From a9840ba26e9f2af18c5f16ee79d0b008f98f22d0 Mon Sep 17 00:00:00 2001 From: Sujatha Sivakumar Date: Mon, 27 Oct 2014 11:49:59 +0530 Subject: [PATCH] Bug#19583563: RPL.RPL_SEMI_SYNC_SHUTDOWN_HANG FAILS SPORADICALLY ON PB2 Problem: ======== CURRENT_TEST: rpl.rpl_semi_sync_shutdown_hang mysqltest: At line 52: query 'reap' failed: 2013: Lost connection to MySQL server during query Analysis: ======== As part of test script with semisync enabled we stop the slave's IO thread and execute a CREATE TABLE on master. This statement will hang waiting for acknowledgement from slave and IO thread has gone away. This CREATE TABLE statement is kept in hang mode using --send command. Then master server goes down because of shutdown causing the CREATE TABLE statement to end successfully. When ever --send is used in MTR it expects a --reap command to be used to capture the return status of the statement that used --send. Because of timing issue some times --reap command is able to get the successful return status of --send command and some times the --reap command is executed post the server has actually shutdown. Because of this the reap command complains that it has lost connection to mysql server during query. --reap commands return code is not important for the actual bugfix verification. The main aim of test script is shutdown should not hang. Fix: === As part of fix --error 0,2013 is added above the --reap command. --- mysql-test/suite/rpl/t/rpl_semi_sync_shutdown_hang.test | 1 + 1 file changed, 1 insertion(+) diff --git a/mysql-test/suite/rpl/t/rpl_semi_sync_shutdown_hang.test b/mysql-test/suite/rpl/t/rpl_semi_sync_shutdown_hang.test index c632ee1b971..397450e19d5 100644 --- a/mysql-test/suite/rpl/t/rpl_semi_sync_shutdown_hang.test +++ b/mysql-test/suite/rpl/t/rpl_semi_sync_shutdown_hang.test @@ -49,6 +49,7 @@ SET GLOBAL rpl_semi_sync_master_timeout = 10000000; --enable_result_log --source include/rpl_connection_master.inc +--error 0,2013 --reap --let $rpl_server_number=1 --source include/rpl_start_server.inc