mariadb/mysql-test/include/handler.inc

1525 lines
40 KiB
PHP
Raw Normal View History

# include/handler.inc
#
# The variables
# $engine_type -- storage engine to be tested
# $other_engine_type -- storage engine <> $engine_type
# $other_handler_engine_type -- storage engine <> $engine_type, if possible
# 1. $other_handler_engine_type must support handler
# 2. $other_handler_engine_type must point to an all
# time available storage engine
# 2006-08 MySQL 5.1 MyISAM and MEMORY only
# have to be set before sourcing this script.
-- source include/not_embedded.inc
#
# test of HANDLER ...
#
# Last update:
# 2006-07-31 ML test refactored (MySQL 5.1)
# code of t/handler.test and t/innodb_handler.test united
# main testing code put into include/handler.inc
#
eval SET SESSION STORAGE_ENGINE = $engine_type;
--disable_warnings
drop table if exists t1,t3,t4,t5;
--enable_warnings
2001-04-08 12:56:12 +02:00
create table t1 (a int, b char(10), key a(a), key b(a,b));
insert into t1 values
2001-04-08 12:56:12 +02:00
(17,"ddd"),(18,"eee"),(19,"fff"),(19,"yyy"),
2001-04-13 16:18:44 +02:00
(14,"aaa"),(15,"bbb"),(16,"ccc"),(16,"xxx"),
(20,"ggg"),(21,"hhh"),(22,"iii");
2001-04-08 12:56:12 +02:00
handler t1 open as t2;
2003-01-10 17:36:59 +02:00
-- error 1064
handler t2 read a=(SELECT 1);
2001-04-08 12:56:12 +02:00
handler t2 read a first;
handler t2 read a next;
handler t2 read a next;
handler t2 read a prev;
handler t2 read a last;
handler t2 read a prev;
handler t2 read a prev;
handler t2 read a first;
handler t2 read a prev;
handler t2 read a last;
handler t2 read a prev;
handler t2 read a next;
handler t2 read a next;
handler t2 read a=(15);
handler t2 read a=(16);
--error 1070
handler t2 read a=(19,"fff");
2001-04-08 12:56:12 +02:00
handler t2 read b=(19,"fff");
handler t2 read b=(19,"yyy");
handler t2 read b=(19);
--error 1109
handler t1 read a last;
2001-04-08 12:56:12 +02:00
handler t2 read a=(11);
handler t2 read a>=(11);
handler t2 read a=(18);
handler t2 read a>=(18);
handler t2 read a>(18);
handler t2 read a<=(18);
handler t2 read a<(18);
2001-04-09 15:37:19 +02:00
handler t2 read a first limit 5;
handler t2 read a next limit 3;
handler t2 read a prev limit 10;
handler t2 read a>=(16) limit 4;
2001-04-13 16:18:44 +02:00
handler t2 read a>=(16) limit 2,2;
2001-04-09 15:37:19 +02:00
handler t2 read a last limit 3;
2001-04-13 16:18:44 +02:00
handler t2 read a=(19);
handler t2 read a=(19) where b="yyy";
handler t2 read first;
handler t2 read next;
handler t2 read next;
--error 1064
handler t2 read last;
2001-04-08 12:56:12 +02:00
handler t2 close;
handler t1 open;
handler t1 read a next; # this used to crash as a bug#5373
handler t1 read a next;
handler t1 close;
handler t1 open;
handler t1 read a prev; # this used to crash as a bug#5373
handler t1 read a prev;
handler t1 close;
handler t1 open as t2;
handler t2 read first;
eval alter table t1 engine = $engine_type;
--error 1109
handler t2 read first;
#
2003-02-22 18:22:39 +01:00
# DROP TABLE / ALTER TABLE
#
handler t1 open as t2;
drop table t1;
create table t1 (a int);
insert into t1 values (17);
--error 1109
handler t2 read first;
2003-02-22 18:22:39 +01:00
handler t1 open as t2;
eval alter table t1 engine=$other_engine_type;
2003-02-22 18:22:39 +01:00
--error 1109
handler t2 read first;
drop table t1;
2002-01-03 14:31:54 +00:00
#
# Test case for the bug #787
#
create table t1 (a int);
insert into t1 values (1),(2),(3),(4),(5),(6);
delete from t1 limit 2;
handler t1 open;
handler t1 read first;
handler t1 read first limit 1,1;
handler t1 read first limit 2,2;
delete from t1 limit 3;
handler t1 read first;
drop table t1;
#
# Test for #751
#
create table t1(a int, index(a));
insert into t1 values (1), (2), (3);
handler t1 open;
--error 1054
handler t1 read a=(W);
--error 1210
handler t1 read a=(a);
drop table t1;
#
# BUG#2304
#
create table t1 (a char(5));
insert into t1 values ("Ok");
handler t1 open as t;
handler t read first;
use mysql;
handler t read first;
handler t close;
handler test.t1 open as t;
handler t read first;
handler t close;
use test;
drop table t1;
2004-05-18 22:59:43 +04:00
#
# BUG#3649
#
create table t1 ( a int, b int, INDEX a (a) );
insert into t1 values (1,2), (2,1);
handler t1 open;
handler t1 read a=(1) where b=2;
handler t1 read a=(1) where b=3;
handler t1 read a=(1) where b=1;
handler t1 close;
drop table t1;
#
# Check if two database names beginning the same are seen as different.
#
# This database begins like the usual 'test' database.
#
--disable_warnings
drop database if exists test_test;
--enable_warnings
create database test_test;
use test_test;
create table t1(table_id char(20) primary key);
insert into t1 values ('test_test.t1');
insert into t1 values ('');
handler t1 open;
handler t1 read first limit 9;
create table t2(table_id char(20) primary key);
insert into t2 values ('test_test.t2');
insert into t2 values ('');
handler t2 open;
handler t2 read first limit 9;
#
# This is the usual 'test' database.
#
use test;
--disable_warnings
drop table if exists t1;
--enable_warnings
create table t1(table_id char(20) primary key);
insert into t1 values ('test.t1');
insert into t1 values ('');
--error 1066
handler t1 open;
#
# Check accesibility of all the tables.
#
use test;
--error 1064
handler test.t1 read first limit 9;
--error 1064
handler test_test.t1 read first limit 9;
handler t1 read first limit 9;
--error 1064
handler test_test.t2 read first limit 9;
handler t2 read first limit 9;
#
# Cleanup.
#
--error 1064
handler test_test.t1 close;
handler t1 close;
drop table test_test.t1;
--error 1064
handler test_test.t2 close;
handler t2 close;
drop table test_test.t2;
drop database test_test;
#
use test;
--error 1064
handler test.t1 close;
--error 1109
handler t1 close;
drop table test.t1;
#
# BUG#4335
#
--disable_warnings
drop database if exists test_test;
drop table if exists t1;
drop table if exists t2;
drop table if exists t3;
--enable_warnings
create database test_test;
use test_test;
create table t1 (c1 char(20));
insert into t1 values ('test_test.t1');
create table t3 (c1 char(20));
insert into t3 values ('test_test.t3');
handler t1 open;
handler t1 read first limit 9;
handler t1 open h1;
handler h1 read first limit 9;
use test;
create table t1 (c1 char(20));
create table t2 (c1 char(20));
create table t3 (c1 char(20));
insert into t1 values ('t1');
insert into t2 values ('t2');
insert into t3 values ('t3');
--error 1066
handler t1 open;
--error 1066
handler t2 open t1;
--error 1066
handler t3 open t1;
handler t1 read first limit 9;
--error 1064
handler test.t1 close;
--error 1066
handler test.t1 open h1;
--error 1066
handler test_test.t1 open h1;
handler test_test.t3 open h3;
handler test.t1 open h2;
handler t1 read first limit 9;
handler h1 read first limit 9;
handler h2 read first limit 9;
handler h3 read first limit 9;
handler h2 read first limit 9;
--error 1064
handler test.h1 close;
handler t1 close;
handler h1 close;
handler h2 close;
--error 1109
handler t1 read first limit 9;
--error 1109
handler h1 read first limit 9;
--error 1109
handler h2 read first limit 9;
handler h3 read first limit 9;
handler h3 read first limit 9;
use test_test;
handler h3 read first limit 9;
--error 1064
handler test.h3 read first limit 9;
handler h3 close;
use test;
drop table t3;
drop table t2;
drop table t1;
drop database test_test;
#
# Test if fix for BUG#4286 correctly closes handler tables.
#
create table t1 (c1 char(20));
insert into t1 values ("t1");
handler t1 open as h1;
handler h1 read first limit 9;
create table t2 (c1 char(20));
insert into t2 values ("t2");
handler t2 open as h2;
handler h2 read first limit 9;
create table t3 (c1 char(20));
insert into t3 values ("t3");
handler t3 open as h3;
handler h3 read first limit 9;
create table t4 (c1 char(20));
insert into t4 values ("t4");
handler t4 open as h4;
handler h4 read first limit 9;
create table t5 (c1 char(20));
insert into t5 values ("t5");
handler t5 open as h5;
handler h5 read first limit 9;
# close first
eval alter table t1 engine=$other_handler_engine_type;
--error 1109
handler h1 read first limit 9;
handler h2 read first limit 9;
handler h3 read first limit 9;
handler h4 read first limit 9;
handler h5 read first limit 9;
# close last
eval alter table t5 engine=$other_handler_engine_type;
--error 1109
handler h1 read first limit 9;
handler h2 read first limit 9;
handler h3 read first limit 9;
handler h4 read first limit 9;
--error 1109
handler h5 read first limit 9;
# close middle
eval alter table t3 engine=$other_handler_engine_type;
--error 1109
handler h1 read first limit 9;
handler h2 read first limit 9;
--error 1109
handler h3 read first limit 9;
handler h4 read first limit 9;
--error 1109
handler h5 read first limit 9;
handler h2 close;
handler h4 close;
# close all depending handler tables
handler t1 open as h1_1;
handler t1 open as h1_2;
handler t1 open as h1_3;
handler h1_1 read first limit 9;
handler h1_2 read first limit 9;
handler h1_3 read first limit 9;
eval alter table t1 engine=$engine_type;
--error 1109
handler h1_1 read first limit 9;
--error 1109
handler h1_2 read first limit 9;
--error 1109
handler h1_3 read first limit 9;
drop table t1;
drop table t2;
drop table t3;
drop table t4;
drop table t5;
#
# Bug#14397 - OPTIMIZE TABLE with an open HANDLER causes a crash
#
create table t1 (c1 int);
insert into t1 values (1);
# client 1
handler t1 open;
handler t1 read first;
# client 2
connect (con2,localhost,root,,);
connection con2;
--exec echo send the below to another connection, do not wait for the result
send optimize table t1;
--sleep 1
# client 1
--exec echo proceed with the normal connection
connection default;
handler t1 read next;
handler t1 close;
# client 2
--exec echo read the result from the other connection
connection con2;
reap;
# client 1
--exec echo proceed with the normal connection
connection default;
drop table t1;
CREATE TABLE t1 ( no1 smallint(5) NOT NULL default '0', no2 int(10) NOT NULL default '0', PRIMARY KEY (no1,no2));
INSERT INTO t1 VALUES (1,274),(1,275),(2,6),(2,8),(4,1),(4,2);
HANDLER t1 OPEN;
HANDLER t1 READ `primary` = (1, 1000);
HANDLER t1 READ `primary` PREV;
DROP TABLE t1;
# End of 4.1 tests
#
# Addendum to Bug#14397 - OPTIMIZE TABLE with an open HANDLER causes a crash
# Show that DROP TABLE can no longer deadlock against
# FLUSH TABLES WITH READ LOCK. This is a 5.0 issue.
#
create table t1 (c1 int);
insert into t1 values (14397);
flush tables with read lock;
# The thread with the global read lock cannot drop the table itself:
--error 1223
drop table t1;
#
# client 2
# We need a second connection to try the drop.
# The drop waits for the global read lock to go away.
# Without the addendum fix it locked LOCK_open before entering the wait loop.
connection con2;
--exec echo send the below to another connection, do not wait for the result
send drop table t1;
--sleep 1
#
# client 1
# Now we need something that wants LOCK_open. A simple table access which
# opens the table does the trick.
--exec echo proceed with the normal connection
connection default;
# This would hang on LOCK_open without the 5.0 addendum fix.
select * from t1;
# Release the read lock. This should make the DROP go through.
unlock tables;
#
# client 2
# Read the result of the drop command.
connection con2;
--exec echo read the result from the other connection
reap;
#
# client 1
# Now back to normal operation. The table should not exist any more.
--exec echo proceed with the normal connection
connection default;
--error 1146
select * from t1;
# Just to be sure and not confuse the next test case writer.
drop table if exists t1;
#
# Bug#25856 - HANDLER table OPEN in one connection lock DROP TABLE in another one
#
--disable_warnings
drop table if exists t1;
--enable_warnings
2007-08-20 17:50:42 -06:00
eval create table t1 (a int) ENGINE=$other_engine_type;
--echo --> client 2
connection con2;
--error 1031
handler t1 open;
--echo --> client 1
connection default;
drop table t1;
disconnect con2;
#
# Bug#30632 HANDLER read failure causes hang
#
--disable_warnings
drop table if exists t1;
--enable_warnings
create table t1 (a int);
handler t1 open as t1_alias;
--error 1176
handler t1_alias read a next;
--error 1054
handler t1_alias READ a next where inexistent > 0;
--error 1176
handler t1_alias read a next;
--error 1054
handler t1_alias READ a next where inexistent > 0;
handler t1_alias close;
drop table t1;
2007-10-10 19:06:53 -03:00
#
# Bug#21587 FLUSH TABLES causes server crash when used with HANDLER statements
#
--disable_warnings
drop table if exists t1,t2;
--enable_warnings
create table t1 (c1 int);
create table t2 (c1 int);
insert into t1 values (1);
insert into t2 values (2);
--echo connection: default
handler t1 open;
handler t1 read first;
connect (flush,localhost,root,,);
connection flush;
--echo connection: flush
--send flush tables;
Initial import of WL#3726 "DDL locking for all metadata objects". Backport of: ------------------------------------------------------------ revno: 2630.4.1 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Fri 2008-05-23 17:54:03 +0400 message: WL#3726 "DDL locking for all metadata objects". After review fixes in progress. ------------------------------------------------------------ This is the first patch in series. It transforms the metadata locking subsystem to use a dedicated module (mdl.h,cc). No significant changes in the locking protocol. The import passes the test suite with the exception of deprecated/removed 6.0 features, and MERGE tables. The latter are subject to a fix by WL#4144. Unfortunately, the original changeset comments got lost in a merge, thus this import has its own (largely insufficient) comments. This patch fixes Bug#25144 "replication / binlog with view breaks". Warning: this patch introduces an incompatible change: Under LOCK TABLES, it's no longer possible to FLUSH a table that was not locked for WRITE. Under LOCK TABLES, it's no longer possible to DROP a table or VIEW that was not locked for WRITE. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.2 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 14:03:45 +0400 message: WL#3726 "DDL locking for all metadata objects". After review fixes in progress. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.3 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 14:08:51 +0400 message: WL#3726 "DDL locking for all metadata objects" Fixed failing Windows builds by adding mdl.cc to the lists of files needed to build server/libmysqld on Windows. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.4 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 21:57:58 +0400 message: WL#3726 "DDL locking for all metadata objects". Fix for assert failures in kill.test which occured when one tried to kill ALTER TABLE statement on merge table while it was waiting in wait_while_table_is_used() for other connections to close this table. These assert failures stemmed from the fact that cleanup code in this case assumed that temporary table representing new version of table was open with adding to THD::temporary_tables list while code which were opening this temporary table wasn't always fulfilling this. This patch changes code that opens new version of table to always do this linking in. It also streamlines cleanup process for cases when error occurs while we have new version of table open. ****** WL#3726 "DDL locking for all metadata objects" Add libmysqld/mdl.cc to .bzrignore. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.6 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sun 2008-05-25 00:33:22 +0400 message: WL#3726 "DDL locking for all metadata objects". Addition to the fix of assert failures in kill.test caused by changes for this worklog. Make sure we close the new table only once.
2009-11-30 18:55:03 +03:00
connect (waiter,localhost,root,,);
connection waiter;
--echo connection: waiter
2007-10-10 19:06:53 -03:00
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Flushing tables";
--source include/wait_condition.inc
Initial import of WL#3726 "DDL locking for all metadata objects". Backport of: ------------------------------------------------------------ revno: 2630.4.1 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Fri 2008-05-23 17:54:03 +0400 message: WL#3726 "DDL locking for all metadata objects". After review fixes in progress. ------------------------------------------------------------ This is the first patch in series. It transforms the metadata locking subsystem to use a dedicated module (mdl.h,cc). No significant changes in the locking protocol. The import passes the test suite with the exception of deprecated/removed 6.0 features, and MERGE tables. The latter are subject to a fix by WL#4144. Unfortunately, the original changeset comments got lost in a merge, thus this import has its own (largely insufficient) comments. This patch fixes Bug#25144 "replication / binlog with view breaks". Warning: this patch introduces an incompatible change: Under LOCK TABLES, it's no longer possible to FLUSH a table that was not locked for WRITE. Under LOCK TABLES, it's no longer possible to DROP a table or VIEW that was not locked for WRITE. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.2 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 14:03:45 +0400 message: WL#3726 "DDL locking for all metadata objects". After review fixes in progress. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.3 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 14:08:51 +0400 message: WL#3726 "DDL locking for all metadata objects" Fixed failing Windows builds by adding mdl.cc to the lists of files needed to build server/libmysqld on Windows. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.4 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 21:57:58 +0400 message: WL#3726 "DDL locking for all metadata objects". Fix for assert failures in kill.test which occured when one tried to kill ALTER TABLE statement on merge table while it was waiting in wait_while_table_is_used() for other connections to close this table. These assert failures stemmed from the fact that cleanup code in this case assumed that temporary table representing new version of table was open with adding to THD::temporary_tables list while code which were opening this temporary table wasn't always fulfilling this. This patch changes code that opens new version of table to always do this linking in. It also streamlines cleanup process for cases when error occurs while we have new version of table open. ****** WL#3726 "DDL locking for all metadata objects" Add libmysqld/mdl.cc to .bzrignore. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.6 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sun 2008-05-25 00:33:22 +0400 message: WL#3726 "DDL locking for all metadata objects". Addition to the fix of assert failures in kill.test caused by changes for this worklog. Make sure we close the new table only once.
2009-11-30 18:55:03 +03:00
connection default;
--echo connection: default
2007-10-10 19:06:53 -03:00
handler t2 open;
handler t2 read first;
handler t1 read next;
handler t1 close;
handler t2 close;
connection flush;
reap;
connection default;
drop table t1,t2;
disconnect flush;
2007-10-10 21:39:22 -03:00
#
# Bug#31409 RENAME TABLE causes server crash or deadlock when used with HANDLER statements
#
--disable_warnings
drop table if exists t1, t0;
2007-10-10 21:39:22 -03:00
--enable_warnings
create table t1 (c1 int);
--echo connection: default
handler t1 open;
handler t1 read first;
connect (flush,localhost,root,,);
connection flush;
--echo connection: flush
--send rename table t1 to t0;
Initial import of WL#3726 "DDL locking for all metadata objects". Backport of: ------------------------------------------------------------ revno: 2630.4.1 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Fri 2008-05-23 17:54:03 +0400 message: WL#3726 "DDL locking for all metadata objects". After review fixes in progress. ------------------------------------------------------------ This is the first patch in series. It transforms the metadata locking subsystem to use a dedicated module (mdl.h,cc). No significant changes in the locking protocol. The import passes the test suite with the exception of deprecated/removed 6.0 features, and MERGE tables. The latter are subject to a fix by WL#4144. Unfortunately, the original changeset comments got lost in a merge, thus this import has its own (largely insufficient) comments. This patch fixes Bug#25144 "replication / binlog with view breaks". Warning: this patch introduces an incompatible change: Under LOCK TABLES, it's no longer possible to FLUSH a table that was not locked for WRITE. Under LOCK TABLES, it's no longer possible to DROP a table or VIEW that was not locked for WRITE. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.2 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 14:03:45 +0400 message: WL#3726 "DDL locking for all metadata objects". After review fixes in progress. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.3 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 14:08:51 +0400 message: WL#3726 "DDL locking for all metadata objects" Fixed failing Windows builds by adding mdl.cc to the lists of files needed to build server/libmysqld on Windows. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.4 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 21:57:58 +0400 message: WL#3726 "DDL locking for all metadata objects". Fix for assert failures in kill.test which occured when one tried to kill ALTER TABLE statement on merge table while it was waiting in wait_while_table_is_used() for other connections to close this table. These assert failures stemmed from the fact that cleanup code in this case assumed that temporary table representing new version of table was open with adding to THD::temporary_tables list while code which were opening this temporary table wasn't always fulfilling this. This patch changes code that opens new version of table to always do this linking in. It also streamlines cleanup process for cases when error occurs while we have new version of table open. ****** WL#3726 "DDL locking for all metadata objects" Add libmysqld/mdl.cc to .bzrignore. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.6 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sun 2008-05-25 00:33:22 +0400 message: WL#3726 "DDL locking for all metadata objects". Addition to the fix of assert failures in kill.test caused by changes for this worklog. Make sure we close the new table only once.
2009-11-30 18:55:03 +03:00
connection waiter;
--echo connection: waiter
2007-10-10 21:39:22 -03:00
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Waiting for table" and info = "rename table t1 to t0";
2007-10-10 21:39:22 -03:00
--source include/wait_condition.inc
Initial import of WL#3726 "DDL locking for all metadata objects". Backport of: ------------------------------------------------------------ revno: 2630.4.1 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Fri 2008-05-23 17:54:03 +0400 message: WL#3726 "DDL locking for all metadata objects". After review fixes in progress. ------------------------------------------------------------ This is the first patch in series. It transforms the metadata locking subsystem to use a dedicated module (mdl.h,cc). No significant changes in the locking protocol. The import passes the test suite with the exception of deprecated/removed 6.0 features, and MERGE tables. The latter are subject to a fix by WL#4144. Unfortunately, the original changeset comments got lost in a merge, thus this import has its own (largely insufficient) comments. This patch fixes Bug#25144 "replication / binlog with view breaks". Warning: this patch introduces an incompatible change: Under LOCK TABLES, it's no longer possible to FLUSH a table that was not locked for WRITE. Under LOCK TABLES, it's no longer possible to DROP a table or VIEW that was not locked for WRITE. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.2 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 14:03:45 +0400 message: WL#3726 "DDL locking for all metadata objects". After review fixes in progress. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.3 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 14:08:51 +0400 message: WL#3726 "DDL locking for all metadata objects" Fixed failing Windows builds by adding mdl.cc to the lists of files needed to build server/libmysqld on Windows. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.4 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 21:57:58 +0400 message: WL#3726 "DDL locking for all metadata objects". Fix for assert failures in kill.test which occured when one tried to kill ALTER TABLE statement on merge table while it was waiting in wait_while_table_is_used() for other connections to close this table. These assert failures stemmed from the fact that cleanup code in this case assumed that temporary table representing new version of table was open with adding to THD::temporary_tables list while code which were opening this temporary table wasn't always fulfilling this. This patch changes code that opens new version of table to always do this linking in. It also streamlines cleanup process for cases when error occurs while we have new version of table open. ****** WL#3726 "DDL locking for all metadata objects" Add libmysqld/mdl.cc to .bzrignore. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.6 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sun 2008-05-25 00:33:22 +0400 message: WL#3726 "DDL locking for all metadata objects". Addition to the fix of assert failures in kill.test caused by changes for this worklog. Make sure we close the new table only once.
2009-11-30 18:55:03 +03:00
connection default;
--echo connection: default
A prerequisite patch for the fix for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". Introduce a notion of a sentinel to MDL_context. A sentinel is a ticket that separates all tickets in the context into two groups: before and after it. Currently we can have (and need) only one designated sentinel -- it separates all locks taken by LOCK TABLE or HANDLER statement, which must survive COMMIT and ROLLBACK and all other locks, which must be released at COMMIT or ROLLBACK. The tricky part is maintaining the sentinel up to date when someone release its corresponding ticket. This can happen, e.g. if someone issues DROP TABLE under LOCK TABLES (generally, see all calls to release_all_locks_for_name()). MDL_context::release_ticket() is modified to take care of it. ****** A fix and a test case for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". An attempt to mix HANDLER SQL statements, which are transaction- agnostic, an open multi-statement transaction, and DDL against the involved tables (in a concurrent connection) could lead to a deadlock. The deadlock would occur when HANDLER OPEN or HANDLER READ would have to wait on a conflicting metadata lock. If the connection that issued HANDLER statement also had other metadata locks (say, acquired in scope of a transaction), a classical deadlock situation of mutual wait could occur. Incompatible change: entering LOCK TABLES mode automatically closes all open HANDLERs in the current connection. Incompatible change: previously an attempt to wait on a lock in a connection that has an open HANDLER statement could wait indefinitely/deadlock. After this patch, an error ER_LOCK_DEADLOCK is produced. The idea of the fix is to merge thd->handler_mdl_context with the main mdl_context of the connection, used for transactional locks. This makes deadlock detection possible, since all waits with locks are "visible" and available to analysis in a single MDL context of the connection. Since HANDLER locks and transactional locks have a different life cycle -- HANDLERs are explicitly open and closed, and so are HANDLER locks, explicitly acquired and released, whereas transactional locks "accumulate" till the end of a transaction and are released only with COMMIT, ROLLBACK and ROLLBACK TO SAVEPOINT, a concept of "sentinel" was introduced to MDL_context. All locks, HANDLER and others, reside in the same linked list. However, a selected element of the list separates locks with different life cycle. HANDLER locks always reside at the end of the list, after the sentinel. Transactional locks are prepended to the beginning of the list, before the sentinel. Thus, ROLLBACK, COMMIT or ROLLBACK TO SAVEPOINT, only release those locks that reside before the sentinel. HANDLER locks must be released explicitly as part of HANDLER CLOSE statement, or an implicit close. The same approach with sentinel is also employed for LOCK TABLES locks. Since HANDLER and LOCK TABLES statement has never worked together, the implementation is made simple and only maintains one sentinel, which is used either for HANDLER locks, or for LOCK TABLES locks.
2009-12-22 19:09:15 +03:00
--echo #
--echo # RENAME placed two pending locks and waits.
--echo # When HANDLER t0 OPEN does open_tables(), it calls
A prerequisite patch for the fix for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". Introduce a notion of a sentinel to MDL_context. A sentinel is a ticket that separates all tickets in the context into two groups: before and after it. Currently we can have (and need) only one designated sentinel -- it separates all locks taken by LOCK TABLE or HANDLER statement, which must survive COMMIT and ROLLBACK and all other locks, which must be released at COMMIT or ROLLBACK. The tricky part is maintaining the sentinel up to date when someone release its corresponding ticket. This can happen, e.g. if someone issues DROP TABLE under LOCK TABLES (generally, see all calls to release_all_locks_for_name()). MDL_context::release_ticket() is modified to take care of it. ****** A fix and a test case for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". An attempt to mix HANDLER SQL statements, which are transaction- agnostic, an open multi-statement transaction, and DDL against the involved tables (in a concurrent connection) could lead to a deadlock. The deadlock would occur when HANDLER OPEN or HANDLER READ would have to wait on a conflicting metadata lock. If the connection that issued HANDLER statement also had other metadata locks (say, acquired in scope of a transaction), a classical deadlock situation of mutual wait could occur. Incompatible change: entering LOCK TABLES mode automatically closes all open HANDLERs in the current connection. Incompatible change: previously an attempt to wait on a lock in a connection that has an open HANDLER statement could wait indefinitely/deadlock. After this patch, an error ER_LOCK_DEADLOCK is produced. The idea of the fix is to merge thd->handler_mdl_context with the main mdl_context of the connection, used for transactional locks. This makes deadlock detection possible, since all waits with locks are "visible" and available to analysis in a single MDL context of the connection. Since HANDLER locks and transactional locks have a different life cycle -- HANDLERs are explicitly open and closed, and so are HANDLER locks, explicitly acquired and released, whereas transactional locks "accumulate" till the end of a transaction and are released only with COMMIT, ROLLBACK and ROLLBACK TO SAVEPOINT, a concept of "sentinel" was introduced to MDL_context. All locks, HANDLER and others, reside in the same linked list. However, a selected element of the list separates locks with different life cycle. HANDLER locks always reside at the end of the list, after the sentinel. Transactional locks are prepended to the beginning of the list, before the sentinel. Thus, ROLLBACK, COMMIT or ROLLBACK TO SAVEPOINT, only release those locks that reside before the sentinel. HANDLER locks must be released explicitly as part of HANDLER CLOSE statement, or an implicit close. The same approach with sentinel is also employed for LOCK TABLES locks. Since HANDLER and LOCK TABLES statement has never worked together, the implementation is made simple and only maintains one sentinel, which is used either for HANDLER locks, or for LOCK TABLES locks.
2009-12-22 19:09:15 +03:00
--echo # mysql_ha_flush(), which in turn closes the open HANDLER for t1.
--echo # RENAME TABLE gets unblocked. If it gets scheduled quickly
--echo # and manages to complete before open_tables()
--echo # of HANDLER t0 OPEN, open_tables() and therefore the whole
--echo # HANDLER t0 OPEN succeeds. Otherwise open_tables()
A prerequisite patch for the fix for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". Introduce a notion of a sentinel to MDL_context. A sentinel is a ticket that separates all tickets in the context into two groups: before and after it. Currently we can have (and need) only one designated sentinel -- it separates all locks taken by LOCK TABLE or HANDLER statement, which must survive COMMIT and ROLLBACK and all other locks, which must be released at COMMIT or ROLLBACK. The tricky part is maintaining the sentinel up to date when someone release its corresponding ticket. This can happen, e.g. if someone issues DROP TABLE under LOCK TABLES (generally, see all calls to release_all_locks_for_name()). MDL_context::release_ticket() is modified to take care of it. ****** A fix and a test case for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". An attempt to mix HANDLER SQL statements, which are transaction- agnostic, an open multi-statement transaction, and DDL against the involved tables (in a concurrent connection) could lead to a deadlock. The deadlock would occur when HANDLER OPEN or HANDLER READ would have to wait on a conflicting metadata lock. If the connection that issued HANDLER statement also had other metadata locks (say, acquired in scope of a transaction), a classical deadlock situation of mutual wait could occur. Incompatible change: entering LOCK TABLES mode automatically closes all open HANDLERs in the current connection. Incompatible change: previously an attempt to wait on a lock in a connection that has an open HANDLER statement could wait indefinitely/deadlock. After this patch, an error ER_LOCK_DEADLOCK is produced. The idea of the fix is to merge thd->handler_mdl_context with the main mdl_context of the connection, used for transactional locks. This makes deadlock detection possible, since all waits with locks are "visible" and available to analysis in a single MDL context of the connection. Since HANDLER locks and transactional locks have a different life cycle -- HANDLERs are explicitly open and closed, and so are HANDLER locks, explicitly acquired and released, whereas transactional locks "accumulate" till the end of a transaction and are released only with COMMIT, ROLLBACK and ROLLBACK TO SAVEPOINT, a concept of "sentinel" was introduced to MDL_context. All locks, HANDLER and others, reside in the same linked list. However, a selected element of the list separates locks with different life cycle. HANDLER locks always reside at the end of the list, after the sentinel. Transactional locks are prepended to the beginning of the list, before the sentinel. Thus, ROLLBACK, COMMIT or ROLLBACK TO SAVEPOINT, only release those locks that reside before the sentinel. HANDLER locks must be released explicitly as part of HANDLER CLOSE statement, or an implicit close. The same approach with sentinel is also employed for LOCK TABLES locks. Since HANDLER and LOCK TABLES statement has never worked together, the implementation is made simple and only maintains one sentinel, which is used either for HANDLER locks, or for LOCK TABLES locks.
2009-12-22 19:09:15 +03:00
--echo # notices a pending or active exclusive metadata lock on t2
--echo # and the whole HANDLER t0 OPEN fails with ER_LOCK_DEADLOCK
A prerequisite patch for the fix for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". Introduce a notion of a sentinel to MDL_context. A sentinel is a ticket that separates all tickets in the context into two groups: before and after it. Currently we can have (and need) only one designated sentinel -- it separates all locks taken by LOCK TABLE or HANDLER statement, which must survive COMMIT and ROLLBACK and all other locks, which must be released at COMMIT or ROLLBACK. The tricky part is maintaining the sentinel up to date when someone release its corresponding ticket. This can happen, e.g. if someone issues DROP TABLE under LOCK TABLES (generally, see all calls to release_all_locks_for_name()). MDL_context::release_ticket() is modified to take care of it. ****** A fix and a test case for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". An attempt to mix HANDLER SQL statements, which are transaction- agnostic, an open multi-statement transaction, and DDL against the involved tables (in a concurrent connection) could lead to a deadlock. The deadlock would occur when HANDLER OPEN or HANDLER READ would have to wait on a conflicting metadata lock. If the connection that issued HANDLER statement also had other metadata locks (say, acquired in scope of a transaction), a classical deadlock situation of mutual wait could occur. Incompatible change: entering LOCK TABLES mode automatically closes all open HANDLERs in the current connection. Incompatible change: previously an attempt to wait on a lock in a connection that has an open HANDLER statement could wait indefinitely/deadlock. After this patch, an error ER_LOCK_DEADLOCK is produced. The idea of the fix is to merge thd->handler_mdl_context with the main mdl_context of the connection, used for transactional locks. This makes deadlock detection possible, since all waits with locks are "visible" and available to analysis in a single MDL context of the connection. Since HANDLER locks and transactional locks have a different life cycle -- HANDLERs are explicitly open and closed, and so are HANDLER locks, explicitly acquired and released, whereas transactional locks "accumulate" till the end of a transaction and are released only with COMMIT, ROLLBACK and ROLLBACK TO SAVEPOINT, a concept of "sentinel" was introduced to MDL_context. All locks, HANDLER and others, reside in the same linked list. However, a selected element of the list separates locks with different life cycle. HANDLER locks always reside at the end of the list, after the sentinel. Transactional locks are prepended to the beginning of the list, before the sentinel. Thus, ROLLBACK, COMMIT or ROLLBACK TO SAVEPOINT, only release those locks that reside before the sentinel. HANDLER locks must be released explicitly as part of HANDLER CLOSE statement, or an implicit close. The same approach with sentinel is also employed for LOCK TABLES locks. Since HANDLER and LOCK TABLES statement has never worked together, the implementation is made simple and only maintains one sentinel, which is used either for HANDLER locks, or for LOCK TABLES locks.
2009-12-22 19:09:15 +03:00
--echo # error.
--echo #
--error 0, ER_LOCK_DEADLOCK
handler t0 open;
A prerequisite patch for the fix for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". Introduce a notion of a sentinel to MDL_context. A sentinel is a ticket that separates all tickets in the context into two groups: before and after it. Currently we can have (and need) only one designated sentinel -- it separates all locks taken by LOCK TABLE or HANDLER statement, which must survive COMMIT and ROLLBACK and all other locks, which must be released at COMMIT or ROLLBACK. The tricky part is maintaining the sentinel up to date when someone release its corresponding ticket. This can happen, e.g. if someone issues DROP TABLE under LOCK TABLES (generally, see all calls to release_all_locks_for_name()). MDL_context::release_ticket() is modified to take care of it. ****** A fix and a test case for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". An attempt to mix HANDLER SQL statements, which are transaction- agnostic, an open multi-statement transaction, and DDL against the involved tables (in a concurrent connection) could lead to a deadlock. The deadlock would occur when HANDLER OPEN or HANDLER READ would have to wait on a conflicting metadata lock. If the connection that issued HANDLER statement also had other metadata locks (say, acquired in scope of a transaction), a classical deadlock situation of mutual wait could occur. Incompatible change: entering LOCK TABLES mode automatically closes all open HANDLERs in the current connection. Incompatible change: previously an attempt to wait on a lock in a connection that has an open HANDLER statement could wait indefinitely/deadlock. After this patch, an error ER_LOCK_DEADLOCK is produced. The idea of the fix is to merge thd->handler_mdl_context with the main mdl_context of the connection, used for transactional locks. This makes deadlock detection possible, since all waits with locks are "visible" and available to analysis in a single MDL context of the connection. Since HANDLER locks and transactional locks have a different life cycle -- HANDLERs are explicitly open and closed, and so are HANDLER locks, explicitly acquired and released, whereas transactional locks "accumulate" till the end of a transaction and are released only with COMMIT, ROLLBACK and ROLLBACK TO SAVEPOINT, a concept of "sentinel" was introduced to MDL_context. All locks, HANDLER and others, reside in the same linked list. However, a selected element of the list separates locks with different life cycle. HANDLER locks always reside at the end of the list, after the sentinel. Transactional locks are prepended to the beginning of the list, before the sentinel. Thus, ROLLBACK, COMMIT or ROLLBACK TO SAVEPOINT, only release those locks that reside before the sentinel. HANDLER locks must be released explicitly as part of HANDLER CLOSE statement, or an implicit close. The same approach with sentinel is also employed for LOCK TABLES locks. Since HANDLER and LOCK TABLES statement has never worked together, the implementation is made simple and only maintains one sentinel, which is used either for HANDLER locks, or for LOCK TABLES locks.
2009-12-22 19:09:15 +03:00
--error 0, ER_UNKNOWN_TABLE
handler t0 close;
A prerequisite patch for the fix for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". Introduce a notion of a sentinel to MDL_context. A sentinel is a ticket that separates all tickets in the context into two groups: before and after it. Currently we can have (and need) only one designated sentinel -- it separates all locks taken by LOCK TABLE or HANDLER statement, which must survive COMMIT and ROLLBACK and all other locks, which must be released at COMMIT or ROLLBACK. The tricky part is maintaining the sentinel up to date when someone release its corresponding ticket. This can happen, e.g. if someone issues DROP TABLE under LOCK TABLES (generally, see all calls to release_all_locks_for_name()). MDL_context::release_ticket() is modified to take care of it. ****** A fix and a test case for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". An attempt to mix HANDLER SQL statements, which are transaction- agnostic, an open multi-statement transaction, and DDL against the involved tables (in a concurrent connection) could lead to a deadlock. The deadlock would occur when HANDLER OPEN or HANDLER READ would have to wait on a conflicting metadata lock. If the connection that issued HANDLER statement also had other metadata locks (say, acquired in scope of a transaction), a classical deadlock situation of mutual wait could occur. Incompatible change: entering LOCK TABLES mode automatically closes all open HANDLERs in the current connection. Incompatible change: previously an attempt to wait on a lock in a connection that has an open HANDLER statement could wait indefinitely/deadlock. After this patch, an error ER_LOCK_DEADLOCK is produced. The idea of the fix is to merge thd->handler_mdl_context with the main mdl_context of the connection, used for transactional locks. This makes deadlock detection possible, since all waits with locks are "visible" and available to analysis in a single MDL context of the connection. Since HANDLER locks and transactional locks have a different life cycle -- HANDLERs are explicitly open and closed, and so are HANDLER locks, explicitly acquired and released, whereas transactional locks "accumulate" till the end of a transaction and are released only with COMMIT, ROLLBACK and ROLLBACK TO SAVEPOINT, a concept of "sentinel" was introduced to MDL_context. All locks, HANDLER and others, reside in the same linked list. However, a selected element of the list separates locks with different life cycle. HANDLER locks always reside at the end of the list, after the sentinel. Transactional locks are prepended to the beginning of the list, before the sentinel. Thus, ROLLBACK, COMMIT or ROLLBACK TO SAVEPOINT, only release those locks that reside before the sentinel. HANDLER locks must be released explicitly as part of HANDLER CLOSE statement, or an implicit close. The same approach with sentinel is also employed for LOCK TABLES locks. Since HANDLER and LOCK TABLES statement has never worked together, the implementation is made simple and only maintains one sentinel, which is used either for HANDLER locks, or for LOCK TABLES locks.
2009-12-22 19:09:15 +03:00
--echo connection: flush
2007-10-10 21:39:22 -03:00
connection flush;
reap;
A prerequisite patch for the fix for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". Introduce a notion of a sentinel to MDL_context. A sentinel is a ticket that separates all tickets in the context into two groups: before and after it. Currently we can have (and need) only one designated sentinel -- it separates all locks taken by LOCK TABLE or HANDLER statement, which must survive COMMIT and ROLLBACK and all other locks, which must be released at COMMIT or ROLLBACK. The tricky part is maintaining the sentinel up to date when someone release its corresponding ticket. This can happen, e.g. if someone issues DROP TABLE under LOCK TABLES (generally, see all calls to release_all_locks_for_name()). MDL_context::release_ticket() is modified to take care of it. ****** A fix and a test case for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". An attempt to mix HANDLER SQL statements, which are transaction- agnostic, an open multi-statement transaction, and DDL against the involved tables (in a concurrent connection) could lead to a deadlock. The deadlock would occur when HANDLER OPEN or HANDLER READ would have to wait on a conflicting metadata lock. If the connection that issued HANDLER statement also had other metadata locks (say, acquired in scope of a transaction), a classical deadlock situation of mutual wait could occur. Incompatible change: entering LOCK TABLES mode automatically closes all open HANDLERs in the current connection. Incompatible change: previously an attempt to wait on a lock in a connection that has an open HANDLER statement could wait indefinitely/deadlock. After this patch, an error ER_LOCK_DEADLOCK is produced. The idea of the fix is to merge thd->handler_mdl_context with the main mdl_context of the connection, used for transactional locks. This makes deadlock detection possible, since all waits with locks are "visible" and available to analysis in a single MDL context of the connection. Since HANDLER locks and transactional locks have a different life cycle -- HANDLERs are explicitly open and closed, and so are HANDLER locks, explicitly acquired and released, whereas transactional locks "accumulate" till the end of a transaction and are released only with COMMIT, ROLLBACK and ROLLBACK TO SAVEPOINT, a concept of "sentinel" was introduced to MDL_context. All locks, HANDLER and others, reside in the same linked list. However, a selected element of the list separates locks with different life cycle. HANDLER locks always reside at the end of the list, after the sentinel. Transactional locks are prepended to the beginning of the list, before the sentinel. Thus, ROLLBACK, COMMIT or ROLLBACK TO SAVEPOINT, only release those locks that reside before the sentinel. HANDLER locks must be released explicitly as part of HANDLER CLOSE statement, or an implicit close. The same approach with sentinel is also employed for LOCK TABLES locks. Since HANDLER and LOCK TABLES statement has never worked together, the implementation is made simple and only maintains one sentinel, which is used either for HANDLER locks, or for LOCK TABLES locks.
2009-12-22 19:09:15 +03:00
--error ER_UNKNOWN_TABLE
handler t1 read next;
--error ER_UNKNOWN_TABLE
handler t1 close;
2007-10-10 21:39:22 -03:00
connection default;
drop table t0;
Initial import of WL#3726 "DDL locking for all metadata objects". Backport of: ------------------------------------------------------------ revno: 2630.4.1 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Fri 2008-05-23 17:54:03 +0400 message: WL#3726 "DDL locking for all metadata objects". After review fixes in progress. ------------------------------------------------------------ This is the first patch in series. It transforms the metadata locking subsystem to use a dedicated module (mdl.h,cc). No significant changes in the locking protocol. The import passes the test suite with the exception of deprecated/removed 6.0 features, and MERGE tables. The latter are subject to a fix by WL#4144. Unfortunately, the original changeset comments got lost in a merge, thus this import has its own (largely insufficient) comments. This patch fixes Bug#25144 "replication / binlog with view breaks". Warning: this patch introduces an incompatible change: Under LOCK TABLES, it's no longer possible to FLUSH a table that was not locked for WRITE. Under LOCK TABLES, it's no longer possible to DROP a table or VIEW that was not locked for WRITE. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.2 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 14:03:45 +0400 message: WL#3726 "DDL locking for all metadata objects". After review fixes in progress. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.3 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 14:08:51 +0400 message: WL#3726 "DDL locking for all metadata objects" Fixed failing Windows builds by adding mdl.cc to the lists of files needed to build server/libmysqld on Windows. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.4 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 21:57:58 +0400 message: WL#3726 "DDL locking for all metadata objects". Fix for assert failures in kill.test which occured when one tried to kill ALTER TABLE statement on merge table while it was waiting in wait_while_table_is_used() for other connections to close this table. These assert failures stemmed from the fact that cleanup code in this case assumed that temporary table representing new version of table was open with adding to THD::temporary_tables list while code which were opening this temporary table wasn't always fulfilling this. This patch changes code that opens new version of table to always do this linking in. It also streamlines cleanup process for cases when error occurs while we have new version of table open. ****** WL#3726 "DDL locking for all metadata objects" Add libmysqld/mdl.cc to .bzrignore. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.6 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sun 2008-05-25 00:33:22 +0400 message: WL#3726 "DDL locking for all metadata objects". Addition to the fix of assert failures in kill.test caused by changes for this worklog. Make sure we close the new table only once.
2009-11-30 18:55:03 +03:00
connection flush;
2007-10-10 21:39:22 -03:00
disconnect flush;
Initial import of WL#3726 "DDL locking for all metadata objects". Backport of: ------------------------------------------------------------ revno: 2630.4.1 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Fri 2008-05-23 17:54:03 +0400 message: WL#3726 "DDL locking for all metadata objects". After review fixes in progress. ------------------------------------------------------------ This is the first patch in series. It transforms the metadata locking subsystem to use a dedicated module (mdl.h,cc). No significant changes in the locking protocol. The import passes the test suite with the exception of deprecated/removed 6.0 features, and MERGE tables. The latter are subject to a fix by WL#4144. Unfortunately, the original changeset comments got lost in a merge, thus this import has its own (largely insufficient) comments. This patch fixes Bug#25144 "replication / binlog with view breaks". Warning: this patch introduces an incompatible change: Under LOCK TABLES, it's no longer possible to FLUSH a table that was not locked for WRITE. Under LOCK TABLES, it's no longer possible to DROP a table or VIEW that was not locked for WRITE. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.2 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 14:03:45 +0400 message: WL#3726 "DDL locking for all metadata objects". After review fixes in progress. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.3 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 14:08:51 +0400 message: WL#3726 "DDL locking for all metadata objects" Fixed failing Windows builds by adding mdl.cc to the lists of files needed to build server/libmysqld on Windows. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.4 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 21:57:58 +0400 message: WL#3726 "DDL locking for all metadata objects". Fix for assert failures in kill.test which occured when one tried to kill ALTER TABLE statement on merge table while it was waiting in wait_while_table_is_used() for other connections to close this table. These assert failures stemmed from the fact that cleanup code in this case assumed that temporary table representing new version of table was open with adding to THD::temporary_tables list while code which were opening this temporary table wasn't always fulfilling this. This patch changes code that opens new version of table to always do this linking in. It also streamlines cleanup process for cases when error occurs while we have new version of table open. ****** WL#3726 "DDL locking for all metadata objects" Add libmysqld/mdl.cc to .bzrignore. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.6 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sun 2008-05-25 00:33:22 +0400 message: WL#3726 "DDL locking for all metadata objects". Addition to the fix of assert failures in kill.test caused by changes for this worklog. Make sure we close the new table only once.
2009-11-30 18:55:03 +03:00
--source include/wait_until_disconnected.inc
connection waiter;
disconnect waiter;
--source include/wait_until_disconnected.inc
connection default;
#
# Bug#30882 Dropping a temporary table inside a stored function may cause a server crash
#
# Test HANDLER statements in conjunction with temporary tables. While the temporary table
# is open by a HANDLER, no other statement can access it.
#
--disable_warnings
drop table if exists t1;
--enable_warnings
create temporary table t1 (a int, b char(1), key a(a), key b(a,b));
insert into t1 values (0,"a"),(1,"b"),(2,"c"),(3,"d"),(4,"e"),
(5,"f"),(6,"g"),(7,"h"),(8,"i"),(9,"j");
select a,b from t1;
handler t1 open as a1;
handler a1 read a first;
handler a1 read a next;
handler a1 read a next;
--error ER_CANT_REOPEN_TABLE
select a,b from t1;
handler a1 read a prev;
handler a1 read a prev;
handler a1 read a=(6) where b="g";
handler a1 close;
select a,b from t1;
handler t1 open as a2;
handler a2 read a first;
handler a2 read a last;
handler a2 read a prev;
handler a2 close;
drop table t1;
#
# Bug#31397 Inconsistent drop table behavior of handler tables.
#
--disable_warnings
drop table if exists t1,t2;
--enable_warnings
create table t1 (a int);
handler t1 open as t1_alias;
drop table t1;
create table t1 (a int);
handler t1 open as t1_alias;
flush tables;
drop table t1;
create table t1 (a int);
handler t1 open as t1_alias;
handler t1_alias close;
drop table t1;
create table t1 (a int);
handler t1 open as t1_alias;
handler t1_alias read first;
drop table t1;
--error ER_UNKNOWN_TABLE
handler t1_alias read next;
# Test that temporary tables associated with handlers are properly dropped.
create table t1 (a int);
create temporary table t2 (a int, key(a));
handler t1 open as a1;
handler t2 open as a2;
handler a2 read a first;
drop table t1, t2;
--error ER_UNKNOWN_TABLE
handler a2 read a next;
--error ER_UNKNOWN_TABLE
handler a1 close;
# Alter table drop handlers
create table t1 (a int, key(a));
create table t2 like t1;
handler t1 open as a1;
handler t2 open as a2;
handler a1 read a first;
handler a2 read a first;
alter table t1 add b int;
--error ER_UNKNOWN_TABLE
handler a1 close;
handler a2 close;
drop table t1, t2;
# Rename table drop handlers
create table t1 (a int, key(a));
handler t1 open as a1;
handler a1 read a first;
rename table t1 to t2;
--error ER_UNKNOWN_TABLE
handler a1 read a first;
drop table t2;
# Optimize table drop handlers
create table t1 (a int, key(a));
create table t2 like t1;
handler t1 open as a1;
handler t2 open as a2;
handler a1 read a first;
handler a2 read a first;
optimize table t1;
--error ER_UNKNOWN_TABLE
handler a1 close;
handler a2 close;
drop table t1, t2;
# Flush tables causes handlers reopen
create table t1 (a int, b char(1), key a(a), key b(a,b));
insert into t1 values (0,"a"),(1,"b"),(2,"c"),(3,"d"),(4,"e"),
(5,"f"),(6,"g"),(7,"h"),(8,"i"),(9,"j");
handler t1 open;
handler t1 read a first;
handler t1 read a next;
flush tables;
handler t1 read a next;
handler t1 read a next;
flush tables with read lock;
handler t1 read a next;
unlock tables;
drop table t1;
--error ER_UNKNOWN_TABLE
handler t1 read a next;
#
# Bug#41110: crash with handler command when used concurrently with alter table
# Bug#41112: crash in mysql_ha_close_table/get_lock_data with alter table
#
Initial import of WL#3726 "DDL locking for all metadata objects". Backport of: ------------------------------------------------------------ revno: 2630.4.1 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Fri 2008-05-23 17:54:03 +0400 message: WL#3726 "DDL locking for all metadata objects". After review fixes in progress. ------------------------------------------------------------ This is the first patch in series. It transforms the metadata locking subsystem to use a dedicated module (mdl.h,cc). No significant changes in the locking protocol. The import passes the test suite with the exception of deprecated/removed 6.0 features, and MERGE tables. The latter are subject to a fix by WL#4144. Unfortunately, the original changeset comments got lost in a merge, thus this import has its own (largely insufficient) comments. This patch fixes Bug#25144 "replication / binlog with view breaks". Warning: this patch introduces an incompatible change: Under LOCK TABLES, it's no longer possible to FLUSH a table that was not locked for WRITE. Under LOCK TABLES, it's no longer possible to DROP a table or VIEW that was not locked for WRITE. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.2 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 14:03:45 +0400 message: WL#3726 "DDL locking for all metadata objects". After review fixes in progress. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.3 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 14:08:51 +0400 message: WL#3726 "DDL locking for all metadata objects" Fixed failing Windows builds by adding mdl.cc to the lists of files needed to build server/libmysqld on Windows. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.4 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 21:57:58 +0400 message: WL#3726 "DDL locking for all metadata objects". Fix for assert failures in kill.test which occured when one tried to kill ALTER TABLE statement on merge table while it was waiting in wait_while_table_is_used() for other connections to close this table. These assert failures stemmed from the fact that cleanup code in this case assumed that temporary table representing new version of table was open with adding to THD::temporary_tables list while code which were opening this temporary table wasn't always fulfilling this. This patch changes code that opens new version of table to always do this linking in. It also streamlines cleanup process for cases when error occurs while we have new version of table open. ****** WL#3726 "DDL locking for all metadata objects" Add libmysqld/mdl.cc to .bzrignore. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.6 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sun 2008-05-25 00:33:22 +0400 message: WL#3726 "DDL locking for all metadata objects". Addition to the fix of assert failures in kill.test caused by changes for this worklog. Make sure we close the new table only once.
2009-11-30 18:55:03 +03:00
connect(con1,localhost,root,,);
connect(con2,localhost,root,,);
connection default;
--disable_warnings
drop table if exists t1;
--enable_warnings
Implement new type-of-operation-aware metadata locks. Add a wait-for graph based deadlock detector to the MDL subsystem. Fixes bug #46272 "MySQL 5.4.4, new MDL: unnecessary deadlock" and bug #37346 "innodb does not detect deadlock between update and alter table". The first bug manifested itself as an unwarranted abort of a transaction with ER_LOCK_DEADLOCK error by a concurrent ALTER statement, when this transaction tried to repeat use of a table, which it has already used in a similar fashion before ALTER started. The second bug showed up as a deadlock between table-level locks and InnoDB row locks, which was "detected" only after innodb_lock_wait_timeout timeout. A transaction would start using the table and modify a few rows. Then ALTER TABLE would come in, and start copying rows into a temporary table. Eventually it would stumble on the modified records and get blocked on a row lock. The first transaction would try to do more updates, and get blocked on thr_lock.c lock. This situation of circular wait would only get resolved by a timeout. Both these bugs stemmed from inadequate solutions to the problem of deadlocks occurring between different locking subsystems. In the first case we tried to avoid deadlocks between metadata locking and table-level locking subsystems, when upgrading shared metadata lock to exclusive one. Transactions holding the shared lock on the table and waiting for some table-level lock used to be aborted too aggressively. We also allowed ALTER TABLE to start in presence of transactions that modify the subject table. ALTER TABLE acquires TL_WRITE_ALLOW_READ lock at start, and that block all writes against the table (naturally, we don't want any writes to be lost when switching the old and the new table). TL_WRITE_ALLOW_READ lock, in turn, would block the started transaction on thr_lock.c lock, should they do more updates. This, again, lead to the need to abort such transactions. The second bug occurred simply because we didn't have any mechanism to detect deadlocks between the table-level locks in thr_lock.c and row-level locks in InnoDB, other than innodb_lock_wait_timeout. This patch solves both these problems by moving lock conflicts which are causing these deadlocks into the metadata locking subsystem, thus making it possible to avoid or detect such deadlocks inside MDL. To do this we introduce new type-of-operation-aware metadata locks, which allow MDL subsystem to know not only the fact that transaction has used or is going to use some object but also what kind of operation it has carried out or going to carry out on the object. This, along with the addition of a special kind of upgradable metadata lock, allows ALTER TABLE to wait until all transactions which has updated the table to go away. This solves the second issue. Another special type of upgradable metadata lock is acquired by LOCK TABLE WRITE. This second lock type allows to solve the first issue, since abortion of table-level locks in event of DDL under LOCK TABLES becomes also unnecessary. Below follows the list of incompatible changes introduced by this patch: - From now on, ALTER TABLE and CREATE/DROP TRIGGER SQL (i.e. those statements that acquire TL_WRITE_ALLOW_READ lock) wait for all transactions which has *updated* the table to complete. - From now on, LOCK TABLES ... WRITE, REPAIR/OPTIMIZE TABLE (i.e. all statements which acquire TL_WRITE table-level lock) wait for all transaction which *updated or read* from the table to complete. As a consequence, innodb_table_locks=0 option no longer applies to LOCK TABLES ... WRITE. - DROP DATABASE, DROP TABLE, RENAME TABLE no longer abort statements or transactions which use tables being dropped or renamed, and instead wait for these transactions to complete. - Since LOCK TABLES WRITE now takes a special metadata lock, not compatible with with reads or writes against the subject table and transaction-wide, thr_lock.c deadlock avoidance algorithm that used to ensure absence of deadlocks between LOCK TABLES WRITE and other statements is no longer sufficient, even for MyISAM. The wait-for graph based deadlock detector of MDL subsystem may sometimes be necessary and is involved. This may lead to ER_LOCK_DEADLOCK error produced for multi-statement transactions even if these only use MyISAM: session 1: session 2: begin; update t1 ... lock table t2 write, t1 write; -- gets a lock on t2, blocks on t1 update t2 ... (ER_LOCK_DEADLOCK) - Finally, support of LOW_PRIORITY option for LOCK TABLES ... WRITE was abandoned. LOCK TABLE ... LOW_PRIORITY WRITE from now on has the same priority as the usual LOCK TABLE ... WRITE. SELECT HIGH PRIORITY no longer trumps LOCK TABLE ... WRITE in the wait queue. - We do not take upgradable metadata locks on implicitly locked tables. So if one has, say, a view v1 that uses table t1, and issues: LOCK TABLE v1 WRITE; FLUSH TABLE t1; -- (or just 'FLUSH TABLES'), an error is produced. In order to be able to perform DDL on a table under LOCK TABLES, the table must be locked explicitly in the LOCK TABLES list.
2010-02-01 14:43:06 +03:00
--echo # First test case which is supposed trigger the execution
--echo # path on which problem was discovered.
create table t1 (a int);
insert into t1 values (1);
handler t1 open;
Initial import of WL#3726 "DDL locking for all metadata objects". Backport of: ------------------------------------------------------------ revno: 2630.4.1 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Fri 2008-05-23 17:54:03 +0400 message: WL#3726 "DDL locking for all metadata objects". After review fixes in progress. ------------------------------------------------------------ This is the first patch in series. It transforms the metadata locking subsystem to use a dedicated module (mdl.h,cc). No significant changes in the locking protocol. The import passes the test suite with the exception of deprecated/removed 6.0 features, and MERGE tables. The latter are subject to a fix by WL#4144. Unfortunately, the original changeset comments got lost in a merge, thus this import has its own (largely insufficient) comments. This patch fixes Bug#25144 "replication / binlog with view breaks". Warning: this patch introduces an incompatible change: Under LOCK TABLES, it's no longer possible to FLUSH a table that was not locked for WRITE. Under LOCK TABLES, it's no longer possible to DROP a table or VIEW that was not locked for WRITE. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.2 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 14:03:45 +0400 message: WL#3726 "DDL locking for all metadata objects". After review fixes in progress. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.3 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 14:08:51 +0400 message: WL#3726 "DDL locking for all metadata objects" Fixed failing Windows builds by adding mdl.cc to the lists of files needed to build server/libmysqld on Windows. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.4 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 21:57:58 +0400 message: WL#3726 "DDL locking for all metadata objects". Fix for assert failures in kill.test which occured when one tried to kill ALTER TABLE statement on merge table while it was waiting in wait_while_table_is_used() for other connections to close this table. These assert failures stemmed from the fact that cleanup code in this case assumed that temporary table representing new version of table was open with adding to THD::temporary_tables list while code which were opening this temporary table wasn't always fulfilling this. This patch changes code that opens new version of table to always do this linking in. It also streamlines cleanup process for cases when error occurs while we have new version of table open. ****** WL#3726 "DDL locking for all metadata objects" Add libmysqld/mdl.cc to .bzrignore. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.6 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sun 2008-05-25 00:33:22 +0400 message: WL#3726 "DDL locking for all metadata objects". Addition to the fix of assert failures in kill.test caused by changes for this worklog. Make sure we close the new table only once.
2009-11-30 18:55:03 +03:00
connection con1;
Implement new type-of-operation-aware metadata locks. Add a wait-for graph based deadlock detector to the MDL subsystem. Fixes bug #46272 "MySQL 5.4.4, new MDL: unnecessary deadlock" and bug #37346 "innodb does not detect deadlock between update and alter table". The first bug manifested itself as an unwarranted abort of a transaction with ER_LOCK_DEADLOCK error by a concurrent ALTER statement, when this transaction tried to repeat use of a table, which it has already used in a similar fashion before ALTER started. The second bug showed up as a deadlock between table-level locks and InnoDB row locks, which was "detected" only after innodb_lock_wait_timeout timeout. A transaction would start using the table and modify a few rows. Then ALTER TABLE would come in, and start copying rows into a temporary table. Eventually it would stumble on the modified records and get blocked on a row lock. The first transaction would try to do more updates, and get blocked on thr_lock.c lock. This situation of circular wait would only get resolved by a timeout. Both these bugs stemmed from inadequate solutions to the problem of deadlocks occurring between different locking subsystems. In the first case we tried to avoid deadlocks between metadata locking and table-level locking subsystems, when upgrading shared metadata lock to exclusive one. Transactions holding the shared lock on the table and waiting for some table-level lock used to be aborted too aggressively. We also allowed ALTER TABLE to start in presence of transactions that modify the subject table. ALTER TABLE acquires TL_WRITE_ALLOW_READ lock at start, and that block all writes against the table (naturally, we don't want any writes to be lost when switching the old and the new table). TL_WRITE_ALLOW_READ lock, in turn, would block the started transaction on thr_lock.c lock, should they do more updates. This, again, lead to the need to abort such transactions. The second bug occurred simply because we didn't have any mechanism to detect deadlocks between the table-level locks in thr_lock.c and row-level locks in InnoDB, other than innodb_lock_wait_timeout. This patch solves both these problems by moving lock conflicts which are causing these deadlocks into the metadata locking subsystem, thus making it possible to avoid or detect such deadlocks inside MDL. To do this we introduce new type-of-operation-aware metadata locks, which allow MDL subsystem to know not only the fact that transaction has used or is going to use some object but also what kind of operation it has carried out or going to carry out on the object. This, along with the addition of a special kind of upgradable metadata lock, allows ALTER TABLE to wait until all transactions which has updated the table to go away. This solves the second issue. Another special type of upgradable metadata lock is acquired by LOCK TABLE WRITE. This second lock type allows to solve the first issue, since abortion of table-level locks in event of DDL under LOCK TABLES becomes also unnecessary. Below follows the list of incompatible changes introduced by this patch: - From now on, ALTER TABLE and CREATE/DROP TRIGGER SQL (i.e. those statements that acquire TL_WRITE_ALLOW_READ lock) wait for all transactions which has *updated* the table to complete. - From now on, LOCK TABLES ... WRITE, REPAIR/OPTIMIZE TABLE (i.e. all statements which acquire TL_WRITE table-level lock) wait for all transaction which *updated or read* from the table to complete. As a consequence, innodb_table_locks=0 option no longer applies to LOCK TABLES ... WRITE. - DROP DATABASE, DROP TABLE, RENAME TABLE no longer abort statements or transactions which use tables being dropped or renamed, and instead wait for these transactions to complete. - Since LOCK TABLES WRITE now takes a special metadata lock, not compatible with with reads or writes against the subject table and transaction-wide, thr_lock.c deadlock avoidance algorithm that used to ensure absence of deadlocks between LOCK TABLES WRITE and other statements is no longer sufficient, even for MyISAM. The wait-for graph based deadlock detector of MDL subsystem may sometimes be necessary and is involved. This may lead to ER_LOCK_DEADLOCK error produced for multi-statement transactions even if these only use MyISAM: session 1: session 2: begin; update t1 ... lock table t2 write, t1 write; -- gets a lock on t2, blocks on t1 update t2 ... (ER_LOCK_DEADLOCK) - Finally, support of LOW_PRIORITY option for LOCK TABLES ... WRITE was abandoned. LOCK TABLE ... LOW_PRIORITY WRITE from now on has the same priority as the usual LOCK TABLE ... WRITE. SELECT HIGH PRIORITY no longer trumps LOCK TABLE ... WRITE in the wait queue. - We do not take upgradable metadata locks on implicitly locked tables. So if one has, say, a view v1 that uses table t1, and issues: LOCK TABLE v1 WRITE; FLUSH TABLE t1; -- (or just 'FLUSH TABLES'), an error is produced. In order to be able to perform DDL on a table under LOCK TABLES, the table must be locked explicitly in the LOCK TABLES list.
2010-02-01 14:43:06 +03:00
lock table t1 write;
send alter table t1 engine=memory;
Initial import of WL#3726 "DDL locking for all metadata objects". Backport of: ------------------------------------------------------------ revno: 2630.4.1 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Fri 2008-05-23 17:54:03 +0400 message: WL#3726 "DDL locking for all metadata objects". After review fixes in progress. ------------------------------------------------------------ This is the first patch in series. It transforms the metadata locking subsystem to use a dedicated module (mdl.h,cc). No significant changes in the locking protocol. The import passes the test suite with the exception of deprecated/removed 6.0 features, and MERGE tables. The latter are subject to a fix by WL#4144. Unfortunately, the original changeset comments got lost in a merge, thus this import has its own (largely insufficient) comments. This patch fixes Bug#25144 "replication / binlog with view breaks". Warning: this patch introduces an incompatible change: Under LOCK TABLES, it's no longer possible to FLUSH a table that was not locked for WRITE. Under LOCK TABLES, it's no longer possible to DROP a table or VIEW that was not locked for WRITE. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.2 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 14:03:45 +0400 message: WL#3726 "DDL locking for all metadata objects". After review fixes in progress. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.3 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 14:08:51 +0400 message: WL#3726 "DDL locking for all metadata objects" Fixed failing Windows builds by adding mdl.cc to the lists of files needed to build server/libmysqld on Windows. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.4 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 21:57:58 +0400 message: WL#3726 "DDL locking for all metadata objects". Fix for assert failures in kill.test which occured when one tried to kill ALTER TABLE statement on merge table while it was waiting in wait_while_table_is_used() for other connections to close this table. These assert failures stemmed from the fact that cleanup code in this case assumed that temporary table representing new version of table was open with adding to THD::temporary_tables list while code which were opening this temporary table wasn't always fulfilling this. This patch changes code that opens new version of table to always do this linking in. It also streamlines cleanup process for cases when error occurs while we have new version of table open. ****** WL#3726 "DDL locking for all metadata objects" Add libmysqld/mdl.cc to .bzrignore. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.6 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sun 2008-05-25 00:33:22 +0400 message: WL#3726 "DDL locking for all metadata objects". Addition to the fix of assert failures in kill.test caused by changes for this worklog. Make sure we close the new table only once.
2009-11-30 18:55:03 +03:00
connection con2;
let $wait_condition=
select count(*) = 1 from information_schema.processlist
Initial import of WL#3726 "DDL locking for all metadata objects". Backport of: ------------------------------------------------------------ revno: 2630.4.1 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Fri 2008-05-23 17:54:03 +0400 message: WL#3726 "DDL locking for all metadata objects". After review fixes in progress. ------------------------------------------------------------ This is the first patch in series. It transforms the metadata locking subsystem to use a dedicated module (mdl.h,cc). No significant changes in the locking protocol. The import passes the test suite with the exception of deprecated/removed 6.0 features, and MERGE tables. The latter are subject to a fix by WL#4144. Unfortunately, the original changeset comments got lost in a merge, thus this import has its own (largely insufficient) comments. This patch fixes Bug#25144 "replication / binlog with view breaks". Warning: this patch introduces an incompatible change: Under LOCK TABLES, it's no longer possible to FLUSH a table that was not locked for WRITE. Under LOCK TABLES, it's no longer possible to DROP a table or VIEW that was not locked for WRITE. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.2 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 14:03:45 +0400 message: WL#3726 "DDL locking for all metadata objects". After review fixes in progress. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.3 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 14:08:51 +0400 message: WL#3726 "DDL locking for all metadata objects" Fixed failing Windows builds by adding mdl.cc to the lists of files needed to build server/libmysqld on Windows. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.4 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 21:57:58 +0400 message: WL#3726 "DDL locking for all metadata objects". Fix for assert failures in kill.test which occured when one tried to kill ALTER TABLE statement on merge table while it was waiting in wait_while_table_is_used() for other connections to close this table. These assert failures stemmed from the fact that cleanup code in this case assumed that temporary table representing new version of table was open with adding to THD::temporary_tables list while code which were opening this temporary table wasn't always fulfilling this. This patch changes code that opens new version of table to always do this linking in. It also streamlines cleanup process for cases when error occurs while we have new version of table open. ****** WL#3726 "DDL locking for all metadata objects" Add libmysqld/mdl.cc to .bzrignore. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.6 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sun 2008-05-25 00:33:22 +0400 message: WL#3726 "DDL locking for all metadata objects". Addition to the fix of assert failures in kill.test caused by changes for this worklog. Make sure we close the new table only once.
2009-11-30 18:55:03 +03:00
where state = "Waiting for table" and info = "alter table t1 engine=memory";
--source include/wait_condition.inc
Initial import of WL#3726 "DDL locking for all metadata objects". Backport of: ------------------------------------------------------------ revno: 2630.4.1 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Fri 2008-05-23 17:54:03 +0400 message: WL#3726 "DDL locking for all metadata objects". After review fixes in progress. ------------------------------------------------------------ This is the first patch in series. It transforms the metadata locking subsystem to use a dedicated module (mdl.h,cc). No significant changes in the locking protocol. The import passes the test suite with the exception of deprecated/removed 6.0 features, and MERGE tables. The latter are subject to a fix by WL#4144. Unfortunately, the original changeset comments got lost in a merge, thus this import has its own (largely insufficient) comments. This patch fixes Bug#25144 "replication / binlog with view breaks". Warning: this patch introduces an incompatible change: Under LOCK TABLES, it's no longer possible to FLUSH a table that was not locked for WRITE. Under LOCK TABLES, it's no longer possible to DROP a table or VIEW that was not locked for WRITE. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.2 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 14:03:45 +0400 message: WL#3726 "DDL locking for all metadata objects". After review fixes in progress. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.3 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 14:08:51 +0400 message: WL#3726 "DDL locking for all metadata objects" Fixed failing Windows builds by adding mdl.cc to the lists of files needed to build server/libmysqld on Windows. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.4 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 21:57:58 +0400 message: WL#3726 "DDL locking for all metadata objects". Fix for assert failures in kill.test which occured when one tried to kill ALTER TABLE statement on merge table while it was waiting in wait_while_table_is_used() for other connections to close this table. These assert failures stemmed from the fact that cleanup code in this case assumed that temporary table representing new version of table was open with adding to THD::temporary_tables list while code which were opening this temporary table wasn't always fulfilling this. This patch changes code that opens new version of table to always do this linking in. It also streamlines cleanup process for cases when error occurs while we have new version of table open. ****** WL#3726 "DDL locking for all metadata objects" Add libmysqld/mdl.cc to .bzrignore. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.6 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sun 2008-05-25 00:33:22 +0400 message: WL#3726 "DDL locking for all metadata objects". Addition to the fix of assert failures in kill.test caused by changes for this worklog. Make sure we close the new table only once.
2009-11-30 18:55:03 +03:00
connection default;
Implement new type-of-operation-aware metadata locks. Add a wait-for graph based deadlock detector to the MDL subsystem. Fixes bug #46272 "MySQL 5.4.4, new MDL: unnecessary deadlock" and bug #37346 "innodb does not detect deadlock between update and alter table". The first bug manifested itself as an unwarranted abort of a transaction with ER_LOCK_DEADLOCK error by a concurrent ALTER statement, when this transaction tried to repeat use of a table, which it has already used in a similar fashion before ALTER started. The second bug showed up as a deadlock between table-level locks and InnoDB row locks, which was "detected" only after innodb_lock_wait_timeout timeout. A transaction would start using the table and modify a few rows. Then ALTER TABLE would come in, and start copying rows into a temporary table. Eventually it would stumble on the modified records and get blocked on a row lock. The first transaction would try to do more updates, and get blocked on thr_lock.c lock. This situation of circular wait would only get resolved by a timeout. Both these bugs stemmed from inadequate solutions to the problem of deadlocks occurring between different locking subsystems. In the first case we tried to avoid deadlocks between metadata locking and table-level locking subsystems, when upgrading shared metadata lock to exclusive one. Transactions holding the shared lock on the table and waiting for some table-level lock used to be aborted too aggressively. We also allowed ALTER TABLE to start in presence of transactions that modify the subject table. ALTER TABLE acquires TL_WRITE_ALLOW_READ lock at start, and that block all writes against the table (naturally, we don't want any writes to be lost when switching the old and the new table). TL_WRITE_ALLOW_READ lock, in turn, would block the started transaction on thr_lock.c lock, should they do more updates. This, again, lead to the need to abort such transactions. The second bug occurred simply because we didn't have any mechanism to detect deadlocks between the table-level locks in thr_lock.c and row-level locks in InnoDB, other than innodb_lock_wait_timeout. This patch solves both these problems by moving lock conflicts which are causing these deadlocks into the metadata locking subsystem, thus making it possible to avoid or detect such deadlocks inside MDL. To do this we introduce new type-of-operation-aware metadata locks, which allow MDL subsystem to know not only the fact that transaction has used or is going to use some object but also what kind of operation it has carried out or going to carry out on the object. This, along with the addition of a special kind of upgradable metadata lock, allows ALTER TABLE to wait until all transactions which has updated the table to go away. This solves the second issue. Another special type of upgradable metadata lock is acquired by LOCK TABLE WRITE. This second lock type allows to solve the first issue, since abortion of table-level locks in event of DDL under LOCK TABLES becomes also unnecessary. Below follows the list of incompatible changes introduced by this patch: - From now on, ALTER TABLE and CREATE/DROP TRIGGER SQL (i.e. those statements that acquire TL_WRITE_ALLOW_READ lock) wait for all transactions which has *updated* the table to complete. - From now on, LOCK TABLES ... WRITE, REPAIR/OPTIMIZE TABLE (i.e. all statements which acquire TL_WRITE table-level lock) wait for all transaction which *updated or read* from the table to complete. As a consequence, innodb_table_locks=0 option no longer applies to LOCK TABLES ... WRITE. - DROP DATABASE, DROP TABLE, RENAME TABLE no longer abort statements or transactions which use tables being dropped or renamed, and instead wait for these transactions to complete. - Since LOCK TABLES WRITE now takes a special metadata lock, not compatible with with reads or writes against the subject table and transaction-wide, thr_lock.c deadlock avoidance algorithm that used to ensure absence of deadlocks between LOCK TABLES WRITE and other statements is no longer sufficient, even for MyISAM. The wait-for graph based deadlock detector of MDL subsystem may sometimes be necessary and is involved. This may lead to ER_LOCK_DEADLOCK error produced for multi-statement transactions even if these only use MyISAM: session 1: session 2: begin; update t1 ... lock table t2 write, t1 write; -- gets a lock on t2, blocks on t1 update t2 ... (ER_LOCK_DEADLOCK) - Finally, support of LOW_PRIORITY option for LOCK TABLES ... WRITE was abandoned. LOCK TABLE ... LOW_PRIORITY WRITE from now on has the same priority as the usual LOCK TABLE ... WRITE. SELECT HIGH PRIORITY no longer trumps LOCK TABLE ... WRITE in the wait queue. - We do not take upgradable metadata locks on implicitly locked tables. So if one has, say, a view v1 that uses table t1, and issues: LOCK TABLE v1 WRITE; FLUSH TABLE t1; -- (or just 'FLUSH TABLES'), an error is produced. In order to be able to perform DDL on a table under LOCK TABLES, the table must be locked explicitly in the LOCK TABLES list.
2010-02-01 14:43:06 +03:00
--error ER_ILLEGAL_HA
handler t1 read a next;
handler t1 close;
connection con1;
--reap
Implement new type-of-operation-aware metadata locks. Add a wait-for graph based deadlock detector to the MDL subsystem. Fixes bug #46272 "MySQL 5.4.4, new MDL: unnecessary deadlock" and bug #37346 "innodb does not detect deadlock between update and alter table". The first bug manifested itself as an unwarranted abort of a transaction with ER_LOCK_DEADLOCK error by a concurrent ALTER statement, when this transaction tried to repeat use of a table, which it has already used in a similar fashion before ALTER started. The second bug showed up as a deadlock between table-level locks and InnoDB row locks, which was "detected" only after innodb_lock_wait_timeout timeout. A transaction would start using the table and modify a few rows. Then ALTER TABLE would come in, and start copying rows into a temporary table. Eventually it would stumble on the modified records and get blocked on a row lock. The first transaction would try to do more updates, and get blocked on thr_lock.c lock. This situation of circular wait would only get resolved by a timeout. Both these bugs stemmed from inadequate solutions to the problem of deadlocks occurring between different locking subsystems. In the first case we tried to avoid deadlocks between metadata locking and table-level locking subsystems, when upgrading shared metadata lock to exclusive one. Transactions holding the shared lock on the table and waiting for some table-level lock used to be aborted too aggressively. We also allowed ALTER TABLE to start in presence of transactions that modify the subject table. ALTER TABLE acquires TL_WRITE_ALLOW_READ lock at start, and that block all writes against the table (naturally, we don't want any writes to be lost when switching the old and the new table). TL_WRITE_ALLOW_READ lock, in turn, would block the started transaction on thr_lock.c lock, should they do more updates. This, again, lead to the need to abort such transactions. The second bug occurred simply because we didn't have any mechanism to detect deadlocks between the table-level locks in thr_lock.c and row-level locks in InnoDB, other than innodb_lock_wait_timeout. This patch solves both these problems by moving lock conflicts which are causing these deadlocks into the metadata locking subsystem, thus making it possible to avoid or detect such deadlocks inside MDL. To do this we introduce new type-of-operation-aware metadata locks, which allow MDL subsystem to know not only the fact that transaction has used or is going to use some object but also what kind of operation it has carried out or going to carry out on the object. This, along with the addition of a special kind of upgradable metadata lock, allows ALTER TABLE to wait until all transactions which has updated the table to go away. This solves the second issue. Another special type of upgradable metadata lock is acquired by LOCK TABLE WRITE. This second lock type allows to solve the first issue, since abortion of table-level locks in event of DDL under LOCK TABLES becomes also unnecessary. Below follows the list of incompatible changes introduced by this patch: - From now on, ALTER TABLE and CREATE/DROP TRIGGER SQL (i.e. those statements that acquire TL_WRITE_ALLOW_READ lock) wait for all transactions which has *updated* the table to complete. - From now on, LOCK TABLES ... WRITE, REPAIR/OPTIMIZE TABLE (i.e. all statements which acquire TL_WRITE table-level lock) wait for all transaction which *updated or read* from the table to complete. As a consequence, innodb_table_locks=0 option no longer applies to LOCK TABLES ... WRITE. - DROP DATABASE, DROP TABLE, RENAME TABLE no longer abort statements or transactions which use tables being dropped or renamed, and instead wait for these transactions to complete. - Since LOCK TABLES WRITE now takes a special metadata lock, not compatible with with reads or writes against the subject table and transaction-wide, thr_lock.c deadlock avoidance algorithm that used to ensure absence of deadlocks between LOCK TABLES WRITE and other statements is no longer sufficient, even for MyISAM. The wait-for graph based deadlock detector of MDL subsystem may sometimes be necessary and is involved. This may lead to ER_LOCK_DEADLOCK error produced for multi-statement transactions even if these only use MyISAM: session 1: session 2: begin; update t1 ... lock table t2 write, t1 write; -- gets a lock on t2, blocks on t1 update t2 ... (ER_LOCK_DEADLOCK) - Finally, support of LOW_PRIORITY option for LOCK TABLES ... WRITE was abandoned. LOCK TABLE ... LOW_PRIORITY WRITE from now on has the same priority as the usual LOCK TABLE ... WRITE. SELECT HIGH PRIORITY no longer trumps LOCK TABLE ... WRITE in the wait queue. - We do not take upgradable metadata locks on implicitly locked tables. So if one has, say, a view v1 that uses table t1, and issues: LOCK TABLE v1 WRITE; FLUSH TABLE t1; -- (or just 'FLUSH TABLES'), an error is produced. In order to be able to perform DDL on a table under LOCK TABLES, the table must be locked explicitly in the LOCK TABLES list.
2010-02-01 14:43:06 +03:00
unlock tables;
drop table t1;
--echo # Now test case which was reported originally but which no longer
--echo # triggers execution path which has caused the problem.
connection default;
create table t1 (a int, key(a));
insert into t1 values (1);
handler t1 open;
connection con1;
send alter table t1 engine=memory;
connection con2;
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Waiting for table" and info = "alter table t1 engine=memory";
--source include/wait_condition.inc
connection default;
--echo # Since S metadata lock was already acquired at HANDLER OPEN time
--echo # and TL_READ lock requested by HANDLER READ is compatible with
--echo # ALTER's TL_WRITE_ALLOW_READ the below statement should succeed
--echo # without waiting. The old version of table should be used in it.
handler t1 read a next;
handler t1 close;
connection con1;
drop table t1;
disconnect con1;
--source include/wait_until_disconnected.inc
Initial import of WL#3726 "DDL locking for all metadata objects". Backport of: ------------------------------------------------------------ revno: 2630.4.1 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Fri 2008-05-23 17:54:03 +0400 message: WL#3726 "DDL locking for all metadata objects". After review fixes in progress. ------------------------------------------------------------ This is the first patch in series. It transforms the metadata locking subsystem to use a dedicated module (mdl.h,cc). No significant changes in the locking protocol. The import passes the test suite with the exception of deprecated/removed 6.0 features, and MERGE tables. The latter are subject to a fix by WL#4144. Unfortunately, the original changeset comments got lost in a merge, thus this import has its own (largely insufficient) comments. This patch fixes Bug#25144 "replication / binlog with view breaks". Warning: this patch introduces an incompatible change: Under LOCK TABLES, it's no longer possible to FLUSH a table that was not locked for WRITE. Under LOCK TABLES, it's no longer possible to DROP a table or VIEW that was not locked for WRITE. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.2 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 14:03:45 +0400 message: WL#3726 "DDL locking for all metadata objects". After review fixes in progress. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.3 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 14:08:51 +0400 message: WL#3726 "DDL locking for all metadata objects" Fixed failing Windows builds by adding mdl.cc to the lists of files needed to build server/libmysqld on Windows. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.4 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sat 2008-05-24 21:57:58 +0400 message: WL#3726 "DDL locking for all metadata objects". Fix for assert failures in kill.test which occured when one tried to kill ALTER TABLE statement on merge table while it was waiting in wait_while_table_is_used() for other connections to close this table. These assert failures stemmed from the fact that cleanup code in this case assumed that temporary table representing new version of table was open with adding to THD::temporary_tables list while code which were opening this temporary table wasn't always fulfilling this. This patch changes code that opens new version of table to always do this linking in. It also streamlines cleanup process for cases when error occurs while we have new version of table open. ****** WL#3726 "DDL locking for all metadata objects" Add libmysqld/mdl.cc to .bzrignore. ****** Backport of: ------------------------------------------------------------ revno: 2630.4.6 committer: Dmitry Lenev <dlenev@mysql.com> branch nick: mysql-6.0-3726-w timestamp: Sun 2008-05-25 00:33:22 +0400 message: WL#3726 "DDL locking for all metadata objects". Addition to the fix of assert failures in kill.test caused by changes for this worklog. Make sure we close the new table only once.
2009-11-30 18:55:03 +03:00
connection con2;
disconnect con2;
--source include/wait_until_disconnected.inc
connection default;
2009-04-17 13:46:27 +05:00
#
# Bug#44151 using handler commands on information_schema tables crashes server
#
USE information_schema;
--error ER_WRONG_USAGE
HANDLER COLUMNS OPEN;
USE test;
A prerequisite patch for the fix for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". Introduce a notion of a sentinel to MDL_context. A sentinel is a ticket that separates all tickets in the context into two groups: before and after it. Currently we can have (and need) only one designated sentinel -- it separates all locks taken by LOCK TABLE or HANDLER statement, which must survive COMMIT and ROLLBACK and all other locks, which must be released at COMMIT or ROLLBACK. The tricky part is maintaining the sentinel up to date when someone release its corresponding ticket. This can happen, e.g. if someone issues DROP TABLE under LOCK TABLES (generally, see all calls to release_all_locks_for_name()). MDL_context::release_ticket() is modified to take care of it. ****** A fix and a test case for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". An attempt to mix HANDLER SQL statements, which are transaction- agnostic, an open multi-statement transaction, and DDL against the involved tables (in a concurrent connection) could lead to a deadlock. The deadlock would occur when HANDLER OPEN or HANDLER READ would have to wait on a conflicting metadata lock. If the connection that issued HANDLER statement also had other metadata locks (say, acquired in scope of a transaction), a classical deadlock situation of mutual wait could occur. Incompatible change: entering LOCK TABLES mode automatically closes all open HANDLERs in the current connection. Incompatible change: previously an attempt to wait on a lock in a connection that has an open HANDLER statement could wait indefinitely/deadlock. After this patch, an error ER_LOCK_DEADLOCK is produced. The idea of the fix is to merge thd->handler_mdl_context with the main mdl_context of the connection, used for transactional locks. This makes deadlock detection possible, since all waits with locks are "visible" and available to analysis in a single MDL context of the connection. Since HANDLER locks and transactional locks have a different life cycle -- HANDLERs are explicitly open and closed, and so are HANDLER locks, explicitly acquired and released, whereas transactional locks "accumulate" till the end of a transaction and are released only with COMMIT, ROLLBACK and ROLLBACK TO SAVEPOINT, a concept of "sentinel" was introduced to MDL_context. All locks, HANDLER and others, reside in the same linked list. However, a selected element of the list separates locks with different life cycle. HANDLER locks always reside at the end of the list, after the sentinel. Transactional locks are prepended to the beginning of the list, before the sentinel. Thus, ROLLBACK, COMMIT or ROLLBACK TO SAVEPOINT, only release those locks that reside before the sentinel. HANDLER locks must be released explicitly as part of HANDLER CLOSE statement, or an implicit close. The same approach with sentinel is also employed for LOCK TABLES locks. Since HANDLER and LOCK TABLES statement has never worked together, the implementation is made simple and only maintains one sentinel, which is used either for HANDLER locks, or for LOCK TABLES locks.
2009-12-22 19:09:15 +03:00
--echo #
--echo # Add test coverage for HANDLER and LOCK TABLES, HANDLER and DDL.
--echo #
--disable_warnings
drop table if exists t1, t2, t3;
--enable_warnings
create table t1 (a int, key a (a));
insert into t1 (a) values (1), (2), (3), (4), (5);
create table t2 (a int, key a (a)) select * from t1;
create temporary table t3 (a int, key a (a)) select * from t2;
handler t1 open;
handler t2 open;
handler t3 open;
--echo #
--echo # LOCK TABLES implicitly closes all handlers.
--echo #
lock table t3 read;
--echo #
--echo # No HANDLER sql is available under lock tables anyway.
--echo #
--error ER_LOCK_OR_ACTIVE_TRANSACTION
handler t1 open;
--error ER_LOCK_OR_ACTIVE_TRANSACTION
handler t1 read next;
--error ER_LOCK_OR_ACTIVE_TRANSACTION
handler t2 close;
--error ER_LOCK_OR_ACTIVE_TRANSACTION
handler t3 open;
--echo # After UNLOCK TABLES no handlers are around, they were
--echo # implicitly closed.
unlock tables;
drop temporary table t3;
--error ER_UNKNOWN_TABLE
handler t1 read next;
--error ER_UNKNOWN_TABLE
handler t2 close;
--error ER_UNKNOWN_TABLE
handler t3 read next;
--echo #
--echo # Other operations also implicitly close handler:
--echo #
--echo # TRUNCATE
--echo #
handler t1 open;
truncate table t1;
--error ER_UNKNOWN_TABLE
handler t1 read next;
handler t1 open;
--echo #
--echo # CREATE TRIGGER
--echo #
create trigger t1_ai after insert on t1 for each row set @a=1;
--error ER_UNKNOWN_TABLE
handler t1 read next;
--echo #
--echo # DROP TRIGGER
--echo #
handler t1 open;
drop trigger t1_ai;
--error ER_UNKNOWN_TABLE
handler t1 read next;
--echo #
--echo # ALTER TABLE
--echo #
handler t1 open;
alter table t1 add column b int;
--error ER_UNKNOWN_TABLE
handler t1 read next;
--echo #
--echo # ANALYZE TABLE
--echo #
handler t1 open;
analyze table t1;
--error ER_UNKNOWN_TABLE
handler t1 read next;
--echo #
--echo # OPTIMIZE TABLE
--echo #
handler t1 open;
optimize table t1;
--error ER_UNKNOWN_TABLE
handler t1 read next;
--echo #
--echo # REPAIR TABLE
--echo #
handler t1 open;
repair table t1;
--error ER_UNKNOWN_TABLE
handler t1 read next;
--echo #
--echo # DROP TABLE, naturally.
--echo #
handler t1 open;
drop table t1;
--error ER_UNKNOWN_TABLE
handler t1 read next;
create table t1 (a int, b int, key a (a)) select a from t2;
--echo #
--echo # RENAME TABLE, naturally
--echo #
handler t1 open;
rename table t1 to t3;
--error ER_UNKNOWN_TABLE
handler t1 read next;
--echo #
--echo # CREATE TABLE (even with IF NOT EXISTS clause,
--echo # and the table exists).
--echo #
handler t2 open;
create table if not exists t2 (a int);
--error ER_UNKNOWN_TABLE
handler t2 read next;
rename table t3 to t1;
drop table t2;
--echo #
--echo # FLUSH TABLE doesn't close the table but loses the position
--echo #
handler t1 open;
handler t1 read a prev;
flush table t1;
handler t1 read a prev;
handler t1 close;
--echo #
--echo # FLUSH TABLES WITH READ LOCK behaves like FLUSH TABLE.
--echo #
handler t1 open;
handler t1 read a prev;
flush tables with read lock;
handler t1 read a prev;
handler t1 close;
unlock tables;
--echo #
--echo # Explore the effect of HANDLER locks on concurrent DDL
--echo #
handler t1 open;
--echo # Establishing auxiliary connections con1, con2, con3
connect(con1, localhost, root,,);
connect(con2, localhost, root,,);
connect(con3, localhost, root,,);
--echo # --> connection con1;
connection con1;
--echo # Sending:
--send drop table t1
--echo # We can't use connection 'default' as wait_condition will
--echo # autoclose handlers.
--echo # --> connection con2
connection con2;
--echo # Waitng for 'drop table t1' to get blocked...
let $wait_condition=select count(*)=1 from information_schema.processlist where state='Waiting for table' and info='drop table t1';
--source include/wait_condition.inc
--echo # --> connection default
connection default;
handler t1 read a prev;
handler t1 read a prev;
handler t1 close;
--echo # --> connection con1
connection con1;
--echo # Reaping 'drop table t1'...
--reap
--echo # --> connection default
connection default;
--echo #
--echo # Explore the effect of HANDLER locks in parallel with SELECT
--echo #
create table t1 (a int, key a (a));
insert into t1 (a) values (1), (2), (3), (4), (5);
begin;
select * from t1;
handler t1 open;
handler t1 read a prev;
handler t1 read a prev;
handler t1 close;
--echo # --> connection con1;
connection con1;
--echo # Sending:
--send drop table t1
--echo # --> connection con2
connection con2;
--echo # Waiting for 'drop table t1' to get blocked...
let $wait_condition=select count(*)=1 from information_schema.processlist where state='Waiting for table' and info='drop table t1';
--source include/wait_condition.inc
--echo # --> connection default
connection default;
--echo # We can still use the table, it's part of the transaction
select * from t1;
--echo # Such are the circumstances that t1 is a part of transaction,
--echo # thus we can reopen it in the handler
handler t1 open;
--echo # We can commit the transaction, it doesn't close the handler
--echo # and doesn't let DROP to proceed.
commit;
handler t1 read a prev;
handler t1 read a prev;
handler t1 read a prev;
handler t1 close;
--echo # --> connection con1
connection con1;
--echo # Now drop can proceed
--echo # Reaping 'drop table t1'...
--reap
--echo # --> connection default
connection default;
--echo #
--echo # Demonstrate that HANDLER locks and transaction locks
--echo # reside in the same context, and we don't back-off
--echo # when have transaction or handler locks.
--echo #
create table t1 (a int, key a (a));
insert into t1 (a) values (1), (2), (3), (4), (5);
create table t0 (a int, key a (a));
insert into t0 (a) values (1), (2), (3), (4), (5);
A prerequisite patch for the fix for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". Introduce a notion of a sentinel to MDL_context. A sentinel is a ticket that separates all tickets in the context into two groups: before and after it. Currently we can have (and need) only one designated sentinel -- it separates all locks taken by LOCK TABLE or HANDLER statement, which must survive COMMIT and ROLLBACK and all other locks, which must be released at COMMIT or ROLLBACK. The tricky part is maintaining the sentinel up to date when someone release its corresponding ticket. This can happen, e.g. if someone issues DROP TABLE under LOCK TABLES (generally, see all calls to release_all_locks_for_name()). MDL_context::release_ticket() is modified to take care of it. ****** A fix and a test case for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". An attempt to mix HANDLER SQL statements, which are transaction- agnostic, an open multi-statement transaction, and DDL against the involved tables (in a concurrent connection) could lead to a deadlock. The deadlock would occur when HANDLER OPEN or HANDLER READ would have to wait on a conflicting metadata lock. If the connection that issued HANDLER statement also had other metadata locks (say, acquired in scope of a transaction), a classical deadlock situation of mutual wait could occur. Incompatible change: entering LOCK TABLES mode automatically closes all open HANDLERs in the current connection. Incompatible change: previously an attempt to wait on a lock in a connection that has an open HANDLER statement could wait indefinitely/deadlock. After this patch, an error ER_LOCK_DEADLOCK is produced. The idea of the fix is to merge thd->handler_mdl_context with the main mdl_context of the connection, used for transactional locks. This makes deadlock detection possible, since all waits with locks are "visible" and available to analysis in a single MDL context of the connection. Since HANDLER locks and transactional locks have a different life cycle -- HANDLERs are explicitly open and closed, and so are HANDLER locks, explicitly acquired and released, whereas transactional locks "accumulate" till the end of a transaction and are released only with COMMIT, ROLLBACK and ROLLBACK TO SAVEPOINT, a concept of "sentinel" was introduced to MDL_context. All locks, HANDLER and others, reside in the same linked list. However, a selected element of the list separates locks with different life cycle. HANDLER locks always reside at the end of the list, after the sentinel. Transactional locks are prepended to the beginning of the list, before the sentinel. Thus, ROLLBACK, COMMIT or ROLLBACK TO SAVEPOINT, only release those locks that reside before the sentinel. HANDLER locks must be released explicitly as part of HANDLER CLOSE statement, or an implicit close. The same approach with sentinel is also employed for LOCK TABLES locks. Since HANDLER and LOCK TABLES statement has never worked together, the implementation is made simple and only maintains one sentinel, which is used either for HANDLER locks, or for LOCK TABLES locks.
2009-12-22 19:09:15 +03:00
begin;
select * from t1;
--echo # --> connection con2
connection con2;
--echo # Sending:
send rename table t0 to t3, t1 to t0, t3 to t1;
A prerequisite patch for the fix for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". Introduce a notion of a sentinel to MDL_context. A sentinel is a ticket that separates all tickets in the context into two groups: before and after it. Currently we can have (and need) only one designated sentinel -- it separates all locks taken by LOCK TABLE or HANDLER statement, which must survive COMMIT and ROLLBACK and all other locks, which must be released at COMMIT or ROLLBACK. The tricky part is maintaining the sentinel up to date when someone release its corresponding ticket. This can happen, e.g. if someone issues DROP TABLE under LOCK TABLES (generally, see all calls to release_all_locks_for_name()). MDL_context::release_ticket() is modified to take care of it. ****** A fix and a test case for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". An attempt to mix HANDLER SQL statements, which are transaction- agnostic, an open multi-statement transaction, and DDL against the involved tables (in a concurrent connection) could lead to a deadlock. The deadlock would occur when HANDLER OPEN or HANDLER READ would have to wait on a conflicting metadata lock. If the connection that issued HANDLER statement also had other metadata locks (say, acquired in scope of a transaction), a classical deadlock situation of mutual wait could occur. Incompatible change: entering LOCK TABLES mode automatically closes all open HANDLERs in the current connection. Incompatible change: previously an attempt to wait on a lock in a connection that has an open HANDLER statement could wait indefinitely/deadlock. After this patch, an error ER_LOCK_DEADLOCK is produced. The idea of the fix is to merge thd->handler_mdl_context with the main mdl_context of the connection, used for transactional locks. This makes deadlock detection possible, since all waits with locks are "visible" and available to analysis in a single MDL context of the connection. Since HANDLER locks and transactional locks have a different life cycle -- HANDLERs are explicitly open and closed, and so are HANDLER locks, explicitly acquired and released, whereas transactional locks "accumulate" till the end of a transaction and are released only with COMMIT, ROLLBACK and ROLLBACK TO SAVEPOINT, a concept of "sentinel" was introduced to MDL_context. All locks, HANDLER and others, reside in the same linked list. However, a selected element of the list separates locks with different life cycle. HANDLER locks always reside at the end of the list, after the sentinel. Transactional locks are prepended to the beginning of the list, before the sentinel. Thus, ROLLBACK, COMMIT or ROLLBACK TO SAVEPOINT, only release those locks that reside before the sentinel. HANDLER locks must be released explicitly as part of HANDLER CLOSE statement, or an implicit close. The same approach with sentinel is also employed for LOCK TABLES locks. Since HANDLER and LOCK TABLES statement has never worked together, the implementation is made simple and only maintains one sentinel, which is used either for HANDLER locks, or for LOCK TABLES locks.
2009-12-22 19:09:15 +03:00
--echo # --> connection con1
connection con1;
Implementation of simple deadlock detection for metadata locks. This change is supposed to reduce number of ER_LOCK_DEADLOCK errors which occur when multi-statement transaction encounters conflicting metadata lock in cases when waiting is possible. The idea is not to fail ER_LOCK_DEADLOCK error immediately when we encounter conflicting metadata lock. Instead we release all metadata locks acquired by current statement and start to wait until conflicting lock go away. To avoid deadlocks we use simple empiric which aborts waiting with ER_LOCK_DEADLOCK error if it turns out that somebody is waiting for metadata locks owned by this transaction. This patch also fixes bug #46273 "MySQL 5.4.4 new MDL: Bug#989 is not fully fixed in case of ALTER". The bug was that concurrent execution of UPDATE or MULTI-UPDATE statement as a part of multi-statement transaction that already has used table being updated and ALTER TABLE statement might have resulted of loss of isolation between this transaction and ALTER TABLE statement, which manifested itself as changes performed by ALTER TABLE becoming visible in transaction and wrong binary log order as a consequence. This problem occurred when UPDATE or MULTI-UPDATE's wait in mysql_lock_tables() call was aborted due to metadata lock upgrade performed by concurrent ALTER TABLE. After such abort all metadata locks held by transaction were released but transaction silently continued to be executed as if nothing has happened. We solve this problem by changing our code not to release all locks in such case. Instead we release only locks which were acquired by current statement and then try to reacquire them by restarting open/lock tables process. We piggyback on simple deadlock detector implementation since this change has to be done anyway for it.
2009-12-30 20:53:30 +03:00
--echo # Waiting for 'rename table ...' to get blocked...
let $wait_condition=select count(*)=1 from information_schema.processlist
where state='Waiting for table' and info='rename table t0 to t3, t1 to t0, t3 to t1';
A prerequisite patch for the fix for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". Introduce a notion of a sentinel to MDL_context. A sentinel is a ticket that separates all tickets in the context into two groups: before and after it. Currently we can have (and need) only one designated sentinel -- it separates all locks taken by LOCK TABLE or HANDLER statement, which must survive COMMIT and ROLLBACK and all other locks, which must be released at COMMIT or ROLLBACK. The tricky part is maintaining the sentinel up to date when someone release its corresponding ticket. This can happen, e.g. if someone issues DROP TABLE under LOCK TABLES (generally, see all calls to release_all_locks_for_name()). MDL_context::release_ticket() is modified to take care of it. ****** A fix and a test case for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". An attempt to mix HANDLER SQL statements, which are transaction- agnostic, an open multi-statement transaction, and DDL against the involved tables (in a concurrent connection) could lead to a deadlock. The deadlock would occur when HANDLER OPEN or HANDLER READ would have to wait on a conflicting metadata lock. If the connection that issued HANDLER statement also had other metadata locks (say, acquired in scope of a transaction), a classical deadlock situation of mutual wait could occur. Incompatible change: entering LOCK TABLES mode automatically closes all open HANDLERs in the current connection. Incompatible change: previously an attempt to wait on a lock in a connection that has an open HANDLER statement could wait indefinitely/deadlock. After this patch, an error ER_LOCK_DEADLOCK is produced. The idea of the fix is to merge thd->handler_mdl_context with the main mdl_context of the connection, used for transactional locks. This makes deadlock detection possible, since all waits with locks are "visible" and available to analysis in a single MDL context of the connection. Since HANDLER locks and transactional locks have a different life cycle -- HANDLERs are explicitly open and closed, and so are HANDLER locks, explicitly acquired and released, whereas transactional locks "accumulate" till the end of a transaction and are released only with COMMIT, ROLLBACK and ROLLBACK TO SAVEPOINT, a concept of "sentinel" was introduced to MDL_context. All locks, HANDLER and others, reside in the same linked list. However, a selected element of the list separates locks with different life cycle. HANDLER locks always reside at the end of the list, after the sentinel. Transactional locks are prepended to the beginning of the list, before the sentinel. Thus, ROLLBACK, COMMIT or ROLLBACK TO SAVEPOINT, only release those locks that reside before the sentinel. HANDLER locks must be released explicitly as part of HANDLER CLOSE statement, or an implicit close. The same approach with sentinel is also employed for LOCK TABLES locks. Since HANDLER and LOCK TABLES statement has never worked together, the implementation is made simple and only maintains one sentinel, which is used either for HANDLER locks, or for LOCK TABLES locks.
