MySQL Bugs: #58745: MTR spends long time checking for warnings

Bug #58745	MTR spends long time checking for warnings
Submitted:	5 Dec 2010 22:56	Modified:	2 May 2011 14:56
Reporter:	Elena Stepanova	Email Updates:
Status:	Closed	Impact on me:	None
Category:	Tools: MTR / mysql-test-run	Severity:	S5 (Performance)
Version:	5.5.8	OS:	Any
Assigned to:	Bjørn Munch	CPU Architecture:	Any

Description:
I have a test case which makes server write a considerable number of warnings in the error log (e.g 5,000). It is by design. While the test case itself takes about nothing, the total time MTR spends running it is ~60 sec. 

The servers were restarted 0 times
Spent 0.000 of 60 seconds executing testcases

With nowarnings it takes < 10 sec.

If the number of warnings is greater, the time grows too. 
It can also fail with 

Timeout for 'check warnings' expired after 90 secondsmysqltest failed but provided no output

(quoted literally, no end-of-line or a space)

How to repeat:
--source include/have_log_bin.inc

SET binlog_format=STATEMENT;
--disable_warnings
DROP TABLE IF EXISTS t, t2;

CREATE TABLE t ( i INT ) ENGINE = MyISAM;
CREATE TABLE t2 ( i INT ) ENGINE = InnoDB;

INSERT INTO t VALUES (1);
BEGIN;

--disable_query_log
let $num=5000;
while ($num)
{
        INSERT INTO t2 SELECT * FROM t;
        dec $num;
}
--enable_query_log
COMMIT;

DROP TABLE t, t2;

--exit

Suggested fix:
While 5,000 warnings is a fairly big amount, 50 seconds to process them seems too long, maybe there is a way to optimize it.

Some speedup may be possible, but it's implemented in a way which is not very optimal, doing the matching against suppressed patterns in SQL. I'm not sure it will be more optimal in Perl though; the number of comparisons needed is number of warnings x number of suppress patterns.

A workaround for a specific test could be to add --testcase-timeout=60 to <test>-master.opt; this will increase the check warnings timeout to 1/10 of that, or 6 minutes. Adjust according to your needs.

Changes to test suite. No changelog entry needed.