&file_entry_buf[DDL_LOG_IO_SIZE_POS]); global_ddl_log.io_size= uint4korr(&file_entry_buf[DDL_LOG_IO_SIZE_POS]);
DBUG_ASSERT(global_ddl_log.io_size <= DBUG_ASSERT(global_ddl_log.io_size <=
sizeof(global_ddl_log.file_entry_buf)); sizeof(global_ddl_log.file_entry_buf));
...@@ -832,6 +832,7 @@ static bool init_ddl_log() ...@@ -832,6 +832,7 @@ static bool init_ddl_log()
goto end; goto end;
global_ddl_log.io_size= IO_SIZE; global_ddl_log.io_size= IO_SIZE;
global_ddl_log.name_len= FN_LEN;
create_ddl_log_file_name(file_name); create_ddl_log_file_name(file_name);
if ((global_ddl_log.file_id= my_create(file_name, if ((global_ddl_log.file_id= my_create(file_name,
CREATE_MODE, CREATE_MODE,
...@@ -884,6 +885,13 @@ static int execute_ddl_log_action(THD *thd, DDL_LOG_ENTRY *ddl_log_entry) ...@@ -884,6 +885,13 @@ static int execute_ddl_log_action(THD *thd, DDL_LOG_ENTRY *ddl_log_entry)
{ {
DBUG_RETURN(FALSE); DBUG_RETURN(FALSE);
} }
DBUG_PRINT("ddl_log",
("execute type %c next %u name '%s' from_name '%s' handler '%s'",
ddl_log_entry->action_type,
ddl_log_entry->next_entry,
ddl_log_entry->name,
ddl_log_entry->from_name,
ddl_log_entry->handler_name));
handler_name.str= (char*)ddl_log_entry->handler_name; handler_name.str= (char*)ddl_log_entry->handler_name;
handler_name.length= strlen(ddl_log_entry->handler_name); handler_name.length= strlen(ddl_log_entry->handler_name);
init_sql_alloc(&mem_root, TABLE_ALLOC_BLOCK_SIZE, 0); init_sql_alloc(&mem_root, TABLE_ALLOC_BLOCK_SIZE, 0);
...@@ -1091,6 +1099,15 @@ bool write_ddl_log_entry(DDL_LOG_ENTRY *ddl_log_entry, ...@@ -1091,6 +1099,15 @@ bool write_ddl_log_entry(DDL_LOG_ENTRY *ddl_log_entry,
DBUG_RETURN(TRUE); DBUG_RETURN(TRUE);
} }
error= FALSE; error= FALSE;
DBUG_PRINT("ddl_log",
("write type %c next %u name '%s' from_name '%s' handler '%s'",
(char) global_ddl_log.file_entry_buf[DDL_LOG_ACTION_TYPE_POS],
ddl_log_entry->next_entry,
(char*) &global_ddl_log.file_entry_buf[DDL_LOG_NAME_POS],
(char*) &global_ddl_log.file_entry_buf[DDL_LOG_NAME_POS
+ FN_LEN],
(char*) &global_ddl_log.file_entry_buf[DDL_LOG_NAME_POS
+ (2*FN_LEN)]));
if (write_ddl_log_file_entry((*active_entry)->entry_pos)) if (write_ddl_log_file_entry((*active_entry)->entry_pos))
{ {
error= TRUE; error= TRUE;
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment