mariadb/mysql-test/suite/innodb/r/read_only_recovery.result
Marko Mäkelä 2f00b73a4b MDEV-11985 Make innodb_read_only shutdown more robust
If InnoDB is started in innodb_read_only mode such that
recovered incomplete transactions exist at startup
(but the redo logs are clean), an assertion will fail at shutdown,
because there would exist some non-prepared transactions.

logs_empty_and_mark_files_at_shutdown(): Do not wait for incomplete
transactions to finish if innodb_read_only or innodb_force_recovery>=3.
Wait for purge to finish in only one place.

trx_sys_close(): Relax the assertion that would fail first.

trx_free_prepared(): Also free recovered TRX_STATE_ACTIVE transactions
if innodb_read_only or innodb_force_recovery>=3.

Also, revert my earlier fix to MySQL 5.7 because this fix is more generic:

Bug#20874411 INNODB SHUTDOWN HANGS IF INNODB_FORCE_RECOVERY>=3
SKIPPED ANY ROLLBACK

trx_undo_fake_prepared(): Remove.

trx_sys_any_active_transactions(): Revert the changes.
2017-02-06 10:45:18 +02:00

36 lines
957 B
Text

connect con1, localhost, root;
CREATE TABLE t(a INT PRIMARY KEY) ENGINE=InnoDB;
BEGIN;
INSERT INTO t VALUES(1),(2);
DELETE FROM t WHERE a=2;
connection default;
# Normal MariaDB shutdown would roll back the above transaction.
# We want the transaction to remain open, so we will kill the server
# after ensuring that any non-transactional files are clean.
FLUSH TABLES;
# Ensure that the above incomplete transaction becomes durable.
SET GLOBAL innodb_flush_log_at_trx_commit=1;
BEGIN;
INSERT INTO t VALUES(0);
ROLLBACK;
# Kill and restart: --innodb-force-recovery=3
disconnect con1;
SELECT * FROM t;
a
SET TRANSACTION ISOLATION LEVEL READ UNCOMMITTED;
SELECT * FROM t;
a
1
# Starting with MariaDB 10.2, innodb_read_only implies READ UNCOMMITTED.
# In earlier versions, this would return the last committed version
# (empty table)!
SELECT * FROM t;
a
1
SET TRANSACTION ISOLATION LEVEL READ UNCOMMITTED;
SELECT * FROM t;
a
1
SELECT * FROM t;
a
DROP TABLE t;