New answers tagged percona
Ultimately the problem was SELinux on CentOS, and the answer was to disable SELinux! I couldn't find any errors in the SELinux log related to MySQL so this took quite some time to track down.
You might check your value for the innodb_stats_sample_pages parameter. It controls how many index dives MySQL performs on a table when updating index statistics, which in turn are used to calculate the cost of a candidate join plan. The default value was 8 for the version we were using. We changed it to 128, and observed fewer unexpected join plans.
Actually, your problem comes from temporary tables, that may be created in MyISAM format (in fact, up to 5.5, this was the only format for temporary tables on disk), not from the system tables in the mysql database, that are mostly also in MyISAM format. This may be caused by a new feature of 5.6, where subquery execution may generate indexes for faster ...
First of all: This is not because of Percona, the differences you are getting are because of the upgrade to 5.6. Percona Server rarely modifies the SQL optimiser from upstream. The changes come from the new optimiser in MySQL 5.6, which Percona Server 5.6 uses and that was heavily optimised between the two version (usually, for the best). The reason why the ...
Top 50 recent answers are included