2009-12-22 19:09:15 +03:00
--source include/wait_condition.inc
--echo # --> connection default
connection default;
--error ER_LOCK_DEADLOCK
handler t0 open;
A prerequisite patch for the fix for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". Introduce a notion of a sentinel to MDL_context. A sentinel is a ticket that separates all tickets in the context into two groups: before and after it. Currently we can have (and need) only one designated sentinel -- it separates all locks taken by LOCK TABLE or HANDLER statement, which must survive COMMIT and ROLLBACK and all other locks, which must be released at COMMIT or ROLLBACK. The tricky part is maintaining the sentinel up to date when someone release its corresponding ticket. This can happen, e.g. if someone issues DROP TABLE under LOCK TABLES (generally, see all calls to release_all_locks_for_name()). MDL_context::release_ticket() is modified to take care of it. ****** A fix and a test case for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". An attempt to mix HANDLER SQL statements, which are transaction- agnostic, an open multi-statement transaction, and DDL against the involved tables (in a concurrent connection) could lead to a deadlock. The deadlock would occur when HANDLER OPEN or HANDLER READ would have to wait on a conflicting metadata lock. If the connection that issued HANDLER statement also had other metadata locks (say, acquired in scope of a transaction), a classical deadlock situation of mutual wait could occur. Incompatible change: entering LOCK TABLES mode automatically closes all open HANDLERs in the current connection. Incompatible change: previously an attempt to wait on a lock in a connection that has an open HANDLER statement could wait indefinitely/deadlock. After this patch, an error ER_LOCK_DEADLOCK is produced. The idea of the fix is to merge thd->handler_mdl_context with the main mdl_context of the connection, used for transactional locks. This makes deadlock detection possible, since all waits with locks are "visible" and available to analysis in a single MDL context of the connection. Since HANDLER locks and transactional locks have a different life cycle -- HANDLERs are explicitly open and closed, and so are HANDLER locks, explicitly acquired and released, whereas transactional locks "accumulate" till the end of a transaction and are released only with COMMIT, ROLLBACK and ROLLBACK TO SAVEPOINT, a concept of "sentinel" was introduced to MDL_context. All locks, HANDLER and others, reside in the same linked list. However, a selected element of the list separates locks with different life cycle. HANDLER locks always reside at the end of the list, after the sentinel. Transactional locks are prepended to the beginning of the list, before the sentinel. Thus, ROLLBACK, COMMIT or ROLLBACK TO SAVEPOINT, only release those locks that reside before the sentinel. HANDLER locks must be released explicitly as part of HANDLER CLOSE statement, or an implicit close. The same approach with sentinel is also employed for LOCK TABLES locks. Since HANDLER and LOCK TABLES statement has never worked together, the implementation is made simple and only maintains one sentinel, which is used either for HANDLER locks, or for LOCK TABLES locks.
2009-12-22 19:09:15 +03:00
--error ER_LOCK_DEADLOCK
select * from t0;
A prerequisite patch for the fix for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". Introduce a notion of a sentinel to MDL_context. A sentinel is a ticket that separates all tickets in the context into two groups: before and after it. Currently we can have (and need) only one designated sentinel -- it separates all locks taken by LOCK TABLE or HANDLER statement, which must survive COMMIT and ROLLBACK and all other locks, which must be released at COMMIT or ROLLBACK. The tricky part is maintaining the sentinel up to date when someone release its corresponding ticket. This can happen, e.g. if someone issues DROP TABLE under LOCK TABLES (generally, see all calls to release_all_locks_for_name()). MDL_context::release_ticket() is modified to take care of it. ****** A fix and a test case for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". An attempt to mix HANDLER SQL statements, which are transaction- agnostic, an open multi-statement transaction, and DDL against the involved tables (in a concurrent connection) could lead to a deadlock. The deadlock would occur when HANDLER OPEN or HANDLER READ would have to wait on a conflicting metadata lock. If the connection that issued HANDLER statement also had other metadata locks (say, acquired in scope of a transaction), a classical deadlock situation of mutual wait could occur. Incompatible change: entering LOCK TABLES mode automatically closes all open HANDLERs in the current connection. Incompatible change: previously an attempt to wait on a lock in a connection that has an open HANDLER statement could wait indefinitely/deadlock. After this patch, an error ER_LOCK_DEADLOCK is produced. The idea of the fix is to merge thd->handler_mdl_context with the main mdl_context of the connection, used for transactional locks. This makes deadlock detection possible, since all waits with locks are "visible" and available to analysis in a single MDL context of the connection. Since HANDLER locks and transactional locks have a different life cycle -- HANDLERs are explicitly open and closed, and so are HANDLER locks, explicitly acquired and released, whereas transactional locks "accumulate" till the end of a transaction and are released only with COMMIT, ROLLBACK and ROLLBACK TO SAVEPOINT, a concept of "sentinel" was introduced to MDL_context. All locks, HANDLER and others, reside in the same linked list. However, a selected element of the list separates locks with different life cycle. HANDLER locks always reside at the end of the list, after the sentinel. Transactional locks are prepended to the beginning of the list, before the sentinel. Thus, ROLLBACK, COMMIT or ROLLBACK TO SAVEPOINT, only release those locks that reside before the sentinel. HANDLER locks must be released explicitly as part of HANDLER CLOSE statement, or an implicit close. The same approach with sentinel is also employed for LOCK TABLES locks. Since HANDLER and LOCK TABLES statement has never worked together, the implementation is made simple and only maintains one sentinel, which is used either for HANDLER locks, or for LOCK TABLES locks.
2009-12-22 19:09:15 +03:00
handler t1 open;
commit;
handler t1 close;
--echo # --> connection con2
connection con2;
Implementation of simple deadlock detection for metadata locks. This change is supposed to reduce number of ER_LOCK_DEADLOCK errors which occur when multi-statement transaction encounters conflicting metadata lock in cases when waiting is possible. The idea is not to fail ER_LOCK_DEADLOCK error immediately when we encounter conflicting metadata lock. Instead we release all metadata locks acquired by current statement and start to wait until conflicting lock go away. To avoid deadlocks we use simple empiric which aborts waiting with ER_LOCK_DEADLOCK error if it turns out that somebody is waiting for metadata locks owned by this transaction. This patch also fixes bug #46273 "MySQL 5.4.4 new MDL: Bug#989 is not fully fixed in case of ALTER". The bug was that concurrent execution of UPDATE or MULTI-UPDATE statement as a part of multi-statement transaction that already has used table being updated and ALTER TABLE statement might have resulted of loss of isolation between this transaction and ALTER TABLE statement, which manifested itself as changes performed by ALTER TABLE becoming visible in transaction and wrong binary log order as a consequence. This problem occurred when UPDATE or MULTI-UPDATE's wait in mysql_lock_tables() call was aborted due to metadata lock upgrade performed by concurrent ALTER TABLE. After such abort all metadata locks held by transaction were released but transaction silently continued to be executed as if nothing has happened. We solve this problem by changing our code not to release all locks in such case. Instead we release only locks which were acquired by current statement and then try to reacquire them by restarting open/lock tables process. We piggyback on simple deadlock detector implementation since this change has to be done anyway for it.
2009-12-30 20:53:30 +03:00
--echo # Reaping 'rename table ...'...
A prerequisite patch for the fix for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". Introduce a notion of a sentinel to MDL_context. A sentinel is a ticket that separates all tickets in the context into two groups: before and after it. Currently we can have (and need) only one designated sentinel -- it separates all locks taken by LOCK TABLE or HANDLER statement, which must survive COMMIT and ROLLBACK and all other locks, which must be released at COMMIT or ROLLBACK. The tricky part is maintaining the sentinel up to date when someone release its corresponding ticket. This can happen, e.g. if someone issues DROP TABLE under LOCK TABLES (generally, see all calls to release_all_locks_for_name()). MDL_context::release_ticket() is modified to take care of it. ****** A fix and a test case for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". An attempt to mix HANDLER SQL statements, which are transaction- agnostic, an open multi-statement transaction, and DDL against the involved tables (in a concurrent connection) could lead to a deadlock. The deadlock would occur when HANDLER OPEN or HANDLER READ would have to wait on a conflicting metadata lock. If the connection that issued HANDLER statement also had other metadata locks (say, acquired in scope of a transaction), a classical deadlock situation of mutual wait could occur. Incompatible change: entering LOCK TABLES mode automatically closes all open HANDLERs in the current connection. Incompatible change: previously an attempt to wait on a lock in a connection that has an open HANDLER statement could wait indefinitely/deadlock. After this patch, an error ER_LOCK_DEADLOCK is produced. The idea of the fix is to merge thd->handler_mdl_context with the main mdl_context of the connection, used for transactional locks. This makes deadlock detection possible, since all waits with locks are "visible" and available to analysis in a single MDL context of the connection. Since HANDLER locks and transactional locks have a different life cycle -- HANDLERs are explicitly open and closed, and so are HANDLER locks, explicitly acquired and released, whereas transactional locks "accumulate" till the end of a transaction and are released only with COMMIT, ROLLBACK and ROLLBACK TO SAVEPOINT, a concept of "sentinel" was introduced to MDL_context. All locks, HANDLER and others, reside in the same linked list. However, a selected element of the list separates locks with different life cycle. HANDLER locks always reside at the end of the list, after the sentinel. Transactional locks are prepended to the beginning of the list, before the sentinel. Thus, ROLLBACK, COMMIT or ROLLBACK TO SAVEPOINT, only release those locks that reside before the sentinel. HANDLER locks must be released explicitly as part of HANDLER CLOSE statement, or an implicit close. The same approach with sentinel is also employed for LOCK TABLES locks. Since HANDLER and LOCK TABLES statement has never worked together, the implementation is made simple and only maintains one sentinel, which is used either for HANDLER locks, or for LOCK TABLES locks.
2009-12-22 19:09:15 +03:00
--reap
--echo # --> connection default
connection default;
handler t1 open;
handler t1 read a prev;
handler t1 close;
drop table t0;
A prerequisite patch for the fix for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". Introduce a notion of a sentinel to MDL_context. A sentinel is a ticket that separates all tickets in the context into two groups: before and after it. Currently we can have (and need) only one designated sentinel -- it separates all locks taken by LOCK TABLE or HANDLER statement, which must survive COMMIT and ROLLBACK and all other locks, which must be released at COMMIT or ROLLBACK. The tricky part is maintaining the sentinel up to date when someone release its corresponding ticket. This can happen, e.g. if someone issues DROP TABLE under LOCK TABLES (generally, see all calls to release_all_locks_for_name()). MDL_context::release_ticket() is modified to take care of it. ****** A fix and a test case for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". An attempt to mix HANDLER SQL statements, which are transaction- agnostic, an open multi-statement transaction, and DDL against the involved tables (in a concurrent connection) could lead to a deadlock. The deadlock would occur when HANDLER OPEN or HANDLER READ would have to wait on a conflicting metadata lock. If the connection that issued HANDLER statement also had other metadata locks (say, acquired in scope of a transaction), a classical deadlock situation of mutual wait could occur. Incompatible change: entering LOCK TABLES mode automatically closes all open HANDLERs in the current connection. Incompatible change: previously an attempt to wait on a lock in a connection that has an open HANDLER statement could wait indefinitely/deadlock. After this patch, an error ER_LOCK_DEADLOCK is produced. The idea of the fix is to merge thd->handler_mdl_context with the main mdl_context of the connection, used for transactional locks. This makes deadlock detection possible, since all waits with locks are "visible" and available to analysis in a single MDL context of the connection. Since HANDLER locks and transactional locks have a different life cycle -- HANDLERs are explicitly open and closed, and so are HANDLER locks, explicitly acquired and released, whereas transactional locks "accumulate" till the end of a transaction and are released only with COMMIT, ROLLBACK and ROLLBACK TO SAVEPOINT, a concept of "sentinel" was introduced to MDL_context. All locks, HANDLER and others, reside in the same linked list. However, a selected element of the list separates locks with different life cycle. HANDLER locks always reside at the end of the list, after the sentinel. Transactional locks are prepended to the beginning of the list, before the sentinel. Thus, ROLLBACK, COMMIT or ROLLBACK TO SAVEPOINT, only release those locks that reside before the sentinel. HANDLER locks must be released explicitly as part of HANDLER CLOSE statement, or an implicit close. The same approach with sentinel is also employed for LOCK TABLES locks. Since HANDLER and LOCK TABLES statement has never worked together, the implementation is made simple and only maintains one sentinel, which is used either for HANDLER locks, or for LOCK TABLES locks.
2009-12-22 19:09:15 +03:00
--echo #
Implementation of simple deadlock detection for metadata locks. This change is supposed to reduce number of ER_LOCK_DEADLOCK errors which occur when multi-statement transaction encounters conflicting metadata lock in cases when waiting is possible. The idea is not to fail ER_LOCK_DEADLOCK error immediately when we encounter conflicting metadata lock. Instead we release all metadata locks acquired by current statement and start to wait until conflicting lock go away. To avoid deadlocks we use simple empiric which aborts waiting with ER_LOCK_DEADLOCK error if it turns out that somebody is waiting for metadata locks owned by this transaction. This patch also fixes bug #46273 "MySQL 5.4.4 new MDL: Bug#989 is not fully fixed in case of ALTER". The bug was that concurrent execution of UPDATE or MULTI-UPDATE statement as a part of multi-statement transaction that already has used table being updated and ALTER TABLE statement might have resulted of loss of isolation between this transaction and ALTER TABLE statement, which manifested itself as changes performed by ALTER TABLE becoming visible in transaction and wrong binary log order as a consequence. This problem occurred when UPDATE or MULTI-UPDATE's wait in mysql_lock_tables() call was aborted due to metadata lock upgrade performed by concurrent ALTER TABLE. After such abort all metadata locks held by transaction were released but transaction silently continued to be executed as if nothing has happened. We solve this problem by changing our code not to release all locks in such case. Instead we release only locks which were acquired by current statement and then try to reacquire them by restarting open/lock tables process. We piggyback on simple deadlock detector implementation since this change has to be done anyway for it.
2009-12-30 20:53:30 +03:00
--echo # Originally there was a deadlock error in this test.
--echo # With implementation of deadlock detector
--echo # we no longer deadlock, but block and wait on a lock.
--echo # The HANDLER is auto-closed as soon as the connection
--echo # sees a pending conflicting lock against it.
A prerequisite patch for the fix for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". Introduce a notion of a sentinel to MDL_context. A sentinel is a ticket that separates all tickets in the context into two groups: before and after it. Currently we can have (and need) only one designated sentinel -- it separates all locks taken by LOCK TABLE or HANDLER statement, which must survive COMMIT and ROLLBACK and all other locks, which must be released at COMMIT or ROLLBACK. The tricky part is maintaining the sentinel up to date when someone release its corresponding ticket. This can happen, e.g. if someone issues DROP TABLE under LOCK TABLES (generally, see all calls to release_all_locks_for_name()). MDL_context::release_ticket() is modified to take care of it. ****** A fix and a test case for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". An attempt to mix HANDLER SQL statements, which are transaction- agnostic, an open multi-statement transaction, and DDL against the involved tables (in a concurrent connection) could lead to a deadlock. The deadlock would occur when HANDLER OPEN or HANDLER READ would have to wait on a conflicting metadata lock. If the connection that issued HANDLER statement also had other metadata locks (say, acquired in scope of a transaction), a classical deadlock situation of mutual wait could occur. Incompatible change: entering LOCK TABLES mode automatically closes all open HANDLERs in the current connection. Incompatible change: previously an attempt to wait on a lock in a connection that has an open HANDLER statement could wait indefinitely/deadlock. After this patch, an error ER_LOCK_DEADLOCK is produced. The idea of the fix is to merge thd->handler_mdl_context with the main mdl_context of the connection, used for transactional locks. This makes deadlock detection possible, since all waits with locks are "visible" and available to analysis in a single MDL context of the connection. Since HANDLER locks and transactional locks have a different life cycle -- HANDLERs are explicitly open and closed, and so are HANDLER locks, explicitly acquired and released, whereas transactional locks "accumulate" till the end of a transaction and are released only with COMMIT, ROLLBACK and ROLLBACK TO SAVEPOINT, a concept of "sentinel" was introduced to MDL_context. All locks, HANDLER and others, reside in the same linked list. However, a selected element of the list separates locks with different life cycle. HANDLER locks always reside at the end of the list, after the sentinel. Transactional locks are prepended to the beginning of the list, before the sentinel. Thus, ROLLBACK, COMMIT or ROLLBACK TO SAVEPOINT, only release those locks that reside before the sentinel. HANDLER locks must be released explicitly as part of HANDLER CLOSE statement, or an implicit close. The same approach with sentinel is also employed for LOCK TABLES locks. Since HANDLER and LOCK TABLES statement has never worked together, the implementation is made simple and only maintains one sentinel, which is used either for HANDLER locks, or for LOCK TABLES locks.
2009-12-22 19:09:15 +03:00
--echo #
create table t2 (a int, key a (a));
handler t1 open;
--echo # --> connection con1
connection con1;
lock tables t2 read;
--echo # --> connection con2
connection con2;
--echo # Sending 'drop table t2'...
--send drop table t2
--echo # --> connection con1
connection con1;
--echo # Waiting for 'drop table t2' to get blocked...
let $wait_condition=select count(*)=1 from information_schema.processlist where state='Waiting for table' and info='drop table t2';
--source include/wait_condition.inc
--echo # --> connection default
connection default;
Implementation of simple deadlock detection for metadata locks. This change is supposed to reduce number of ER_LOCK_DEADLOCK errors which occur when multi-statement transaction encounters conflicting metadata lock in cases when waiting is possible. The idea is not to fail ER_LOCK_DEADLOCK error immediately when we encounter conflicting metadata lock. Instead we release all metadata locks acquired by current statement and start to wait until conflicting lock go away. To avoid deadlocks we use simple empiric which aborts waiting with ER_LOCK_DEADLOCK error if it turns out that somebody is waiting for metadata locks owned by this transaction. This patch also fixes bug #46273 "MySQL 5.4.4 new MDL: Bug#989 is not fully fixed in case of ALTER". The bug was that concurrent execution of UPDATE or MULTI-UPDATE statement as a part of multi-statement transaction that already has used table being updated and ALTER TABLE statement might have resulted of loss of isolation between this transaction and ALTER TABLE statement, which manifested itself as changes performed by ALTER TABLE becoming visible in transaction and wrong binary log order as a consequence. This problem occurred when UPDATE or MULTI-UPDATE's wait in mysql_lock_tables() call was aborted due to metadata lock upgrade performed by concurrent ALTER TABLE. After such abort all metadata locks held by transaction were released but transaction silently continued to be executed as if nothing has happened. We solve this problem by changing our code not to release all locks in such case. Instead we release only locks which were acquired by current statement and then try to reacquire them by restarting open/lock tables process. We piggyback on simple deadlock detector implementation since this change has to be done anyway for it.
2009-12-30 20:53:30 +03:00
--echo # Sending 'select * from t2'
send select * from t2;
A prerequisite patch for the fix for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". Introduce a notion of a sentinel to MDL_context. A sentinel is a ticket that separates all tickets in the context into two groups: before and after it. Currently we can have (and need) only one designated sentinel -- it separates all locks taken by LOCK TABLE or HANDLER statement, which must survive COMMIT and ROLLBACK and all other locks, which must be released at COMMIT or ROLLBACK. The tricky part is maintaining the sentinel up to date when someone release its corresponding ticket. This can happen, e.g. if someone issues DROP TABLE under LOCK TABLES (generally, see all calls to release_all_locks_for_name()). MDL_context::release_ticket() is modified to take care of it. ****** A fix and a test case for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". An attempt to mix HANDLER SQL statements, which are transaction- agnostic, an open multi-statement transaction, and DDL against the involved tables (in a concurrent connection) could lead to a deadlock. The deadlock would occur when HANDLER OPEN or HANDLER READ would have to wait on a conflicting metadata lock. If the connection that issued HANDLER statement also had other metadata locks (say, acquired in scope of a transaction), a classical deadlock situation of mutual wait could occur. Incompatible change: entering LOCK TABLES mode automatically closes all open HANDLERs in the current connection. Incompatible change: previously an attempt to wait on a lock in a connection that has an open HANDLER statement could wait indefinitely/deadlock. After this patch, an error ER_LOCK_DEADLOCK is produced. The idea of the fix is to merge thd->handler_mdl_context with the main mdl_context of the connection, used for transactional locks. This makes deadlock detection possible, since all waits with locks are "visible" and available to analysis in a single MDL context of the connection. Since HANDLER locks and transactional locks have a different life cycle -- HANDLERs are explicitly open and closed, and so are HANDLER locks, explicitly acquired and released, whereas transactional locks "accumulate" till the end of a transaction and are released only with COMMIT, ROLLBACK and ROLLBACK TO SAVEPOINT, a concept of "sentinel" was introduced to MDL_context. All locks, HANDLER and others, reside in the same linked list. However, a selected element of the list separates locks with different life cycle. HANDLER locks always reside at the end of the list, after the sentinel. Transactional locks are prepended to the beginning of the list, before the sentinel. Thus, ROLLBACK, COMMIT or ROLLBACK TO SAVEPOINT, only release those locks that reside before the sentinel. HANDLER locks must be released explicitly as part of HANDLER CLOSE statement, or an implicit close. The same approach with sentinel is also employed for LOCK TABLES locks. Since HANDLER and LOCK TABLES statement has never worked together, the implementation is made simple and only maintains one sentinel, which is used either for HANDLER locks, or for LOCK TABLES locks.
2009-12-22 19:09:15 +03:00
--echo # --> connection con1
connection con1;
Implementation of simple deadlock detection for metadata locks. This change is supposed to reduce number of ER_LOCK_DEADLOCK errors which occur when multi-statement transaction encounters conflicting metadata lock in cases when waiting is possible. The idea is not to fail ER_LOCK_DEADLOCK error immediately when we encounter conflicting metadata lock. Instead we release all metadata locks acquired by current statement and start to wait until conflicting lock go away. To avoid deadlocks we use simple empiric which aborts waiting with ER_LOCK_DEADLOCK error if it turns out that somebody is waiting for metadata locks owned by this transaction. This patch also fixes bug #46273 "MySQL 5.4.4 new MDL: Bug#989 is not fully fixed in case of ALTER". The bug was that concurrent execution of UPDATE or MULTI-UPDATE statement as a part of multi-statement transaction that already has used table being updated and ALTER TABLE statement might have resulted of loss of isolation between this transaction and ALTER TABLE statement, which manifested itself as changes performed by ALTER TABLE becoming visible in transaction and wrong binary log order as a consequence. This problem occurred when UPDATE or MULTI-UPDATE's wait in mysql_lock_tables() call was aborted due to metadata lock upgrade performed by concurrent ALTER TABLE. After such abort all metadata locks held by transaction were released but transaction silently continued to be executed as if nothing has happened. We solve this problem by changing our code not to release all locks in such case. Instead we release only locks which were acquired by current statement and then try to reacquire them by restarting open/lock tables process. We piggyback on simple deadlock detector implementation since this change has to be done anyway for it.
2009-12-30 20:53:30 +03:00
--echo # Waiting for 'select * from t2' to get blocked...
let $wait_condition=select count(*)=1 from information_schema.processlist where state='Waiting for table' and info='select * from t2';
A prerequisite patch for the fix for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". Introduce a notion of a sentinel to MDL_context. A sentinel is a ticket that separates all tickets in the context into two groups: before and after it. Currently we can have (and need) only one designated sentinel -- it separates all locks taken by LOCK TABLE or HANDLER statement, which must survive COMMIT and ROLLBACK and all other locks, which must be released at COMMIT or ROLLBACK. The tricky part is maintaining the sentinel up to date when someone release its corresponding ticket. This can happen, e.g. if someone issues DROP TABLE under LOCK TABLES (generally, see all calls to release_all_locks_for_name()). MDL_context::release_ticket() is modified to take care of it. ****** A fix and a test case for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". An attempt to mix HANDLER SQL statements, which are transaction- agnostic, an open multi-statement transaction, and DDL against the involved tables (in a concurrent connection) could lead to a deadlock. The deadlock would occur when HANDLER OPEN or HANDLER READ would have to wait on a conflicting metadata lock. If the connection that issued HANDLER statement also had other metadata locks (say, acquired in scope of a transaction), a classical deadlock situation of mutual wait could occur. Incompatible change: entering LOCK TABLES mode automatically closes all open HANDLERs in the current connection. Incompatible change: previously an attempt to wait on a lock in a connection that has an open HANDLER statement could wait indefinitely/deadlock. After this patch, an error ER_LOCK_DEADLOCK is produced. The idea of the fix is to merge thd->handler_mdl_context with the main mdl_context of the connection, used for transactional locks. This makes deadlock detection possible, since all waits with locks are "visible" and available to analysis in a single MDL context of the connection. Since HANDLER locks and transactional locks have a different life cycle -- HANDLERs are explicitly open and closed, and so are HANDLER locks, explicitly acquired and released, whereas transactional locks "accumulate" till the end of a transaction and are released only with COMMIT, ROLLBACK and ROLLBACK TO SAVEPOINT, a concept of "sentinel" was introduced to MDL_context. All locks, HANDLER and others, reside in the same linked list. However, a selected element of the list separates locks with different life cycle. HANDLER locks always reside at the end of the list, after the sentinel. Transactional locks are prepended to the beginning of the list, before the sentinel. Thus, ROLLBACK, COMMIT or ROLLBACK TO SAVEPOINT, only release those locks that reside before the sentinel. HANDLER locks must be released explicitly as part of HANDLER CLOSE statement, or an implicit close. The same approach with sentinel is also employed for LOCK TABLES locks. Since HANDLER and LOCK TABLES statement has never worked together, the implementation is made simple and only maintains one sentinel, which is used either for HANDLER locks, or for LOCK TABLES locks.
2009-12-22 19:09:15 +03:00
unlock tables;
--echo # --> connection con2
connection con2;
--echo # Reaping 'drop table t2'...
--reap
--echo # --> connection default
connection default;
Implementation of simple deadlock detection for metadata locks. This change is supposed to reduce number of ER_LOCK_DEADLOCK errors which occur when multi-statement transaction encounters conflicting metadata lock in cases when waiting is possible. The idea is not to fail ER_LOCK_DEADLOCK error immediately when we encounter conflicting metadata lock. Instead we release all metadata locks acquired by current statement and start to wait until conflicting lock go away. To avoid deadlocks we use simple empiric which aborts waiting with ER_LOCK_DEADLOCK error if it turns out that somebody is waiting for metadata locks owned by this transaction. This patch also fixes bug #46273 "MySQL 5.4.4 new MDL: Bug#989 is not fully fixed in case of ALTER". The bug was that concurrent execution of UPDATE or MULTI-UPDATE statement as a part of multi-statement transaction that already has used table being updated and ALTER TABLE statement might have resulted of loss of isolation between this transaction and ALTER TABLE statement, which manifested itself as changes performed by ALTER TABLE becoming visible in transaction and wrong binary log order as a consequence. This problem occurred when UPDATE or MULTI-UPDATE's wait in mysql_lock_tables() call was aborted due to metadata lock upgrade performed by concurrent ALTER TABLE. After such abort all metadata locks held by transaction were released but transaction silently continued to be executed as if nothing has happened. We solve this problem by changing our code not to release all locks in such case. Instead we release only locks which were acquired by current statement and then try to reacquire them by restarting open/lock tables process. We piggyback on simple deadlock detector implementation since this change has to be done anyway for it.
