[rootnik]

Originally Posted by bada_bing

Has anyone figured out why this hack sucks up processor utilization?

Most of it has to do with the way messages are deleted.

When you prune messages, or your members delete messages, it just deletes the message ID (not the entire message)

So when the hourly cleanup script runs, it shifts through every message, for every user, and see's if it has a message ID. If it doesn't, it deletes the message.

The longer you go without pruning, the harder this hourly check is on your CPU. I've had over 3,000,000 rows examined by the cleanup query, all because I had a few thousand pager messages either in the system or deleted. This happened with a 3000 member board with about 50 people online at a time. This will shut ANY cpu down until the query finally times out... I had it hit my server so hard one time that I had to call the datacenter to request a reboot.

The last version of this hacks includes an option to automaticly delete messages that have been read. If you enable this it helps with the load problems because the cleanup script is not trying to do too much after your big 'prunes'.