MySQL Bugs: #38228: binlog_format_basic At line 82: query 'SELECT ... failed: 2013: Lost connection

Bug #38228	binlog_format_basic At line 82: query 'SELECT ... failed: 2013: Lost connection
Submitted:	18 Jul 2008 16:09	Modified:	19 Dec 2008 9:57
Reporter:	Andrei Elkin	Email Updates:
Status:	Can't repeat	Impact on me:	None
Category:	MySQL Server	Severity:	S2 (Serious)
Version:	5.1-bugteam	OS:	Any
Assigned to:	Sven Sandberg	CPU Architecture:	Any
Tags:	binlog_format_basic, crash, pushbuild, sporadic, test failure

Description:
        tfl_name: main.binlog_format_basic
    tfl_tst_type: n_mix
        psh_tree: bzr_mysql-5.1-bugteam
       psh_order: 67
tfl_bld_platform: powermacg5
        tfl_text: main.binlog_format_basic       [ fail ]
mysqltest: At line 82: query 'SELECT @@SESSION.binlog_format = VARIABLE_VALUE 
FROM INFORMATION_SCHEMA.SESSION_VARIABLES 
WHERE VARIABLE_NAME='binlog_format'' failed: 2013: Lost connection to MySQL server during query

The result from queries just before the failure was:
< snip >
SELECT @@GLOBAL.binlog_format = VARIABLE_VALUE
FROM INFORMATION_SCHEMA.GLOBAL_VARIABLES
WHERE VARIABLE_NAME='binlog_format';
@@GLOBAL.binlog_format = VARIABLE_VALUE
1
1 Expected
SELECT COUNT(@@GLOBAL.binlog_format);
COUNT(@@GLOBAL.binlog_format)
1
1 Expected
SELECT COUNT(VARIABLE_VALUE)
FROM INFORMATION_SCHEMA.GLOBAL_VARIABLES 
WHERE VARIABLE_NAME='binlog_format';
COUNT(VARIABLE_VALUE)
1
1 Expected
'#---------------------BS_STVARS_002_04----------------------#'
SELECT @@SESSION.binlog_format = VARIABLE_VALUE 
FROM INFORMATION_SCHEMA.SESSION_VARIABLES 
WHERE VARIABLE_NAME='binlog_format';

More results from queries before failure can be found in /Users/henry/pb/bzr_mysql-5.1-bugteam/67/mysql-5.1.28-pb67/mysql-test/var-n_mix-7/log/binlog_format_basic.log

How to repeat:
https://intranet.mysql.com/secure/pushbuild/xref.pl?startdate=&enddate=&dir=&plat=powermac...
https://intranet.mysql.com/secure/pushbuild/showpush.pl?dir=bzr_mysql-5.1-bugteam&order=67

Difficult to analyze since little debug info is left on server.

Setting to "To be fixed later".

Solving this bug is dependent on having the code for BUG#38181 in the
source code.  Until BUG#38181 is fixed, there is no way to work on
this bug.

After BUG#38181 has been pushed and if the failure re-occur, 
then this bug can be set back to "Verified".

Here's a better xref link: http://tinyurl.com/6odxa8
The failure has only happened four times (three of which were in the same push; a push in which almost all tests seem to have failed).

Setting to Can't repeat, since this has not happened since July (except one push when most tests failed, so there was something wrong with that push, unrelated to the present bug).