2009-12-30 20:53:30 +03:00
--echo # Reaping 'select * from t2'
--error ER_NO_SUCH_TABLE
reap;
A prerequisite patch for the fix for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". Introduce a notion of a sentinel to MDL_context. A sentinel is a ticket that separates all tickets in the context into two groups: before and after it. Currently we can have (and need) only one designated sentinel -- it separates all locks taken by LOCK TABLE or HANDLER statement, which must survive COMMIT and ROLLBACK and all other locks, which must be released at COMMIT or ROLLBACK. The tricky part is maintaining the sentinel up to date when someone release its corresponding ticket. This can happen, e.g. if someone issues DROP TABLE under LOCK TABLES (generally, see all calls to release_all_locks_for_name()). MDL_context::release_ticket() is modified to take care of it. ****** A fix and a test case for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". An attempt to mix HANDLER SQL statements, which are transaction- agnostic, an open multi-statement transaction, and DDL against the involved tables (in a concurrent connection) could lead to a deadlock. The deadlock would occur when HANDLER OPEN or HANDLER READ would have to wait on a conflicting metadata lock. If the connection that issued HANDLER statement also had other metadata locks (say, acquired in scope of a transaction), a classical deadlock situation of mutual wait could occur. Incompatible change: entering LOCK TABLES mode automatically closes all open HANDLERs in the current connection. Incompatible change: previously an attempt to wait on a lock in a connection that has an open HANDLER statement could wait indefinitely/deadlock. After this patch, an error ER_LOCK_DEADLOCK is produced. The idea of the fix is to merge thd->handler_mdl_context with the main mdl_context of the connection, used for transactional locks. This makes deadlock detection possible, since all waits with locks are "visible" and available to analysis in a single MDL context of the connection. Since HANDLER locks and transactional locks have a different life cycle -- HANDLERs are explicitly open and closed, and so are HANDLER locks, explicitly acquired and released, whereas transactional locks "accumulate" till the end of a transaction and are released only with COMMIT, ROLLBACK and ROLLBACK TO SAVEPOINT, a concept of "sentinel" was introduced to MDL_context. All locks, HANDLER and others, reside in the same linked list. However, a selected element of the list separates locks with different life cycle. HANDLER locks always reside at the end of the list, after the sentinel. Transactional locks are prepended to the beginning of the list, before the sentinel. Thus, ROLLBACK, COMMIT or ROLLBACK TO SAVEPOINT, only release those locks that reside before the sentinel. HANDLER locks must be released explicitly as part of HANDLER CLOSE statement, or an implicit close. The same approach with sentinel is also employed for LOCK TABLES locks. Since HANDLER and LOCK TABLES statement has never worked together, the implementation is made simple and only maintains one sentinel, which is used either for HANDLER locks, or for LOCK TABLES locks.
2009-12-22 19:09:15 +03:00
handler t1 close;
--echo #
--echo # ROLLBACK TO SAVEPOINT releases transactional locks,
--echo # but has no effect on open HANDLERs
--echo #
create table t2 like t1;
create table t3 like t1;
begin;
--echo # Have something before the savepoint
select * from t3;
savepoint sv;
handler t1 open;
handler t1 read a first;
handler t1 read a next;
select * from t2;
--echo # --> connection con1
connection con1;
--echo # Sending:
--send drop table t1
--echo # --> connection con2
connection con2;
--echo # Sending:
--send drop table t2
--echo # --> connection default
connection default;
--echo # Let DROP TABLE statements sync in. We must use
--echo # a separate connection for that, because otherwise SELECT
--echo # will auto-close the HANDLERs, becaues there are pending
--echo # exclusive locks against them.
--echo # --> connection con3
connection con3;
--echo # Waiting for 'drop table t1' to get blocked...
let $wait_condition=select count(*)=1 from information_schema.processlist where state='Waiting for table' and info='drop table t1';
--source include/wait_condition.inc
--echo # Waiting for 'drop table t2' to get blocked...
let $wait_condition=select count(*)=1 from information_schema.processlist where state='Waiting for table' and info='drop table t2';
--source include/wait_condition.inc
--echo # Demonstrate that t2 lock was released and t2 was dropped
--echo # after ROLLBACK TO SAVEPOINT
--echo # --> connection default
connection default;
rollback to savepoint sv;
--echo # --> connection con2
connection con2;
--echo # Reaping 'drop table t2'...
--reap
--echo # Demonstrate that ROLLBACK TO SAVEPOINT didn't release the handler
--echo # lock.
--echo # --> connection default
connection default;
handler t1 read a next;
handler t1 read a next;
--echo # Demonstrate that the drop will go through as soon as we close the
--echo # HANDLER
handler t1 close;
--echo # connection con1
connection con1;
--echo # Reaping 'drop table t1'...
--reap
--echo # --> connection default
connection default;
commit;
drop table t3;
--echo #
--echo # A few special cases when using SAVEPOINT/ROLLBACK TO
--echo # SAVEPOINT and HANDLER.
--echo #
--echo # Show that rollback to the savepoint taken in the beginning
--echo # of the transaction doesn't release mdl lock on
--echo # the HANDLER that was opened later.
--echo #
create table t1 (a int, key a(a));
insert into t1 (a) values (1), (2), (3), (4), (5);
create table t2 like t1;
begin;
savepoint sv;
handler t1 open;
handler t1 read a first;
handler t1 read a next;
select * from t2;
--echo # --> connection con1
connection con1;
--echo # Sending:
--send drop table t1
--echo # --> connection con2
connection con2;
--echo # Sending:
--send drop table t2
--echo # --> connection default
connection default;
--echo # Let DROP TABLE statements sync in. We must use
--echo # a separate connection for that, because otherwise SELECT
--echo # will auto-close the HANDLERs, becaues there are pending
--echo # exclusive locks against them.
--echo # --> connection con3
connection con3;
--echo # Waiting for 'drop table t1' to get blocked...
let $wait_condition=select count(*)=1 from information_schema.processlist where state='Waiting for table' and info='drop table t1';
--source include/wait_condition.inc
--echo # Waiting for 'drop table t2' to get blocked...
let $wait_condition=select count(*)=1 from information_schema.processlist where state='Waiting for table' and info='drop table t2';
--source include/wait_condition.inc
--echo # Demonstrate that t2 lock was released and t2 was dropped
--echo # after ROLLBACK TO SAVEPOINT
--echo # --> connection default
connection default;
rollback to savepoint sv;
--echo # --> connection con2
connection con2;
--echo # Reaping 'drop table t2'...
--reap
--echo # Demonstrate that ROLLBACK TO SAVEPOINT didn't release the handler
--echo # lock.
--echo # --> connection default
connection default;
handler t1 read a next;
handler t1 read a next;
--echo # Demonstrate that the drop will go through as soon as we close the
--echo # HANDLER
handler t1 close;
--echo # connection con1
connection con1;
--echo # Reaping 'drop table t1'...
--reap
--echo # --> connection default
connection default;
commit;
--echo #
--echo # Show that rollback to the savepoint taken in the beginning
--echo # of the transaction works properly (no valgrind warnins, etc),
--echo # even though it's done after the HANDLER mdl lock that was there
--echo # at the beginning is released and added again.
--echo #
create table t1 (a int, key a(a));
insert into t1 (a) values (1), (2), (3), (4), (5);
create table t2 like t1;
create table t3 like t1;
insert into t3 (a) select a from t1;
begin;
handler t1 open;
savepoint sv;
handler t1 read a first;
select * from t2;
handler t1 close;
handler t3 open;
handler t3 read a first;
rollback to savepoint sv;
--echo # --> connection con1
connection con1;
drop table t1, t2;
--echo # Sending:
--send drop table t3
--echo # Let DROP TABLE statement sync in.
--echo # --> connection con2
connection con2;
--echo # Waiting for 'drop table t3' to get blocked...
let $wait_condition=select count(*)=1 from information_schema.processlist where state='Waiting for table' and info='drop table t3';
--source include/wait_condition.inc
--echo # Demonstrate that ROLLBACK TO SAVEPOINT didn't release the handler
--echo # lock.
--echo # --> connection default
connection default;
handler t3 read a next;
--echo # Demonstrate that the drop will go through as soon as we close the
--echo # HANDLER
handler t3 close;
--echo # connection con1
connection con1;
--echo # Reaping 'drop table t3'...
--reap
--echo # --> connection default
connection default;
commit;
--echo #
--echo # If we have to wait on an exclusive locks while having
--echo # an open HANDLER, ER_LOCK_DEADLOCK is reported.
--echo #
create table t1 (a int, key a(a));
create table t2 like t1;
handler t1 open;
--echo # --> connection con1
connection con1;
Implement new type-of-operation-aware metadata locks. Add a wait-for graph based deadlock detector to the MDL subsystem. Fixes bug #46272 "MySQL 5.4.4, new MDL: unnecessary deadlock" and bug #37346 "innodb does not detect deadlock between update and alter table". The first bug manifested itself as an unwarranted abort of a transaction with ER_LOCK_DEADLOCK error by a concurrent ALTER statement, when this transaction tried to repeat use of a table, which it has already used in a similar fashion before ALTER started. The second bug showed up as a deadlock between table-level locks and InnoDB row locks, which was "detected" only after innodb_lock_wait_timeout timeout. A transaction would start using the table and modify a few rows. Then ALTER TABLE would come in, and start copying rows into a temporary table. Eventually it would stumble on the modified records and get blocked on a row lock. The first transaction would try to do more updates, and get blocked on thr_lock.c lock. This situation of circular wait would only get resolved by a timeout. Both these bugs stemmed from inadequate solutions to the problem of deadlocks occurring between different locking subsystems. In the first case we tried to avoid deadlocks between metadata locking and table-level locking subsystems, when upgrading shared metadata lock to exclusive one. Transactions holding the shared lock on the table and waiting for some table-level lock used to be aborted too aggressively. We also allowed ALTER TABLE to start in presence of transactions that modify the subject table. ALTER TABLE acquires TL_WRITE_ALLOW_READ lock at start, and that block all writes against the table (naturally, we don't want any writes to be lost when switching the old and the new table). TL_WRITE_ALLOW_READ lock, in turn, would block the started transaction on thr_lock.c lock, should they do more updates. This, again, lead to the need to abort such transactions. The second bug occurred simply because we didn't have any mechanism to detect deadlocks between the table-level locks in thr_lock.c and row-level locks in InnoDB, other than innodb_lock_wait_timeout. This patch solves both these problems by moving lock conflicts which are causing these deadlocks into the metadata locking subsystem, thus making it possible to avoid or detect such deadlocks inside MDL. To do this we introduce new type-of-operation-aware metadata locks, which allow MDL subsystem to know not only the fact that transaction has used or is going to use some object but also what kind of operation it has carried out or going to carry out on the object. This, along with the addition of a special kind of upgradable metadata lock, allows ALTER TABLE to wait until all transactions which has updated the table to go away. This solves the second issue. Another special type of upgradable metadata lock is acquired by LOCK TABLE WRITE. This second lock type allows to solve the first issue, since abortion of table-level locks in event of DDL under LOCK TABLES becomes also unnecessary. Below follows the list of incompatible changes introduced by this patch: - From now on, ALTER TABLE and CREATE/DROP TRIGGER SQL (i.e. those statements that acquire TL_WRITE_ALLOW_READ lock) wait for all transactions which has *updated* the table to complete. - From now on, LOCK TABLES ... WRITE, REPAIR/OPTIMIZE TABLE (i.e. all statements which acquire TL_WRITE table-level lock) wait for all transaction which *updated or read* from the table to complete. As a consequence, innodb_table_locks=0 option no longer applies to LOCK TABLES ... WRITE. - DROP DATABASE, DROP TABLE, RENAME TABLE no longer abort statements or transactions which use tables being dropped or renamed, and instead wait for these transactions to complete. - Since LOCK TABLES WRITE now takes a special metadata lock, not compatible with with reads or writes against the subject table and transaction-wide, thr_lock.c deadlock avoidance algorithm that used to ensure absence of deadlocks between LOCK TABLES WRITE and other statements is no longer sufficient, even for MyISAM. The wait-for graph based deadlock detector of MDL subsystem may sometimes be necessary and is involved. This may lead to ER_LOCK_DEADLOCK error produced for multi-statement transactions even if these only use MyISAM: session 1: session 2: begin; update t1 ... lock table t2 write, t1 write; -- gets a lock on t2, blocks on t1 update t2 ... (ER_LOCK_DEADLOCK) - Finally, support of LOW_PRIORITY option for LOCK TABLES ... WRITE was abandoned. LOCK TABLE ... LOW_PRIORITY WRITE from now on has the same priority as the usual LOCK TABLE ... WRITE. SELECT HIGH PRIORITY no longer trumps LOCK TABLE ... WRITE in the wait queue. - We do not take upgradable metadata locks on implicitly locked tables. So if one has, say, a view v1 that uses table t1, and issues: LOCK TABLE v1 WRITE; FLUSH TABLE t1; -- (or just 'FLUSH TABLES'), an error is produced. In order to be able to perform DDL on a table under LOCK TABLES, the table must be locked explicitly in the LOCK TABLES list.
2010-02-01 14:43:06 +03:00
lock table t1 write, t2 write;
A prerequisite patch for the fix for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". Introduce a notion of a sentinel to MDL_context. A sentinel is a ticket that separates all tickets in the context into two groups: before and after it. Currently we can have (and need) only one designated sentinel -- it separates all locks taken by LOCK TABLE or HANDLER statement, which must survive COMMIT and ROLLBACK and all other locks, which must be released at COMMIT or ROLLBACK. The tricky part is maintaining the sentinel up to date when someone release its corresponding ticket. This can happen, e.g. if someone issues DROP TABLE under LOCK TABLES (generally, see all calls to release_all_locks_for_name()). MDL_context::release_ticket() is modified to take care of it. ****** A fix and a test case for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". An attempt to mix HANDLER SQL statements, which are transaction- agnostic, an open multi-statement transaction, and DDL against the involved tables (in a concurrent connection) could lead to a deadlock. The deadlock would occur when HANDLER OPEN or HANDLER READ would have to wait on a conflicting metadata lock. If the connection that issued HANDLER statement also had other metadata locks (say, acquired in scope of a transaction), a classical deadlock situation of mutual wait could occur. Incompatible change: entering LOCK TABLES mode automatically closes all open HANDLERs in the current connection. Incompatible change: previously an attempt to wait on a lock in a connection that has an open HANDLER statement could wait indefinitely/deadlock. After this patch, an error ER_LOCK_DEADLOCK is produced. The idea of the fix is to merge thd->handler_mdl_context with the main mdl_context of the connection, used for transactional locks. This makes deadlock detection possible, since all waits with locks are "visible" and available to analysis in a single MDL context of the connection. Since HANDLER locks and transactional locks have a different life cycle -- HANDLERs are explicitly open and closed, and so are HANDLER locks, explicitly acquired and released, whereas transactional locks "accumulate" till the end of a transaction and are released only with COMMIT, ROLLBACK and ROLLBACK TO SAVEPOINT, a concept of "sentinel" was introduced to MDL_context. All locks, HANDLER and others, reside in the same linked list. However, a selected element of the list separates locks with different life cycle. HANDLER locks always reside at the end of the list, after the sentinel. Transactional locks are prepended to the beginning of the list, before the sentinel. Thus, ROLLBACK, COMMIT or ROLLBACK TO SAVEPOINT, only release those locks that reside before the sentinel. HANDLER locks must be released explicitly as part of HANDLER CLOSE statement, or an implicit close. The same approach with sentinel is also employed for LOCK TABLES locks. Since HANDLER and LOCK TABLES statement has never worked together, the implementation is made simple and only maintains one sentinel, which is used either for HANDLER locks, or for LOCK TABLES locks.
2009-12-22 19:09:15 +03:00
--echo # --> connection default
connection default;
Implement new type-of-operation-aware metadata locks. Add a wait-for graph based deadlock detector to the MDL subsystem. Fixes bug #46272 "MySQL 5.4.4, new MDL: unnecessary deadlock" and bug #37346 "innodb does not detect deadlock between update and alter table". The first bug manifested itself as an unwarranted abort of a transaction with ER_LOCK_DEADLOCK error by a concurrent ALTER statement, when this transaction tried to repeat use of a table, which it has already used in a similar fashion before ALTER started. The second bug showed up as a deadlock between table-level locks and InnoDB row locks, which was "detected" only after innodb_lock_wait_timeout timeout. A transaction would start using the table and modify a few rows. Then ALTER TABLE would come in, and start copying rows into a temporary table. Eventually it would stumble on the modified records and get blocked on a row lock. The first transaction would try to do more updates, and get blocked on thr_lock.c lock. This situation of circular wait would only get resolved by a timeout. Both these bugs stemmed from inadequate solutions to the problem of deadlocks occurring between different locking subsystems. In the first case we tried to avoid deadlocks between metadata locking and table-level locking subsystems, when upgrading shared metadata lock to exclusive one. Transactions holding the shared lock on the table and waiting for some table-level lock used to be aborted too aggressively. We also allowed ALTER TABLE to start in presence of transactions that modify the subject table. ALTER TABLE acquires TL_WRITE_ALLOW_READ lock at start, and that block all writes against the table (naturally, we don't want any writes to be lost when switching the old and the new table). TL_WRITE_ALLOW_READ lock, in turn, would block the started transaction on thr_lock.c lock, should they do more updates. This, again, lead to the need to abort such transactions. The second bug occurred simply because we didn't have any mechanism to detect deadlocks between the table-level locks in thr_lock.c and row-level locks in InnoDB, other than innodb_lock_wait_timeout. This patch solves both these problems by moving lock conflicts which are causing these deadlocks into the metadata locking subsystem, thus making it possible to avoid or detect such deadlocks inside MDL. To do this we introduce new type-of-operation-aware metadata locks, which allow MDL subsystem to know not only the fact that transaction has used or is going to use some object but also what kind of operation it has carried out or going to carry out on the object. This, along with the addition of a special kind of upgradable metadata lock, allows ALTER TABLE to wait until all transactions which has updated the table to go away. This solves the second issue. Another special type of upgradable metadata lock is acquired by LOCK TABLE WRITE. This second lock type allows to solve the first issue, since abortion of table-level locks in event of DDL under LOCK TABLES becomes also unnecessary. Below follows the list of incompatible changes introduced by this patch: - From now on, ALTER TABLE and CREATE/DROP TRIGGER SQL (i.e. those statements that acquire TL_WRITE_ALLOW_READ lock) wait for all transactions which has *updated* the table to complete. - From now on, LOCK TABLES ... WRITE, REPAIR/OPTIMIZE TABLE (i.e. all statements which acquire TL_WRITE table-level lock) wait for all transaction which *updated or read* from the table to complete. As a consequence, innodb_table_locks=0 option no longer applies to LOCK TABLES ... WRITE. - DROP DATABASE, DROP TABLE, RENAME TABLE no longer abort statements or transactions which use tables being dropped or renamed, and instead wait for these transactions to complete. - Since LOCK TABLES WRITE now takes a special metadata lock, not compatible with with reads or writes against the subject table and transaction-wide, thr_lock.c deadlock avoidance algorithm that used to ensure absence of deadlocks between LOCK TABLES WRITE and other statements is no longer sufficient, even for MyISAM. The wait-for graph based deadlock detector of MDL subsystem may sometimes be necessary and is involved. This may lead to ER_LOCK_DEADLOCK error produced for multi-statement transactions even if these only use MyISAM: session 1: session 2: begin; update t1 ... lock table t2 write, t1 write; -- gets a lock on t2, blocks on t1 update t2 ... (ER_LOCK_DEADLOCK) - Finally, support of LOW_PRIORITY option for LOCK TABLES ... WRITE was abandoned. LOCK TABLE ... LOW_PRIORITY WRITE from now on has the same priority as the usual LOCK TABLE ... WRITE. SELECT HIGH PRIORITY no longer trumps LOCK TABLE ... WRITE in the wait queue. - We do not take upgradable metadata locks on implicitly locked tables. So if one has, say, a view v1 that uses table t1, and issues: LOCK TABLE v1 WRITE; FLUSH TABLE t1; -- (or just 'FLUSH TABLES'), an error is produced. In order to be able to perform DDL on a table under LOCK TABLES, the table must be locked explicitly in the LOCK TABLES list.
2010-02-01 14:43:06 +03:00
send drop table t2;
--echo # --> connection con2
connection con2;
--echo # Waiting for 'drop table t2' to get blocked...
let $wait_condition=select count(*)=1 from information_schema.processlist where state='Waiting for table' and info='drop table t2';
--source include/wait_condition.inc
--echo # --> connection con1
connection con1;
A prerequisite patch for the fix for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". Introduce a notion of a sentinel to MDL_context. A sentinel is a ticket that separates all tickets in the context into two groups: before and after it. Currently we can have (and need) only one designated sentinel -- it separates all locks taken by LOCK TABLE or HANDLER statement, which must survive COMMIT and ROLLBACK and all other locks, which must be released at COMMIT or ROLLBACK. The tricky part is maintaining the sentinel up to date when someone release its corresponding ticket. This can happen, e.g. if someone issues DROP TABLE under LOCK TABLES (generally, see all calls to release_all_locks_for_name()). MDL_context::release_ticket() is modified to take care of it. ****** A fix and a test case for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". An attempt to mix HANDLER SQL statements, which are transaction- agnostic, an open multi-statement transaction, and DDL against the involved tables (in a concurrent connection) could lead to a deadlock. The deadlock would occur when HANDLER OPEN or HANDLER READ would have to wait on a conflicting metadata lock. If the connection that issued HANDLER statement also had other metadata locks (say, acquired in scope of a transaction), a classical deadlock situation of mutual wait could occur. Incompatible change: entering LOCK TABLES mode automatically closes all open HANDLERs in the current connection. Incompatible change: previously an attempt to wait on a lock in a connection that has an open HANDLER statement could wait indefinitely/deadlock. After this patch, an error ER_LOCK_DEADLOCK is produced. The idea of the fix is to merge thd->handler_mdl_context with the main mdl_context of the connection, used for transactional locks. This makes deadlock detection possible, since all waits with locks are "visible" and available to analysis in a single MDL context of the connection. Since HANDLER locks and transactional locks have a different life cycle -- HANDLERs are explicitly open and closed, and so are HANDLER locks, explicitly acquired and released, whereas transactional locks "accumulate" till the end of a transaction and are released only with COMMIT, ROLLBACK and ROLLBACK TO SAVEPOINT, a concept of "sentinel" was introduced to MDL_context. All locks, HANDLER and others, reside in the same linked list. However, a selected element of the list separates locks with different life cycle. HANDLER locks always reside at the end of the list, after the sentinel. Transactional locks are prepended to the beginning of the list, before the sentinel. Thus, ROLLBACK, COMMIT or ROLLBACK TO SAVEPOINT, only release those locks that reside before the sentinel. HANDLER locks must be released explicitly as part of HANDLER CLOSE statement, or an implicit close. The same approach with sentinel is also employed for LOCK TABLES locks. Since HANDLER and LOCK TABLES statement has never worked together, the implementation is made simple and only maintains one sentinel, which is used either for HANDLER locks, or for LOCK TABLES locks.
2009-12-22 19:09:15 +03:00
--error ER_LOCK_DEADLOCK
Implement new type-of-operation-aware metadata locks. Add a wait-for graph based deadlock detector to the MDL subsystem. Fixes bug #46272 "MySQL 5.4.4, new MDL: unnecessary deadlock" and bug #37346 "innodb does not detect deadlock between update and alter table". The first bug manifested itself as an unwarranted abort of a transaction with ER_LOCK_DEADLOCK error by a concurrent ALTER statement, when this transaction tried to repeat use of a table, which it has already used in a similar fashion before ALTER started. The second bug showed up as a deadlock between table-level locks and InnoDB row locks, which was "detected" only after innodb_lock_wait_timeout timeout. A transaction would start using the table and modify a few rows. Then ALTER TABLE would come in, and start copying rows into a temporary table. Eventually it would stumble on the modified records and get blocked on a row lock. The first transaction would try to do more updates, and get blocked on thr_lock.c lock. This situation of circular wait would only get resolved by a timeout. Both these bugs stemmed from inadequate solutions to the problem of deadlocks occurring between different locking subsystems. In the first case we tried to avoid deadlocks between metadata locking and table-level locking subsystems, when upgrading shared metadata lock to exclusive one. Transactions holding the shared lock on the table and waiting for some table-level lock used to be aborted too aggressively. We also allowed ALTER TABLE to start in presence of transactions that modify the subject table. ALTER TABLE acquires TL_WRITE_ALLOW_READ lock at start, and that block all writes against the table (naturally, we don't want any writes to be lost when switching the old and the new table). TL_WRITE_ALLOW_READ lock, in turn, would block the started transaction on thr_lock.c lock, should they do more updates. This, again, lead to the need to abort such transactions. The second bug occurred simply because we didn't have any mechanism to detect deadlocks between the table-level locks in thr_lock.c and row-level locks in InnoDB, other than innodb_lock_wait_timeout. This patch solves both these problems by moving lock conflicts which are causing these deadlocks into the metadata locking subsystem, thus making it possible to avoid or detect such deadlocks inside MDL. To do this we introduce new type-of-operation-aware metadata locks, which allow MDL subsystem to know not only the fact that transaction has used or is going to use some object but also what kind of operation it has carried out or going to carry out on the object. This, along with the addition of a special kind of upgradable metadata lock, allows ALTER TABLE to wait until all transactions which has updated the table to go away. This solves the second issue. Another special type of upgradable metadata lock is acquired by LOCK TABLE WRITE. This second lock type allows to solve the first issue, since abortion of table-level locks in event of DDL under LOCK TABLES becomes also unnecessary. Below follows the list of incompatible changes introduced by this patch: - From now on, ALTER TABLE and CREATE/DROP TRIGGER SQL (i.e. those statements that acquire TL_WRITE_ALLOW_READ lock) wait for all transactions which has *updated* the table to complete. - From now on, LOCK TABLES ... WRITE, REPAIR/OPTIMIZE TABLE (i.e. all statements which acquire TL_WRITE table-level lock) wait for all transaction which *updated or read* from the table to complete. As a consequence, innodb_table_locks=0 option no longer applies to LOCK TABLES ... WRITE. - DROP DATABASE, DROP TABLE, RENAME TABLE no longer abort statements or transactions which use tables being dropped or renamed, and instead wait for these transactions to complete. - Since LOCK TABLES WRITE now takes a special metadata lock, not compatible with with reads or writes against the subject table and transaction-wide, thr_lock.c deadlock avoidance algorithm that used to ensure absence of deadlocks between LOCK TABLES WRITE and other statements is no longer sufficient, even for MyISAM. The wait-for graph based deadlock detector of MDL subsystem may sometimes be necessary and is involved. This may lead to ER_LOCK_DEADLOCK error produced for multi-statement transactions even if these only use MyISAM: session 1: session 2: begin; update t1 ... lock table t2 write, t1 write; -- gets a lock on t2, blocks on t1 update t2 ... (ER_LOCK_DEADLOCK) - Finally, support of LOW_PRIORITY option for LOCK TABLES ... WRITE was abandoned. LOCK TABLE ... LOW_PRIORITY WRITE from now on has the same priority as the usual LOCK TABLE ... WRITE. SELECT HIGH PRIORITY no longer trumps LOCK TABLE ... WRITE in the wait queue. - We do not take upgradable metadata locks on implicitly locked tables. So if one has, say, a view v1 that uses table t1, and issues: LOCK TABLE v1 WRITE; FLUSH TABLE t1; -- (or just 'FLUSH TABLES'), an error is produced. In order to be able to perform DDL on a table under LOCK TABLES, the table must be locked explicitly in the LOCK TABLES list.
