mariadb/mysql-test/suite/perfschema/t/global_read_lock.test
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

77 lines
2 KiB
Text

# Tests for PERFORMANCE_SCHEMA
#
# Test the effect of a flush tables with read lock on setup_ tables.
--source include/not_embedded.inc
--source include/have_perfschema.inc
use performance_schema;
# Make test robust against errors in other tests.
# Ensure that instrumentation is turned on when we create new connection.
update performance_schema.setup_instruments set enabled='YES';
create user pfsuser@localhost;
grant SELECT, UPDATE, LOCK TABLES on performance_schema.* to pfsuser@localhost;
flush privileges;
connect (con1, localhost, pfsuser, ,"*NO-ONE*");
lock tables performance_schema.setup_instruments read;
--disable_result_log
select * from performance_schema.setup_instruments;
--enable_result_log
unlock tables;
lock tables performance_schema.setup_instruments write;
update performance_schema.setup_instruments set enabled='NO';
update performance_schema.setup_instruments set enabled='YES';
unlock tables;
connection default;
flush tables with read lock;
connection con1;
lock tables performance_schema.setup_instruments read;
--disable_result_log
select * from performance_schema.setup_instruments;
--enable_result_log
unlock tables;
# This will block
--send
lock tables performance_schema.setup_instruments write;
connection default;
let $wait_condition= select 1 from performance_schema.events_waits_current where event_name like "wait/synch/cond/sql/MDL_context::COND_wait_status";
--source include/wait_condition.inc
# Observe the blocked thread in the performance schema :)
select event_name,
left(source, locate(":", source)) as short_source,
if(timer_end IS NULL, NULL, "SET") as timer_end,
if(timer_wait IS NULL, NULL, "SET") as timer_wait,
operation
from performance_schema.events_waits_current
where event_name like "wait/synch/cond/sql/MDL_context::COND_wait_status";
unlock tables;
connection con1;
--reap
update performance_schema.setup_instruments set enabled='NO';
update performance_schema.setup_instruments set enabled='YES';
unlock tables;
disconnect con1;
connection default;
drop user pfsuser@localhost;
flush privileges;