paulsm4 |
04-25-2011 09:45 PM |
Hi -
You definitely need to drill down deeper. This includes:
* "Explain plan" for critical queries
* "show status" during heavy loads
* use "iostat", "vmstat" and "top" during heavy loads
* etc
However...
Consider this advice (for a system running Drupal ... but it sounds like it might be applicable to your site, too):
Quote:
http://groups.drupal.org/node/19185
Re: query cache, you would want to increase the query cache limit. Cache table queries can easily contain more than 1M of data in a large site, and you want as much of the cache in memory as possible -- so a few suggestions:
1. Make sure all large queries are handled by memcache, then tune query cache limit for the 'smaller' intensive queries. Redirect memory savings to memcache.
2. Increase query cache limit after inspecting your slow log, then repeat. It is likely that you will see a few HUGE dumps in there. That is what you are trying find a place for or clean up. Probably variable, menu_router, og_ancestry, node_access.. and then your custom views.
3. Reduce larger query sizes by investigating the nature of the tables and the data. i.e., maybe something in the query itself will catch your eye -- "why is that node loading on this page?" or "why is this cache refreshing on every page load?" .. ;)
|
|