2010-02-01 14:43:06 +03:00
drop table t1;
unlock tables;
--echo # --> connection default
connection default;
reap;
A prerequisite patch for the fix for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". Introduce a notion of a sentinel to MDL_context. A sentinel is a ticket that separates all tickets in the context into two groups: before and after it. Currently we can have (and need) only one designated sentinel -- it separates all locks taken by LOCK TABLE or HANDLER statement, which must survive COMMIT and ROLLBACK and all other locks, which must be released at COMMIT or ROLLBACK. The tricky part is maintaining the sentinel up to date when someone release its corresponding ticket. This can happen, e.g. if someone issues DROP TABLE under LOCK TABLES (generally, see all calls to release_all_locks_for_name()). MDL_context::release_ticket() is modified to take care of it. ****** A fix and a test case for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". An attempt to mix HANDLER SQL statements, which are transaction- agnostic, an open multi-statement transaction, and DDL against the involved tables (in a concurrent connection) could lead to a deadlock. The deadlock would occur when HANDLER OPEN or HANDLER READ would have to wait on a conflicting metadata lock. If the connection that issued HANDLER statement also had other metadata locks (say, acquired in scope of a transaction), a classical deadlock situation of mutual wait could occur. Incompatible change: entering LOCK TABLES mode automatically closes all open HANDLERs in the current connection. Incompatible change: previously an attempt to wait on a lock in a connection that has an open HANDLER statement could wait indefinitely/deadlock. After this patch, an error ER_LOCK_DEADLOCK is produced. The idea of the fix is to merge thd->handler_mdl_context with the main mdl_context of the connection, used for transactional locks. This makes deadlock detection possible, since all waits with locks are "visible" and available to analysis in a single MDL context of the connection. Since HANDLER locks and transactional locks have a different life cycle -- HANDLERs are explicitly open and closed, and so are HANDLER locks, explicitly acquired and released, whereas transactional locks "accumulate" till the end of a transaction and are released only with COMMIT, ROLLBACK and ROLLBACK TO SAVEPOINT, a concept of "sentinel" was introduced to MDL_context. All locks, HANDLER and others, reside in the same linked list. However, a selected element of the list separates locks with different life cycle. HANDLER locks always reside at the end of the list, after the sentinel. Transactional locks are prepended to the beginning of the list, before the sentinel. Thus, ROLLBACK, COMMIT or ROLLBACK TO SAVEPOINT, only release those locks that reside before the sentinel. HANDLER locks must be released explicitly as part of HANDLER CLOSE statement, or an implicit close. The same approach with sentinel is also employed for LOCK TABLES locks. Since HANDLER and LOCK TABLES statement has never worked together, the implementation is made simple and only maintains one sentinel, which is used either for HANDLER locks, or for LOCK TABLES locks.
2009-12-22 19:09:15 +03:00
--echo # Demonstrate that there is no deadlock with FLUSH TABLE,
--echo # even though it is waiting for the other table to go away
Implement new type-of-operation-aware metadata locks. Add a wait-for graph based deadlock detector to the MDL subsystem. Fixes bug #46272 "MySQL 5.4.4, new MDL: unnecessary deadlock" and bug #37346 "innodb does not detect deadlock between update and alter table". The first bug manifested itself as an unwarranted abort of a transaction with ER_LOCK_DEADLOCK error by a concurrent ALTER statement, when this transaction tried to repeat use of a table, which it has already used in a similar fashion before ALTER started. The second bug showed up as a deadlock between table-level locks and InnoDB row locks, which was "detected" only after innodb_lock_wait_timeout timeout. A transaction would start using the table and modify a few rows. Then ALTER TABLE would come in, and start copying rows into a temporary table. Eventually it would stumble on the modified records and get blocked on a row lock. The first transaction would try to do more updates, and get blocked on thr_lock.c lock. This situation of circular wait would only get resolved by a timeout. Both these bugs stemmed from inadequate solutions to the problem of deadlocks occurring between different locking subsystems. In the first case we tried to avoid deadlocks between metadata locking and table-level locking subsystems, when upgrading shared metadata lock to exclusive one. Transactions holding the shared lock on the table and waiting for some table-level lock used to be aborted too aggressively. We also allowed ALTER TABLE to start in presence of transactions that modify the subject table. ALTER TABLE acquires TL_WRITE_ALLOW_READ lock at start, and that block all writes against the table (naturally, we don't want any writes to be lost when switching the old and the new table). TL_WRITE_ALLOW_READ lock, in turn, would block the started transaction on thr_lock.c lock, should they do more updates. This, again, lead to the need to abort such transactions. The second bug occurred simply because we didn't have any mechanism to detect deadlocks between the table-level locks in thr_lock.c and row-level locks in InnoDB, other than innodb_lock_wait_timeout. This patch solves both these problems by moving lock conflicts which are causing these deadlocks into the metadata locking subsystem, thus making it possible to avoid or detect such deadlocks inside MDL. To do this we introduce new type-of-operation-aware metadata locks, which allow MDL subsystem to know not only the fact that transaction has used or is going to use some object but also what kind of operation it has carried out or going to carry out on the object. This, along with the addition of a special kind of upgradable metadata lock, allows ALTER TABLE to wait until all transactions which has updated the table to go away. This solves the second issue. Another special type of upgradable metadata lock is acquired by LOCK TABLE WRITE. This second lock type allows to solve the first issue, since abortion of table-level locks in event of DDL under LOCK TABLES becomes also unnecessary. Below follows the list of incompatible changes introduced by this patch: - From now on, ALTER TABLE and CREATE/DROP TRIGGER SQL (i.e. those statements that acquire TL_WRITE_ALLOW_READ lock) wait for all transactions which has *updated* the table to complete. - From now on, LOCK TABLES ... WRITE, REPAIR/OPTIMIZE TABLE (i.e. all statements which acquire TL_WRITE table-level lock) wait for all transaction which *updated or read* from the table to complete. As a consequence, innodb_table_locks=0 option no longer applies to LOCK TABLES ... WRITE. - DROP DATABASE, DROP TABLE, RENAME TABLE no longer abort statements or transactions which use tables being dropped or renamed, and instead wait for these transactions to complete. - Since LOCK TABLES WRITE now takes a special metadata lock, not compatible with with reads or writes against the subject table and transaction-wide, thr_lock.c deadlock avoidance algorithm that used to ensure absence of deadlocks between LOCK TABLES WRITE and other statements is no longer sufficient, even for MyISAM. The wait-for graph based deadlock detector of MDL subsystem may sometimes be necessary and is involved. This may lead to ER_LOCK_DEADLOCK error produced for multi-statement transactions even if these only use MyISAM: session 1: session 2: begin; update t1 ... lock table t2 write, t1 write; -- gets a lock on t2, blocks on t1 update t2 ... (ER_LOCK_DEADLOCK) - Finally, support of LOW_PRIORITY option for LOCK TABLES ... WRITE was abandoned. LOCK TABLE ... LOW_PRIORITY WRITE from now on has the same priority as the usual LOCK TABLE ... WRITE. SELECT HIGH PRIORITY no longer trumps LOCK TABLE ... WRITE in the wait queue. - We do not take upgradable metadata locks on implicitly locked tables. So if one has, say, a view v1 that uses table t1, and issues: LOCK TABLE v1 WRITE; FLUSH TABLE t1; -- (or just 'FLUSH TABLES'), an error is produced. In order to be able to perform DDL on a table under LOCK TABLES, the table must be locked explicitly in the LOCK TABLES list.
2010-02-01 14:43:06 +03:00
create table t2 like t1;
A prerequisite patch for the fix for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". Introduce a notion of a sentinel to MDL_context. A sentinel is a ticket that separates all tickets in the context into two groups: before and after it. Currently we can have (and need) only one designated sentinel -- it separates all locks taken by LOCK TABLE or HANDLER statement, which must survive COMMIT and ROLLBACK and all other locks, which must be released at COMMIT or ROLLBACK. The tricky part is maintaining the sentinel up to date when someone release its corresponding ticket. This can happen, e.g. if someone issues DROP TABLE under LOCK TABLES (generally, see all calls to release_all_locks_for_name()). MDL_context::release_ticket() is modified to take care of it. ****** A fix and a test case for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". An attempt to mix HANDLER SQL statements, which are transaction- agnostic, an open multi-statement transaction, and DDL against the involved tables (in a concurrent connection) could lead to a deadlock. The deadlock would occur when HANDLER OPEN or HANDLER READ would have to wait on a conflicting metadata lock. If the connection that issued HANDLER statement also had other metadata locks (say, acquired in scope of a transaction), a classical deadlock situation of mutual wait could occur. Incompatible change: entering LOCK TABLES mode automatically closes all open HANDLERs in the current connection. Incompatible change: previously an attempt to wait on a lock in a connection that has an open HANDLER statement could wait indefinitely/deadlock. After this patch, an error ER_LOCK_DEADLOCK is produced. The idea of the fix is to merge thd->handler_mdl_context with the main mdl_context of the connection, used for transactional locks. This makes deadlock detection possible, since all waits with locks are "visible" and available to analysis in a single MDL context of the connection. Since HANDLER locks and transactional locks have a different life cycle -- HANDLERs are explicitly open and closed, and so are HANDLER locks, explicitly acquired and released, whereas transactional locks "accumulate" till the end of a transaction and are released only with COMMIT, ROLLBACK and ROLLBACK TO SAVEPOINT, a concept of "sentinel" was introduced to MDL_context. All locks, HANDLER and others, reside in the same linked list. However, a selected element of the list separates locks with different life cycle. HANDLER locks always reside at the end of the list, after the sentinel. Transactional locks are prepended to the beginning of the list, before the sentinel. Thus, ROLLBACK, COMMIT or ROLLBACK TO SAVEPOINT, only release those locks that reside before the sentinel. HANDLER locks must be released explicitly as part of HANDLER CLOSE statement, or an implicit close. The same approach with sentinel is also employed for LOCK TABLES locks. Since HANDLER and LOCK TABLES statement has never worked together, the implementation is made simple and only maintains one sentinel, which is used either for HANDLER locks, or for LOCK TABLES locks.
2009-12-22 19:09:15 +03:00
--echo # Sending:
--send flush table t2
--echo # --> connection con2
connection con2;
drop table t1;
--echo # --> connection con1
connection con1;
unlock tables;
--echo # --> connection default
connection default;
--echo # Reaping 'flush table t2'...
--reap
drop table t2;
--echo #
--echo # Bug #46224 HANDLER statements within a transaction might
--echo # lead to deadlocks
--echo #
create table t1 (a int, key a(a));
Implement new type-of-operation-aware metadata locks. Add a wait-for graph based deadlock detector to the MDL subsystem. Fixes bug #46272 "MySQL 5.4.4, new MDL: unnecessary deadlock" and bug #37346 "innodb does not detect deadlock between update and alter table". The first bug manifested itself as an unwarranted abort of a transaction with ER_LOCK_DEADLOCK error by a concurrent ALTER statement, when this transaction tried to repeat use of a table, which it has already used in a similar fashion before ALTER started. The second bug showed up as a deadlock between table-level locks and InnoDB row locks, which was "detected" only after innodb_lock_wait_timeout timeout. A transaction would start using the table and modify a few rows. Then ALTER TABLE would come in, and start copying rows into a temporary table. Eventually it would stumble on the modified records and get blocked on a row lock. The first transaction would try to do more updates, and get blocked on thr_lock.c lock. This situation of circular wait would only get resolved by a timeout. Both these bugs stemmed from inadequate solutions to the problem of deadlocks occurring between different locking subsystems. In the first case we tried to avoid deadlocks between metadata locking and table-level locking subsystems, when upgrading shared metadata lock to exclusive one. Transactions holding the shared lock on the table and waiting for some table-level lock used to be aborted too aggressively. We also allowed ALTER TABLE to start in presence of transactions that modify the subject table. ALTER TABLE acquires TL_WRITE_ALLOW_READ lock at start, and that block all writes against the table (naturally, we don't want any writes to be lost when switching the old and the new table). TL_WRITE_ALLOW_READ lock, in turn, would block the started transaction on thr_lock.c lock, should they do more updates. This, again, lead to the need to abort such transactions. The second bug occurred simply because we didn't have any mechanism to detect deadlocks between the table-level locks in thr_lock.c and row-level locks in InnoDB, other than innodb_lock_wait_timeout. This patch solves both these problems by moving lock conflicts which are causing these deadlocks into the metadata locking subsystem, thus making it possible to avoid or detect such deadlocks inside MDL. To do this we introduce new type-of-operation-aware metadata locks, which allow MDL subsystem to know not only the fact that transaction has used or is going to use some object but also what kind of operation it has carried out or going to carry out on the object. This, along with the addition of a special kind of upgradable metadata lock, allows ALTER TABLE to wait until all transactions which has updated the table to go away. This solves the second issue. Another special type of upgradable metadata lock is acquired by LOCK TABLE WRITE. This second lock type allows to solve the first issue, since abortion of table-level locks in event of DDL under LOCK TABLES becomes also unnecessary. Below follows the list of incompatible changes introduced by this patch: - From now on, ALTER TABLE and CREATE/DROP TRIGGER SQL (i.e. those statements that acquire TL_WRITE_ALLOW_READ lock) wait for all transactions which has *updated* the table to complete. - From now on, LOCK TABLES ... WRITE, REPAIR/OPTIMIZE TABLE (i.e. all statements which acquire TL_WRITE table-level lock) wait for all transaction which *updated or read* from the table to complete. As a consequence, innodb_table_locks=0 option no longer applies to LOCK TABLES ... WRITE. - DROP DATABASE, DROP TABLE, RENAME TABLE no longer abort statements or transactions which use tables being dropped or renamed, and instead wait for these transactions to complete. - Since LOCK TABLES WRITE now takes a special metadata lock, not compatible with with reads or writes against the subject table and transaction-wide, thr_lock.c deadlock avoidance algorithm that used to ensure absence of deadlocks between LOCK TABLES WRITE and other statements is no longer sufficient, even for MyISAM. The wait-for graph based deadlock detector of MDL subsystem may sometimes be necessary and is involved. This may lead to ER_LOCK_DEADLOCK error produced for multi-statement transactions even if these only use MyISAM: session 1: session 2: begin; update t1 ... lock table t2 write, t1 write; -- gets a lock on t2, blocks on t1 update t2 ... (ER_LOCK_DEADLOCK) - Finally, support of LOW_PRIORITY option for LOCK TABLES ... WRITE was abandoned. LOCK TABLE ... LOW_PRIORITY WRITE from now on has the same priority as the usual LOCK TABLE ... WRITE. SELECT HIGH PRIORITY no longer trumps LOCK TABLE ... WRITE in the wait queue. - We do not take upgradable metadata locks on implicitly locked tables. So if one has, say, a view v1 that uses table t1, and issues: LOCK TABLE v1 WRITE; FLUSH TABLE t1; -- (or just 'FLUSH TABLES'), an error is produced. In order to be able to perform DDL on a table under LOCK TABLES, the table must be locked explicitly in the LOCK TABLES list.
2010-02-01 14:43:06 +03:00
insert into t1 values (1), (2);
A prerequisite patch for the fix for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". Introduce a notion of a sentinel to MDL_context. A sentinel is a ticket that separates all tickets in the context into two groups: before and after it. Currently we can have (and need) only one designated sentinel -- it separates all locks taken by LOCK TABLE or HANDLER statement, which must survive COMMIT and ROLLBACK and all other locks, which must be released at COMMIT or ROLLBACK. The tricky part is maintaining the sentinel up to date when someone release its corresponding ticket. This can happen, e.g. if someone issues DROP TABLE under LOCK TABLES (generally, see all calls to release_all_locks_for_name()). MDL_context::release_ticket() is modified to take care of it. ****** A fix and a test case for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". An attempt to mix HANDLER SQL statements, which are transaction- agnostic, an open multi-statement transaction, and DDL against the involved tables (in a concurrent connection) could lead to a deadlock. The deadlock would occur when HANDLER OPEN or HANDLER READ would have to wait on a conflicting metadata lock. If the connection that issued HANDLER statement also had other metadata locks (say, acquired in scope of a transaction), a classical deadlock situation of mutual wait could occur. Incompatible change: entering LOCK TABLES mode automatically closes all open HANDLERs in the current connection. Incompatible change: previously an attempt to wait on a lock in a connection that has an open HANDLER statement could wait indefinitely/deadlock. After this patch, an error ER_LOCK_DEADLOCK is produced. The idea of the fix is to merge thd->handler_mdl_context with the main mdl_context of the connection, used for transactional locks. This makes deadlock detection possible, since all waits with locks are "visible" and available to analysis in a single MDL context of the connection. Since HANDLER locks and transactional locks have a different life cycle -- HANDLERs are explicitly open and closed, and so are HANDLER locks, explicitly acquired and released, whereas transactional locks "accumulate" till the end of a transaction and are released only with COMMIT, ROLLBACK and ROLLBACK TO SAVEPOINT, a concept of "sentinel" was introduced to MDL_context. All locks, HANDLER and others, reside in the same linked list. However, a selected element of the list separates locks with different life cycle. HANDLER locks always reside at the end of the list, after the sentinel. Transactional locks are prepended to the beginning of the list, before the sentinel. Thus, ROLLBACK, COMMIT or ROLLBACK TO SAVEPOINT, only release those locks that reside before the sentinel. HANDLER locks must be released explicitly as part of HANDLER CLOSE statement, or an implicit close. The same approach with sentinel is also employed for LOCK TABLES locks. Since HANDLER and LOCK TABLES statement has never worked together, the implementation is made simple and only maintains one sentinel, which is used either for HANDLER locks, or for LOCK TABLES locks.
2009-12-22 19:09:15 +03:00
--echo # --> connection default
connection default;
begin;
select * from t1;
handler t1 open;
--echo # --> connection con1
connection con1;
Implement new type-of-operation-aware metadata locks. Add a wait-for graph based deadlock detector to the MDL subsystem. Fixes bug #46272 "MySQL 5.4.4, new MDL: unnecessary deadlock" and bug #37346 "innodb does not detect deadlock between update and alter table". The first bug manifested itself as an unwarranted abort of a transaction with ER_LOCK_DEADLOCK error by a concurrent ALTER statement, when this transaction tried to repeat use of a table, which it has already used in a similar fashion before ALTER started. The second bug showed up as a deadlock between table-level locks and InnoDB row locks, which was "detected" only after innodb_lock_wait_timeout timeout. A transaction would start using the table and modify a few rows. Then ALTER TABLE would come in, and start copying rows into a temporary table. Eventually it would stumble on the modified records and get blocked on a row lock. The first transaction would try to do more updates, and get blocked on thr_lock.c lock. This situation of circular wait would only get resolved by a timeout. Both these bugs stemmed from inadequate solutions to the problem of deadlocks occurring between different locking subsystems. In the first case we tried to avoid deadlocks between metadata locking and table-level locking subsystems, when upgrading shared metadata lock to exclusive one. Transactions holding the shared lock on the table and waiting for some table-level lock used to be aborted too aggressively. We also allowed ALTER TABLE to start in presence of transactions that modify the subject table. ALTER TABLE acquires TL_WRITE_ALLOW_READ lock at start, and that block all writes against the table (naturally, we don't want any writes to be lost when switching the old and the new table). TL_WRITE_ALLOW_READ lock, in turn, would block the started transaction on thr_lock.c lock, should they do more updates. This, again, lead to the need to abort such transactions. The second bug occurred simply because we didn't have any mechanism to detect deadlocks between the table-level locks in thr_lock.c and row-level locks in InnoDB, other than innodb_lock_wait_timeout. This patch solves both these problems by moving lock conflicts which are causing these deadlocks into the metadata locking subsystem, thus making it possible to avoid or detect such deadlocks inside MDL. To do this we introduce new type-of-operation-aware metadata locks, which allow MDL subsystem to know not only the fact that transaction has used or is going to use some object but also what kind of operation it has carried out or going to carry out on the object. This, along with the addition of a special kind of upgradable metadata lock, allows ALTER TABLE to wait until all transactions which has updated the table to go away. This solves the second issue. Another special type of upgradable metadata lock is acquired by LOCK TABLE WRITE. This second lock type allows to solve the first issue, since abortion of table-level locks in event of DDL under LOCK TABLES becomes also unnecessary. Below follows the list of incompatible changes introduced by this patch: - From now on, ALTER TABLE and CREATE/DROP TRIGGER SQL (i.e. those statements that acquire TL_WRITE_ALLOW_READ lock) wait for all transactions which has *updated* the table to complete. - From now on, LOCK TABLES ... WRITE, REPAIR/OPTIMIZE TABLE (i.e. all statements which acquire TL_WRITE table-level lock) wait for all transaction which *updated or read* from the table to complete. As a consequence, innodb_table_locks=0 option no longer applies to LOCK TABLES ... WRITE. - DROP DATABASE, DROP TABLE, RENAME TABLE no longer abort statements or transactions which use tables being dropped or renamed, and instead wait for these transactions to complete. - Since LOCK TABLES WRITE now takes a special metadata lock, not compatible with with reads or writes against the subject table and transaction-wide, thr_lock.c deadlock avoidance algorithm that used to ensure absence of deadlocks between LOCK TABLES WRITE and other statements is no longer sufficient, even for MyISAM. The wait-for graph based deadlock detector of MDL subsystem may sometimes be necessary and is involved. This may lead to ER_LOCK_DEADLOCK error produced for multi-statement transactions even if these only use MyISAM: session 1: session 2: begin; update t1 ... lock table t2 write, t1 write; -- gets a lock on t2, blocks on t1 update t2 ... (ER_LOCK_DEADLOCK) - Finally, support of LOW_PRIORITY option for LOCK TABLES ... WRITE was abandoned. LOCK TABLE ... LOW_PRIORITY WRITE from now on has the same priority as the usual LOCK TABLE ... WRITE. SELECT HIGH PRIORITY no longer trumps LOCK TABLE ... WRITE in the wait queue. - We do not take upgradable metadata locks on implicitly locked tables. So if one has, say, a view v1 that uses table t1, and issues: LOCK TABLE v1 WRITE; FLUSH TABLE t1; -- (or just 'FLUSH TABLES'), an error is produced. In order to be able to perform DDL on a table under LOCK TABLES, the table must be locked explicitly in the LOCK TABLES list.
2010-02-01 14:43:06 +03:00
--echo # Sending:
--send lock tables t1 write
--echo # --> connection con2
connection con2;
--echo # Check that 'lock tables t1 write' waits until transaction which
--echo # has read from the table commits.
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Waiting for table" and info = "lock tables t1 write";
--source include/wait_condition.inc
--echo # --> connection default
connection default;
--echo # The below 'handler t1 read ...' should not be blocked as
--echo # 'lock tables t1 write' has not succeeded yet.
handler t1 read a next;
--echo # Unblock 'lock tables t1 write'.
commit;
--echo # --> connection con1
connection con1;
--echo # Reap 'lock tables t1 write'.
