Bug #48876 Test "query_cache" uses a value which exceeds platform (?) limit
Submitted: 18 Nov 2009 18:22 Modified: 7 Aug 2013 15:37
Reporter: Joerg Bruehe Email Updates:
Status: Can't repeat Impact on me:
None 
Category:Tests: Server Severity:S3 (Non-critical)
Version:5.5.0-beta OS:FreeBSD (6.3 x86)
Assigned to: CPU Architecture:Any
Triage: Triaged: D3 (Medium)

[18 Nov 2009 18:22] Joerg Bruehe
Description:
This test failure seems to be new in 5.5.0-beta,
at least I don't have a previous matching report accessible.

=====
main.query_cache                         [ retry-fail ]
        Test ended at YYYY-MM-DD HH:MM:SS

CURRENT_TEST: main.query_cache
--- /PATH/mysql-test/r/query_cache.result
+++ /PATH/mysql-test/r/query_cache.reject
@@ -1709,6 +1709,8 @@
 SET GLOBAL query_cache_size= default;
 End of 5.0 tests
 SET GLOBAL query_cache_size=1024*1024*512;
+Warnings:
+Warning        1282    Query cache failed to set size 536870912; new query cache size is 0
 CREATE TABLE t1 (a ENUM('rainbow'));
 INSERT INTO t1 VALUES (),(),(),(),();
 SELECT 1 FROM t1 GROUP BY (SELECT 1 FROM t1 ORDER BY AVG(LAST_INSERT_ID()));

mysqltest: Result content mismatch
=====

Until now, I see it only on FreeBSD 6.3, x86;
especially, the test passes on FreeBSD 6.3 (x86_64), on FreeBSD 7.0 (any CPU), and all other operating systems.

Maybe it is "just" a machine issue,
that would show a mismatch between test development assumptions and available hardware.

How to repeat:
Found while doing a release build.

Suggested fix:
Reduce the value ?
[7 Aug 2013 15:37] Erlend Dahl
No longer seen in our testing of 5.5, 5.6 and trunk.