Details
-
Bug
-
Status: Closed (View Workflow)
-
Major
-
Resolution: Fixed
-
10.6, 10.7
Description
The test case for reproducing purposes only, not suitable for the test suite!
The test is concurrent and non-deterministic. It fails for me, but it can vary on different machines and builds. There is no need to run it with --repeat, as it already has a loop inside, which can be adjusted, although it can be converted into a one-loop-multiple-repeat variation. Also, a random grammar or mysqlslap can be created based on the same idea
--source include/have_innodb.inc
|
|
|
SET innodb_lock_wait_timeout= 1, lock_wait_timeout= 1; |
|
|
CREATE TABLE t1 ( |
pk INT AUTO_INCREMENT, |
a INT, |
b VARCHAR(10), |
PRIMARY KEY (pk), |
KEY (a), |
KEY (b) |
) ENGINE=InnoDB;
|
|
|
CREATE TABLE t2 ( |
pk INT AUTO_INCREMENT, |
a INT, |
b VARCHAR(10), |
PRIMARY KEY (pk), |
KEY (a), |
FOREIGN KEY (b) REFERENCES t1(b) |
) ENGINE=InnoDB;
|
|
|
ALTER TABLE t1 ADD FOREIGN KEY (a) REFERENCES t2(a); |
|
|
--connect (con1,localhost,root,,test)
|
|
|
SET innodb_lock_wait_timeout= 1, lock_wait_timeout= 1; |
|
|
--let $run = 1000
|
while ($run)
|
{
|
--connection default |
ALTER TABLE t2 ADD INDEX IF NOT EXISTS idx (a); |
ALTER TABLE t1 ADD INDEX IF NOT EXISTS idx (a); |
|
|
--connection con1 |
--send |
DROP INDEX idx ON t2; |
--connection default |
--error 0,ER_LOCK_WAIT_TIMEOUT |
DROP INDEX idx ON t1; |
--connection con1 |
--error 0,ER_LOCK_WAIT_TIMEOUT |
--reap |
--dec $run |
}
|
|
|
# Cleanup
|
SET FOREIGN_KEY_CHECKS= OFF; |
DROP TABLE t1, t2; |
|
10.6 1193a793 |
mariadbd: /data/src/10.6/storage/innobase/handler/handler0alter.cc:8683: bool rollback_inplace_alter_table(Alter_inplace_info*, const TABLE*, row_prebuilt_t*): Assertion `ctx->trx->state == TRX_STATE_ACTIVE' failed.
|
211026 2:02:34 [ERROR] mysqld got signal 6 ;
|
|
|
#6 0x00007f8a52536729 in __assert_fail_base (fmt=0x7f8a526cc588 "%s%s%s:%u: %s%sAssertion `%s' failed.\n%n", assertion=0x5567109861d0 "ctx->trx->state == TRX_STATE_ACTIVE", file=0x5567109808d0 "/data/src/10.6/storage/innobase/handler/handler0alter.cc", line=8683, function=<optimized out>) at assert.c:92
|
#7 0x00007f8a52547f36 in __GI___assert_fail (assertion=0x5567109861d0 "ctx->trx->state == TRX_STATE_ACTIVE", file=0x5567109808d0 "/data/src/10.6/storage/innobase/handler/handler0alter.cc", line=8683, function=0x556710987028 "bool rollback_inplace_alter_table(Alter_inplace_info*, const TABLE*, row_prebuilt_t*)") at assert.c:101
|
#8 0x0000556710105b9c in rollback_inplace_alter_table (ha_alter_info=0x7f8a47765c70, table=0x7f8a1cc9ae78, prebuilt=0x7f8a1cc9b968) at /data/src/10.6/storage/innobase/handler/handler0alter.cc:8683
|
#9 0x00005567100f4f7a in ha_innobase::commit_inplace_alter_table (this=0x7f8a1cc99980, altered_table=0x7f8a47765d30, ha_alter_info=0x7f8a47765c70, commit=false) at /data/src/10.6/storage/innobase/handler/handler0alter.cc:10791
|
#10 0x000055670fc71fb0 in handler::ha_commit_inplace_alter_table (this=0x7f8a1cc99980, altered_table=0x7f8a47765d30, ha_alter_info=0x7f8a47765c70, commit=false) at /data/src/10.6/sql/handler.cc:5194
|
#11 0x000055670f9ce31f in mysql_inplace_alter_table (thd=0x7f8a1c000db8, table_list=0x7f8a1c0156b8, table=0x7f8a1cc9ae78, altered_table=0x7f8a47765d30, ha_alter_info=0x7f8a47765c70, target_mdl_request=0x7f8a47766530, ddl_log_state=0x7f8a47765c10, trigger_param=0x7f8a47766100, alter_ctx=0x7f8a47767090) at /data/src/10.6/sql/sql_table.cc:7535
|
#12 0x000055670f9d6674 in mysql_alter_table (thd=0x7f8a1c000db8, new_db=0x7f8a1c0156d0, new_name=0x7f8a1c0156e0, create_info=0x7f8a47768000, table_list=0x7f8a1c0156b8, alter_info=0x7f8a47767f10, order_num=0, order=0x0, ignore=false, if_exists=false) at /data/src/10.6/sql/sql_table.cc:10204
|
#13 0x000055670f8cadc7 in mysql_execute_command (thd=0x7f8a1c000db8, is_called_from_prepared_stmt=false) at /data/src/10.6/sql/sql_parse.cc:4211
|
#14 0x000055670f8d7cb3 in mysql_parse (thd=0x7f8a1c000db8, rawbuf=0x7f8a1c0155c0 "DROP INDEX idx ON t1", length=20, parser_state=0x7f8a47768480) at /data/src/10.6/sql/sql_parse.cc:8030
|
#15 0x000055670f8c40e3 in dispatch_command (command=COM_QUERY, thd=0x7f8a1c000db8, packet=0x7f8a1c00b879 "DROP INDEX idx ON t1", packet_length=20, blocking=true) at /data/src/10.6/sql/sql_parse.cc:1896
|
#16 0x000055670f8c2a7f in do_command (thd=0x7f8a1c000db8, blocking=true) at /data/src/10.6/sql/sql_parse.cc:1404
|
#17 0x000055670fa81123 in do_handle_one_connection (connect=0x5567126809d8, put_in_cache=true) at /data/src/10.6/sql/sql_connect.cc:1418
|
#18 0x000055670fa80db3 in handle_one_connection (arg=0x556712632e08) at /data/src/10.6/sql/sql_connect.cc:1312
|
#19 0x000055670ffb204d in pfs_spawn_thread (arg=0x5567126b2a28) at /data/src/10.6/storage/perfschema/pfs.cc:2201
|
#20 0x00007f8a52a60609 in start_thread (arg=<optimized out>) at pthread_create.c:477
|
#21 0x00007f8a52633293 in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:95
|
Reproducible on 10.6/10.7 debug builds.
Not reproducible on 10.5.
The failure apparently started happening after this commit:
commit c3c53926c467c95386ae98d61ada87294bd61478
|
Author: Marko Mäkelä
|
Date: Mon Oct 18 18:03:12 2021 +0300
|
|
|
MDEV-26554: Races between INSERT on child and DDL on parent table
|
|
The SQL layer never acquires metadata locks (MDL) on the tables
|