Okay, so I need some help from you WP experts out there. While I've been looking under the hood now of WP for a few months, I haven't figured out all the nuances yet that make MU tick.. Here's the backstory and problem.
- 22 sites set up as subdomains, all can be found under competitor.com
- getting ready to migrate our last site over, velonews.competitor.com, which is our largest, 15-20mm pageviews per month with about 600k uniques. Total combined traffic comes to about 25mm pageviews with about 800k uniques.
- Running on 4 webservers (dual proc, dual core xeon with 4GB ram) and a db box (dual proc dual core xeon with 8gb of ram).
- Using W3T Cache, Page caching with memcache as well as db caching with memcache, and minifying.
- Mysql query caching is turned on.
Here's the issue we have:
- We have successfully handled 400k pageviews in a single day.
- When we flip the DNS for the velonews site, we begin to get the traffic and the web boxes handle things fine, but our DB box starts to explode.
- Within a few minutes, on a regular schedule, hundreds of db connections are opened up, begin to stack, don't close and bogs down the db.
- The wp_options table is seeing a lot of activity with _transient_doing_cron and a number, as well as _transient_rss, which appears to be caches of some rss feeds
- The 'transient' records are stacking up in the wp_options table, causing it to bloat and slow down.
So we're not sure first off, why the db connections are backing up, and secondly what's causing the high number of concurrent connections to open, time out, and remain open. We hit about 500 concurrent connections open, and not being serviced, the db slows to a crawl, and dies out.
So looking for help. Anyone have any ideas, suggestions, help, anything they can offer? We're getting desperate for help. If anyone has any suggestions, please let me know.
Thanks everyone !