mariadb/mysql-test/suite/storage_engine/trx/transaction_isolation.inc
Sergei Golubchik bead24b7f3 mariadb-test: wait on disconnect
Remove one of the major sources of race condiitons in mariadb-test.
Normally, mariadb_close() sends COM_QUIT to the server and immediately
disconnects. In mariadb-test it means the test can switch to another
connection and sends queries to the server before the server even
started parsing the COM_QUIT packet and these queries can see the
connection as fully active, as it didn't reach dispatch_command yet.

This is a major source of instability in tests and many - but not all,
still less than a half - tests employ workarounds. The correct one
is a pair count_sessions.inc/wait_until_count_sessions.inc.
Also very popular was wait_until_disconnected.inc, which was completely
useless, because it verifies that the connection is closed, and after
disconnect it always is, it didn't verify whether the server processed
COM_QUIT. Sadly the placebo was as widely used as the real thing.

Let's fix this by making mariadb-test `disconnect` command _to wait_ for
the server to confirm. This makes almost all workarounds redundant.

In some cases count_sessions.inc/wait_until_count_sessions.inc is still
needed, though, as only `disconnect` command is changed:

 * after external tools, like `exec $MYSQL`
 * after failed `connect` command
 * replication, after `STOP SLAVE`
 * Federated/CONNECT/SPIDER/etc after `DROP TABLE`

and also in some XA tests, because an XA transaction is dissociated from
the THD very late, after the server has closed the client connection.

Collateral cleanups: fix comments, remove some redundant statements:
 * DROP IF EXISTS if nothing is known to exist
 * DROP table/view before DROP DATABASE
 * REVOKE privileges before DROP USER
 etc
2025-07-16 09:14:33 +07:00

87 lines
2.2 KiB
PHP

#
# Basic check for transaction isolation.
# The results should be different depending on the isolation level.
# For some isolation levels, some statements will end with a timeout.
# If the engine has its own timeout parameters, reduce them to minimum,
# otherwise the test will take very long.
# If the timeout value is greater than the testcase-timeout the test is run with,
# it might fail due to the testcase timeout.
#
--source ../have_engine.inc
connect (con1,localhost,root,,);
eval SET SESSION TRANSACTION ISOLATION LEVEL $trx_isolation;
connect (con2,localhost,root,,);
eval SET SESSION TRANSACTION ISOLATION LEVEL $trx_isolation;
connection con1;
let $create_definition = a $int_col;
--source ../create_table.inc
START TRANSACTION;
--sorted_result
SELECT a FROM t1; # First snapshot
connection con2;
BEGIN;
--let $error_codes = 0,ER_LOCK_WAIT_TIMEOUT
INSERT INTO t1 (a) VALUES(1);
--source ../strict_check_errors.inc
connection con1;
--sorted_result
SELECT a FROM t1; # Second snapshot
connection con2;
--let $error_codes = 0,ER_LOCK_WAIT_TIMEOUT
INSERT INTO t1 (a) VALUES (2);
--source ../strict_check_errors.inc
connection con1;
--sorted_result
SELECT a FROM t1; # Third snapshot
--let $error_codes = 0,ER_LOCK_WAIT_TIMEOUT
INSERT INTO t1 (a) SELECT a+100 FROM t1;
--source ../strict_check_errors.inc
--sorted_result
SELECT a FROM t1;
connection con2;
--sorted_result
SELECT a FROM t1; # Inside the transaction
COMMIT;
--sorted_result
SELECT a FROM t1; # Outside the transaction
connection con1;
--sorted_result
SELECT a FROM t1; # Inside the transaction
# Note: INSERT .. SELECT might be tricky, for example for InnoDB
# even with REPEATABLE-READ it works as if it is executed with READ COMMITTED.
# The test will have a 'logical' result for repeatable read, even although
# we currently don't have an engine which works this way.
--let $error_codes = 0,ER_LOCK_WAIT_TIMEOUT
INSERT INTO t1 (a) SELECT a+200 FROM t1;
--source ../strict_check_errors.inc
--sorted_result
SELECT a FROM t1;
COMMIT;
--sorted_result
SELECT a FROM t1; # Outside the transaction
connection con2;
--sorted_result
SELECT a FROM t1; # After both transactions have committed
connection default;
disconnect con1;
disconnect con2;
DROP TABLE t1;