--reap
A prerequisite patch for the fix for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". Introduce a notion of a sentinel to MDL_context. A sentinel is a ticket that separates all tickets in the context into two groups: before and after it. Currently we can have (and need) only one designated sentinel -- it separates all locks taken by LOCK TABLE or HANDLER statement, which must survive COMMIT and ROLLBACK and all other locks, which must be released at COMMIT or ROLLBACK. The tricky part is maintaining the sentinel up to date when someone release its corresponding ticket. This can happen, e.g. if someone issues DROP TABLE under LOCK TABLES (generally, see all calls to release_all_locks_for_name()). MDL_context::release_ticket() is modified to take care of it. ****** A fix and a test case for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". An attempt to mix HANDLER SQL statements, which are transaction- agnostic, an open multi-statement transaction, and DDL against the involved tables (in a concurrent connection) could lead to a deadlock. The deadlock would occur when HANDLER OPEN or HANDLER READ would have to wait on a conflicting metadata lock. If the connection that issued HANDLER statement also had other metadata locks (say, acquired in scope of a transaction), a classical deadlock situation of mutual wait could occur. Incompatible change: entering LOCK TABLES mode automatically closes all open HANDLERs in the current connection. Incompatible change: previously an attempt to wait on a lock in a connection that has an open HANDLER statement could wait indefinitely/deadlock. After this patch, an error ER_LOCK_DEADLOCK is produced. The idea of the fix is to merge thd->handler_mdl_context with the main mdl_context of the connection, used for transactional locks. This makes deadlock detection possible, since all waits with locks are "visible" and available to analysis in a single MDL context of the connection. Since HANDLER locks and transactional locks have a different life cycle -- HANDLERs are explicitly open and closed, and so are HANDLER locks, explicitly acquired and released, whereas transactional locks "accumulate" till the end of a transaction and are released only with COMMIT, ROLLBACK and ROLLBACK TO SAVEPOINT, a concept of "sentinel" was introduced to MDL_context. All locks, HANDLER and others, reside in the same linked list. However, a selected element of the list separates locks with different life cycle. HANDLER locks always reside at the end of the list, after the sentinel. Transactional locks are prepended to the beginning of the list, before the sentinel. Thus, ROLLBACK, COMMIT or ROLLBACK TO SAVEPOINT, only release those locks that reside before the sentinel. HANDLER locks must be released explicitly as part of HANDLER CLOSE statement, or an implicit close. The same approach with sentinel is also employed for LOCK TABLES locks. Since HANDLER and LOCK TABLES statement has never worked together, the implementation is made simple and only maintains one sentinel, which is used either for HANDLER locks, or for LOCK TABLES locks.
2009-12-22 19:09:15 +03:00
--echo # --> connection default
connection default;
--echo # Sending:
--send handler t1 read a next
--echo # --> connection con1
connection con1;
--echo # Waiting for 'handler t1 read a next' to get blocked...
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Table lock" and info = "handler t1 read a next";
--source include/wait_condition.inc
Implement new type-of-operation-aware metadata locks. Add a wait-for graph based deadlock detector to the MDL subsystem. Fixes bug #46272 "MySQL 5.4.4, new MDL: unnecessary deadlock" and bug #37346 "innodb does not detect deadlock between update and alter table". The first bug manifested itself as an unwarranted abort of a transaction with ER_LOCK_DEADLOCK error by a concurrent ALTER statement, when this transaction tried to repeat use of a table, which it has already used in a similar fashion before ALTER started. The second bug showed up as a deadlock between table-level locks and InnoDB row locks, which was "detected" only after innodb_lock_wait_timeout timeout. A transaction would start using the table and modify a few rows. Then ALTER TABLE would come in, and start copying rows into a temporary table. Eventually it would stumble on the modified records and get blocked on a row lock. The first transaction would try to do more updates, and get blocked on thr_lock.c lock. This situation of circular wait would only get resolved by a timeout. Both these bugs stemmed from inadequate solutions to the problem of deadlocks occurring between different locking subsystems. In the first case we tried to avoid deadlocks between metadata locking and table-level locking subsystems, when upgrading shared metadata lock to exclusive one. Transactions holding the shared lock on the table and waiting for some table-level lock used to be aborted too aggressively. We also allowed ALTER TABLE to start in presence of transactions that modify the subject table. ALTER TABLE acquires TL_WRITE_ALLOW_READ lock at start, and that block all writes against the table (naturally, we don't want any writes to be lost when switching the old and the new table). TL_WRITE_ALLOW_READ lock, in turn, would block the started transaction on thr_lock.c lock, should they do more updates. This, again, lead to the need to abort such transactions. The second bug occurred simply because we didn't have any mechanism to detect deadlocks between the table-level locks in thr_lock.c and row-level locks in InnoDB, other than innodb_lock_wait_timeout. This patch solves both these problems by moving lock conflicts which are causing these deadlocks into the metadata locking subsystem, thus making it possible to avoid or detect such deadlocks inside MDL. To do this we introduce new type-of-operation-aware metadata locks, which allow MDL subsystem to know not only the fact that transaction has used or is going to use some object but also what kind of operation it has carried out or going to carry out on the object. This, along with the addition of a special kind of upgradable metadata lock, allows ALTER TABLE to wait until all transactions which has updated the table to go away. This solves the second issue. Another special type of upgradable metadata lock is acquired by LOCK TABLE WRITE. This second lock type allows to solve the first issue, since abortion of table-level locks in event of DDL under LOCK TABLES becomes also unnecessary. Below follows the list of incompatible changes introduced by this patch: - From now on, ALTER TABLE and CREATE/DROP TRIGGER SQL (i.e. those statements that acquire TL_WRITE_ALLOW_READ lock) wait for all transactions which has *updated* the table to complete. - From now on, LOCK TABLES ... WRITE, REPAIR/OPTIMIZE TABLE (i.e. all statements which acquire TL_WRITE table-level lock) wait for all transaction which *updated or read* from the table to complete. As a consequence, innodb_table_locks=0 option no longer applies to LOCK TABLES ... WRITE. - DROP DATABASE, DROP TABLE, RENAME TABLE no longer abort statements or transactions which use tables being dropped or renamed, and instead wait for these transactions to complete. - Since LOCK TABLES WRITE now takes a special metadata lock, not compatible with with reads or writes against the subject table and transaction-wide, thr_lock.c deadlock avoidance algorithm that used to ensure absence of deadlocks between LOCK TABLES WRITE and other statements is no longer sufficient, even for MyISAM. The wait-for graph based deadlock detector of MDL subsystem may sometimes be necessary and is involved. This may lead to ER_LOCK_DEADLOCK error produced for multi-statement transactions even if these only use MyISAM: session 1: session 2: begin; update t1 ... lock table t2 write, t1 write; -- gets a lock on t2, blocks on t1 update t2 ... (ER_LOCK_DEADLOCK) - Finally, support of LOW_PRIORITY option for LOCK TABLES ... WRITE was abandoned. LOCK TABLE ... LOW_PRIORITY WRITE from now on has the same priority as the usual LOCK TABLE ... WRITE. SELECT HIGH PRIORITY no longer trumps LOCK TABLE ... WRITE in the wait queue. - We do not take upgradable metadata locks on implicitly locked tables. So if one has, say, a view v1 that uses table t1, and issues: LOCK TABLE v1 WRITE; FLUSH TABLE t1; -- (or just 'FLUSH TABLES'), an error is produced. In order to be able to perform DDL on a table under LOCK TABLES, the table must be locked explicitly in the LOCK TABLES list.
2010-02-01 14:43:06 +03:00
--echo # The below 'drop table t1' should be able to proceed without
--echo # waiting as it will force HANDLER to be closed.
drop table t1;
unlock tables;
A prerequisite patch for the fix for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". Introduce a notion of a sentinel to MDL_context. A sentinel is a ticket that separates all tickets in the context into two groups: before and after it. Currently we can have (and need) only one designated sentinel -- it separates all locks taken by LOCK TABLE or HANDLER statement, which must survive COMMIT and ROLLBACK and all other locks, which must be released at COMMIT or ROLLBACK. The tricky part is maintaining the sentinel up to date when someone release its corresponding ticket. This can happen, e.g. if someone issues DROP TABLE under LOCK TABLES (generally, see all calls to release_all_locks_for_name()). MDL_context::release_ticket() is modified to take care of it. ****** A fix and a test case for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". An attempt to mix HANDLER SQL statements, which are transaction- agnostic, an open multi-statement transaction, and DDL against the involved tables (in a concurrent connection) could lead to a deadlock. The deadlock would occur when HANDLER OPEN or HANDLER READ would have to wait on a conflicting metadata lock. If the connection that issued HANDLER statement also had other metadata locks (say, acquired in scope of a transaction), a classical deadlock situation of mutual wait could occur. Incompatible change: entering LOCK TABLES mode automatically closes all open HANDLERs in the current connection. Incompatible change: previously an attempt to wait on a lock in a connection that has an open HANDLER statement could wait indefinitely/deadlock. After this patch, an error ER_LOCK_DEADLOCK is produced. The idea of the fix is to merge thd->handler_mdl_context with the main mdl_context of the connection, used for transactional locks. This makes deadlock detection possible, since all waits with locks are "visible" and available to analysis in a single MDL context of the connection. Since HANDLER locks and transactional locks have a different life cycle -- HANDLERs are explicitly open and closed, and so are HANDLER locks, explicitly acquired and released, whereas transactional locks "accumulate" till the end of a transaction and are released only with COMMIT, ROLLBACK and ROLLBACK TO SAVEPOINT, a concept of "sentinel" was introduced to MDL_context. All locks, HANDLER and others, reside in the same linked list. However, a selected element of the list separates locks with different life cycle. HANDLER locks always reside at the end of the list, after the sentinel. Transactional locks are prepended to the beginning of the list, before the sentinel. Thus, ROLLBACK, COMMIT or ROLLBACK TO SAVEPOINT, only release those locks that reside before the sentinel. HANDLER locks must be released explicitly as part of HANDLER CLOSE statement, or an implicit close. The same approach with sentinel is also employed for LOCK TABLES locks. Since HANDLER and LOCK TABLES statement has never worked together, the implementation is made simple and only maintains one sentinel, which is used either for HANDLER locks, or for LOCK TABLES locks.
2009-12-22 19:09:15 +03:00
--echo # --> connection default
connection default;
--echo # Reaping 'handler t1 read a next'...
Implement new type-of-operation-aware metadata locks. Add a wait-for graph based deadlock detector to the MDL subsystem. Fixes bug #46272 "MySQL 5.4.4, new MDL: unnecessary deadlock" and bug #37346 "innodb does not detect deadlock between update and alter table". The first bug manifested itself as an unwarranted abort of a transaction with ER_LOCK_DEADLOCK error by a concurrent ALTER statement, when this transaction tried to repeat use of a table, which it has already used in a similar fashion before ALTER started. The second bug showed up as a deadlock between table-level locks and InnoDB row locks, which was "detected" only after innodb_lock_wait_timeout timeout. A transaction would start using the table and modify a few rows. Then ALTER TABLE would come in, and start copying rows into a temporary table. Eventually it would stumble on the modified records and get blocked on a row lock. The first transaction would try to do more updates, and get blocked on thr_lock.c lock. This situation of circular wait would only get resolved by a timeout. Both these bugs stemmed from inadequate solutions to the problem of deadlocks occurring between different locking subsystems. In the first case we tried to avoid deadlocks between metadata locking and table-level locking subsystems, when upgrading shared metadata lock to exclusive one. Transactions holding the shared lock on the table and waiting for some table-level lock used to be aborted too aggressively. We also allowed ALTER TABLE to start in presence of transactions that modify the subject table. ALTER TABLE acquires TL_WRITE_ALLOW_READ lock at start, and that block all writes against the table (naturally, we don't want any writes to be lost when switching the old and the new table). TL_WRITE_ALLOW_READ lock, in turn, would block the started transaction on thr_lock.c lock, should they do more updates. This, again, lead to the need to abort such transactions. The second bug occurred simply because we didn't have any mechanism to detect deadlocks between the table-level locks in thr_lock.c and row-level locks in InnoDB, other than innodb_lock_wait_timeout. This patch solves both these problems by moving lock conflicts which are causing these deadlocks into the metadata locking subsystem, thus making it possible to avoid or detect such deadlocks inside MDL. To do this we introduce new type-of-operation-aware metadata locks, which allow MDL subsystem to know not only the fact that transaction has used or is going to use some object but also what kind of operation it has carried out or going to carry out on the object. This, along with the addition of a special kind of upgradable metadata lock, allows ALTER TABLE to wait until all transactions which has updated the table to go away. This solves the second issue. Another special type of upgradable metadata lock is acquired by LOCK TABLE WRITE. This second lock type allows to solve the first issue, since abortion of table-level locks in event of DDL under LOCK TABLES becomes also unnecessary. Below follows the list of incompatible changes introduced by this patch: - From now on, ALTER TABLE and CREATE/DROP TRIGGER SQL (i.e. those statements that acquire TL_WRITE_ALLOW_READ lock) wait for all transactions which has *updated* the table to complete. - From now on, LOCK TABLES ... WRITE, REPAIR/OPTIMIZE TABLE (i.e. all statements which acquire TL_WRITE table-level lock) wait for all transaction which *updated or read* from the table to complete. As a consequence, innodb_table_locks=0 option no longer applies to LOCK TABLES ... WRITE. - DROP DATABASE, DROP TABLE, RENAME TABLE no longer abort statements or transactions which use tables being dropped or renamed, and instead wait for these transactions to complete. - Since LOCK TABLES WRITE now takes a special metadata lock, not compatible with with reads or writes against the subject table and transaction-wide, thr_lock.c deadlock avoidance algorithm that used to ensure absence of deadlocks between LOCK TABLES WRITE and other statements is no longer sufficient, even for MyISAM. The wait-for graph based deadlock detector of MDL subsystem may sometimes be necessary and is involved. This may lead to ER_LOCK_DEADLOCK error produced for multi-statement transactions even if these only use MyISAM: session 1: session 2: begin; update t1 ... lock table t2 write, t1 write; -- gets a lock on t2, blocks on t1 update t2 ... (ER_LOCK_DEADLOCK) - Finally, support of LOW_PRIORITY option for LOCK TABLES ... WRITE was abandoned. LOCK TABLE ... LOW_PRIORITY WRITE from now on has the same priority as the usual LOCK TABLE ... WRITE. SELECT HIGH PRIORITY no longer trumps LOCK TABLE ... WRITE in the wait queue. - We do not take upgradable metadata locks on implicitly locked tables. So if one has, say, a view v1 that uses table t1, and issues: LOCK TABLE v1 WRITE; FLUSH TABLE t1; -- (or just 'FLUSH TABLES'), an error is produced. In order to be able to perform DDL on a table under LOCK TABLES, the table must be locked explicitly in the LOCK TABLES list.
2010-02-01 14:43:06 +03:00
--error ER_NO_SUCH_TABLE
A prerequisite patch for the fix for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". Introduce a notion of a sentinel to MDL_context. A sentinel is a ticket that separates all tickets in the context into two groups: before and after it. Currently we can have (and need) only one designated sentinel -- it separates all locks taken by LOCK TABLE or HANDLER statement, which must survive COMMIT and ROLLBACK and all other locks, which must be released at COMMIT or ROLLBACK. The tricky part is maintaining the sentinel up to date when someone release its corresponding ticket. This can happen, e.g. if someone issues DROP TABLE under LOCK TABLES (generally, see all calls to release_all_locks_for_name()). MDL_context::release_ticket() is modified to take care of it. ****** A fix and a test case for Bug#46224 "HANDLER statements within a transaction might lead to deadlocks". An attempt to mix HANDLER SQL statements, which are transaction- agnostic, an open multi-statement transaction, and DDL against the involved tables (in a concurrent connection) could lead to a deadlock. The deadlock would occur when HANDLER OPEN or HANDLER READ would have to wait on a conflicting metadata lock. If the connection that issued HANDLER statement also had other metadata locks (say, acquired in scope of a transaction), a classical deadlock situation of mutual wait could occur. Incompatible change: entering LOCK TABLES mode automatically closes all open HANDLERs in the current connection. Incompatible change: previously an attempt to wait on a lock in a connection that has an open HANDLER statement could wait indefinitely/deadlock. After this patch, an error ER_LOCK_DEADLOCK is produced. The idea of the fix is to merge thd->handler_mdl_context with the main mdl_context of the connection, used for transactional locks. This makes deadlock detection possible, since all waits with locks are "visible" and available to analysis in a single MDL context of the connection. Since HANDLER locks and transactional locks have a different life cycle -- HANDLERs are explicitly open and closed, and so are HANDLER locks, explicitly acquired and released, whereas transactional locks "accumulate" till the end of a transaction and are released only with COMMIT, ROLLBACK and ROLLBACK TO SAVEPOINT, a concept of "sentinel" was introduced to MDL_context. All locks, HANDLER and others, reside in the same linked list. However, a selected element of the list separates locks with different life cycle. HANDLER locks always reside at the end of the list, after the sentinel. Transactional locks are prepended to the beginning of the list, before the sentinel. Thus, ROLLBACK, COMMIT or ROLLBACK TO SAVEPOINT, only release those locks that reside before the sentinel. HANDLER locks must be released explicitly as part of HANDLER CLOSE statement, or an implicit close. The same approach with sentinel is also employed for LOCK TABLES locks. Since HANDLER and LOCK TABLES statement has never worked together, the implementation is made simple and only maintains one sentinel, which is used either for HANDLER locks, or for LOCK TABLES locks.
2009-12-22 19:09:15 +03:00
--reap
handler t1 close;
--echo # --> connection con1
connection con1;
disconnect con1;
--source include/wait_until_disconnected.inc
--echo # --> connection con2
connection con2;
disconnect con2;
--source include/wait_until_disconnected.inc
--echo # --> connection con3
connection con3;
disconnect con3;
--source include/wait_until_disconnected.inc
connection default;
--echo #
--echo # A temporary table test.
--echo # Check that we don't loose positions of HANDLER opened
--echo # against a temporary table.
--echo #
create table t1 (a int, b int, key a (a));
insert into t1 (a) values (1), (2), (3), (4), (5);
create temporary table t2 (a int, b int, key a (a));
insert into t2 (a) select a from t1;
handler t1 open;
handler t1 read a next;
handler t2 open;
handler t2 read a next;
flush table t1;
handler t2 read a next;
--echo # Sic: the position is lost
handler t1 read a next;
select * from t1;
--echo # Sic: the position is not lost
handler t2 read a next;
--error ER_CANT_REOPEN_TABLE
select * from t2;
handler t2 read a next;
drop table t1;
drop temporary table t2;
--echo #
--echo # A test for lock_table_names()/unlock_table_names() function.
--echo # It should work properly in presence of open HANDLER.
--echo #
create table t1 (a int, b int, key a (a));
create table t2 like t1;
create table t3 like t1;
create table t4 like t1;
handler t1 open;
handler t2 open;
rename table t4 to t5, t3 to t4, t5 to t3;
handler t1 read first;
handler t2 read first;
drop table t1, t2, t3, t4;
Implement new type-of-operation-aware metadata locks. Add a wait-for graph based deadlock detector to the MDL subsystem. Fixes bug #46272 "MySQL 5.4.4, new MDL: unnecessary deadlock" and bug #37346 "innodb does not detect deadlock between update and alter table". The first bug manifested itself as an unwarranted abort of a transaction with ER_LOCK_DEADLOCK error by a concurrent ALTER statement, when this transaction tried to repeat use of a table, which it has already used in a similar fashion before ALTER started. The second bug showed up as a deadlock between table-level locks and InnoDB row locks, which was "detected" only after innodb_lock_wait_timeout timeout. A transaction would start using the table and modify a few rows. Then ALTER TABLE would come in, and start copying rows into a temporary table. Eventually it would stumble on the modified records and get blocked on a row lock. The first transaction would try to do more updates, and get blocked on thr_lock.c lock. This situation of circular wait would only get resolved by a timeout. Both these bugs stemmed from inadequate solutions to the problem of deadlocks occurring between different locking subsystems. In the first case we tried to avoid deadlocks between metadata locking and table-level locking subsystems, when upgrading shared metadata lock to exclusive one. Transactions holding the shared lock on the table and waiting for some table-level lock used to be aborted too aggressively. We also allowed ALTER TABLE to start in presence of transactions that modify the subject table. ALTER TABLE acquires TL_WRITE_ALLOW_READ lock at start, and that block all writes against the table (naturally, we don't want any writes to be lost when switching the old and the new table). TL_WRITE_ALLOW_READ lock, in turn, would block the started transaction on thr_lock.c lock, should they do more updates. This, again, lead to the need to abort such transactions. The second bug occurred simply because we didn't have any mechanism to detect deadlocks between the table-level locks in thr_lock.c and row-level locks in InnoDB, other than innodb_lock_wait_timeout. This patch solves both these problems by moving lock conflicts which are causing these deadlocks into the metadata locking subsystem, thus making it possible to avoid or detect such deadlocks inside MDL. To do this we introduce new type-of-operation-aware metadata locks, which allow MDL subsystem to know not only the fact that transaction has used or is going to use some object but also what kind of operation it has carried out or going to carry out on the object. This, along with the addition of a special kind of upgradable metadata lock, allows ALTER TABLE to wait until all transactions which has updated the table to go away. This solves the second issue. Another special type of upgradable metadata lock is acquired by LOCK TABLE WRITE. This second lock type allows to solve the first issue, since abortion of table-level locks in event of DDL under LOCK TABLES becomes also unnecessary. Below follows the list of incompatible changes introduced by this patch: - From now on, ALTER TABLE and CREATE/DROP TRIGGER SQL (i.e. those statements that acquire TL_WRITE_ALLOW_READ lock) wait for all transactions which has *updated* the table to complete. - From now on, LOCK TABLES ... WRITE, REPAIR/OPTIMIZE TABLE (i.e. all statements which acquire TL_WRITE table-level lock) wait for all transaction which *updated or read* from the table to complete. As a consequence, innodb_table_locks=0 option no longer applies to LOCK TABLES ... WRITE. - DROP DATABASE, DROP TABLE, RENAME TABLE no longer abort statements or transactions which use tables being dropped or renamed, and instead wait for these transactions to complete. - Since LOCK TABLES WRITE now takes a special metadata lock, not compatible with with reads or writes against the subject table and transaction-wide, thr_lock.c deadlock avoidance algorithm that used to ensure absence of deadlocks between LOCK TABLES WRITE and other statements is no longer sufficient, even for MyISAM. The wait-for graph based deadlock detector of MDL subsystem may sometimes be necessary and is involved. This may lead to ER_LOCK_DEADLOCK error produced for multi-statement transactions even if these only use MyISAM: session 1: session 2: begin; update t1 ... lock table t2 write, t1 write; -- gets a lock on t2, blocks on t1 update t2 ... (ER_LOCK_DEADLOCK) - Finally, support of LOW_PRIORITY option for LOCK TABLES ... WRITE was abandoned. LOCK TABLE ... LOW_PRIORITY WRITE from now on has the same priority as the usual LOCK TABLE ... WRITE. SELECT HIGH PRIORITY no longer trumps LOCK TABLE ... WRITE in the wait queue. - We do not take upgradable metadata locks on implicitly locked tables. So if one has, say, a view v1 that uses table t1, and issues: LOCK TABLE v1 WRITE; FLUSH TABLE t1; -- (or just 'FLUSH TABLES'), an error is produced. In order to be able to perform DDL on a table under LOCK TABLES, the table must be locked explicitly in the LOCK TABLES list.
2010-02-01 14:43:06 +03:00
--echo #
--echo # A test for FLUSH TABLES WITH READ LOCK and HANDLER statements.
--echo #
set autocommit=0;
create table t1 (a int, b int, key a (a));
insert into t1 (a, b) values (1, 1), (2, 1), (3, 2), (4, 2), (5, 5);
create table t2 like t1;
insert into t2 (a, b) select a, b from t1;
create table t3 like t1;
insert into t3 (a, b) select a, b from t1;
commit;
flush tables with read lock;
handler t1 open;
lock table t1 read;
--error ER_LOCK_OR_ACTIVE_TRANSACTION
handler t1 read next;
--echo # This implicitly leaves LOCK TABLES but doesn't drop the GLR
--error ER_NO_SUCH_TABLE
lock table not_exists_write read;
--echo # We still have the read lock.
--error ER_CANT_UPDATE_WITH_READLOCK
drop table t1;
handler t1 open;
select a from t2;
handler t1 read next;
flush tables with read lock;
handler t2 open;
flush tables with read lock;
handler t1 read next;
select a from t3;
handler t2 read next;
handler t1 close;
rollback;
handler t2 close;
--error ER_CANT_UPDATE_WITH_READLOCK
drop table t1;
commit;
flush tables;
--error ER_CANT_UPDATE_WITH_READLOCK
drop table t1;
unlock tables;
drop table t1;
set autocommit=default;
drop table t2, t3;
--echo #
--echo # HANDLER statement and operation-type aware metadata locks.
--echo # Check that when we clone a ticket for HANDLER we downrade
--echo # the lock.
--echo #
--echo # Establish an auxiliary connection con1.
connect (con1,localhost,root,,);
--echo # -> connection default
connection default;
create table t1 (a int, b int, key a (a));
insert into t1 (a, b) values (1, 1), (2, 1), (3, 2), (4, 2), (5, 5);
begin;
insert into t1 (a, b) values (6, 6);
handler t1 open;
handler t1 read a last;
insert into t1 (a, b) values (7, 7);
handler t1 read a last;
commit;
--echo # -> connection con1
connection con1;
--echo # Demonstrate that the HANDLER doesn't hold MDL_SHARED_WRITE.
lock table t1 write;
unlock tables;
--echo # -> connection default
connection default;
handler t1 read a prev;
handler t1 close;
--echo # Cleanup.
drop table t1;
--echo # -> connection con1
connection con1;
disconnect con1;
--source include/wait_until_disconnected.inc
--echo # -> connection default
connection default;
--echo #
--echo # A test for Bug#50555 "handler commands crash server in
--echo # my_hash_first()".
--echo #
--error ER_UNKNOWN_TABLE
handler no_such_table read no_such_index first;
--error ER_UNKNOWN_TABLE
handler no_such_table close;
--echo #
--echo # Bug#50907 Assertion `hash_tables->table->next == __null' on
--echo # HANDLER OPEN
--echo #
--disable_warnings
DROP TABLE IF EXISTS t1, t2;
--enable_warnings
CREATE TEMPORARY TABLE t1 (i INT);
CREATE TEMPORARY TABLE t2 (i INT);
# This used to trigger the assert
HANDLER t2 OPEN;
# This also used to trigger the assert
HANDLER t2 READ FIRST;
HANDLER t2 CLOSE;
DROP TABLE t1, t2;