mariadb/mysql-test/main/wait_timeout.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

135 lines
4 KiB
Text

# Last modification:
# 2009-01-19 H.Hunger Fix Bug#39108 main.wait_timeout fails sporadically
# - Increase wait timeout to 2 seconds
# - Eliminated the corresponding opt file,
# set global wait timeout within the test.
# - Replaced sleeps by wait condition
# - Minor improvements
###############################################################################
# These tests cannot run with the embedded server
-- source include/not_embedded.inc
#-- source include/one_thread_per_connection.inc
#
# Bug#8731 wait_timeout does not work on Mac OS X
#
let $start_value= `SELECT @@global.wait_timeout`;
SET @@global.wait_timeout= 2;
disconnect default;
# Connect with another connection and reset counters
connect (wait_con,localhost,root,,test,,);
connection wait_con;
--disable_query_log
SET SESSION wait_timeout=100;
let $retries=300;
SET @aborted_clients= 0;
--enable_query_log
# Disable reconnect and do the query
connect (default,localhost,root,,test,,);
connection default;
--echo --disable_reconnect;
--disable_reconnect
SELECT 1;
# Switch to wait_con and wait until server has aborted the connection
connection wait_con;
--disable_query_log
while (!`select @aborted_clients`)
{
real_sleep 0.1;
let $aborted_clients = `SHOW STATUS LIKE 'aborted_clients'`;
eval SET @aborted_clients= SUBSTRING('$aborted_clients', 16)+0;
dec $retries;
if (!$retries)
{
die Failed to detect that client has been aborted;
}
}
--enable_query_log
# The server has disconnected, add small sleep to make sure
# the disconnect has reached client
let $wait_condition= SELECT COUNT(*)=1 FROM information_schema.processlist;
--source include/wait_condition.inc
connection default;
# When the connection is closed in this way, the error code should
# be consistent see Bug#2845 for an explanation
# depending on platform/client, either errno 2006 or 2013 can occur below
--error 2006,2013,5014
SELECT 2;
--echo --enable_reconnect;
--enable_reconnect
SELECT 3;
# Disconnect so that we will not be confused by a future abort from this
# connection.
disconnect default;
#
# Do the same test as above on a TCP connection
# (which we get by specifying an ip adress)
# Connect with another connection and reset counters
connection wait_con;
--disable_query_log
FLUSH STATUS; # Reset counters
let $retries=300;
SET @aborted_clients= 0;
--enable_query_log
connect (con1,127.0.0.1,root,,test,$MASTER_MYPORT,);
--echo --disable_reconnect;
--disable_reconnect
SELECT 1;
# Switch to wait_con and wait until server has aborted the connection
connection wait_con;
--disable_query_log
while (!`select @aborted_clients`)
{
real_sleep 0.1;
let $aborted_clients = `SHOW STATUS LIKE 'aborted_clients'`;
eval SET @aborted_clients= SUBSTRING('$aborted_clients', 16)+0;
dec $retries;
if (!$retries)
{
die Failed to detect that client has been aborted;
}
}
--enable_query_log
# The server has disconnected, add small sleep to make sure
# the disconnect has reached client
let $wait_condition= SELECT COUNT(*)=1 FROM information_schema.processlist;
--source include/wait_condition.inc
disconnect wait_con;
connection con1;
# When the connection is closed in this way, the error code should
# be consistent see Bug#2845 for an explanation
# depending on platform/client, either errno 2006 or 2013 can occur below
--error 2006,2013,5014
SELECT 2;
--echo --enable_reconnect;
--enable_reconnect
SELECT 3;
--replace_result $start_value <start_value>
eval SET @@global.wait_timeout= $start_value;
disconnect con1;
connect (default,localhost,root,,test,,);
#
# MDEV-7775 Wrong error message (Unknown error) when idle sessions are killed after wait_timeout
#
set global log_warnings=2;
connect (foo,localhost,root);
set @@wait_timeout=1;
sleep 2;
connection default;
let SEARCH_FILE=$MYSQLTEST_VARDIR/log/mysqld.1.err;
let SEARCH_PATTERN= Aborted.*Got timeout reading communication packets;
source include/search_pattern_in_file.inc;
set global log_warnings=@@log_warnings;