I just had this happen, with the following as first lines of the errorlog: 100216 17:48:01 [ERROR] /usr/sbin/mysqld: Incorrect key file for table > '/var/tmp/#sql_46b2_0.MYI'; try to repair it > 100216 17:48:02 [ERROR] /usr/sbin/mysqld: Incorrect key file for table > '/var/tmp/#sql_46b2_0.MYI'; try to repair it > *** glibc detected *** /usr/sbin/mysqld: double free or corruption (out): > 0x00002aaadd97b630 *** >
I know what I was doing, though, so I suspect the following scenario: * I was doing stuff, which filled up the /var/tmp filesystem * MySQL can't write some tmp tables, notices they're incorrect and tries to repair them * In the repair thread, it hits the glibc issue. I have also traced the thread (and thus, query) that caused it, and like Ćukasz' post above, it's a pretty complex query with joins, wich indeed leads to the probable use of tmp tables. On the MySQL boards I find reference to this bug (on ndb), and it maybe being fixed with a number of memory leak fixes in 5.1.10: http://bugs.mysql.com/bug.php?id=19154 -- Bier met grenadyn Is als mosterd by den wyn Sy die't drinkt, is eene kwezel Hy die't drinkt, is ras een ezel