AppDB Issues

[Jeremy Newman]

As many of you have noticed we've been having a bit of server trauma 
this past month. The issue boils down to a bug somewhere in mysqld. We 
are still trying to find the exact cause of the bug. We do have some 
leads. I just wanted to do a brain dump on wine-devel here before I go 
on Vacation for a week.

The problem:
mysqld will randomly eat up all resources on the system until the server 
is completely unresponsive to anything but pings.

My band-aids thus far:
1. mysql is now running at a nice of 1. This should allow us to at least 
ssh in and kill it if it gets out of control again.

2. I've removed InnoDB table support to save on memory. Not many tables 
where using it anyway, and nothing is using transaction support at this 
time.

3. I've patched the appdb to use memcached for session management. This 
will reduce some load on the DB as the session table was busiest and 
slowest table of the appdb.

These fixes should give longer uptime, but they are not solving the core 
issue.

While I'm gone, if the server goes offline again, please contact 
Alexandre or Jeremy White. They can kick proverbial butt of the server 
to get it going again.

Lastly, sorry for the issues.

-Newman

"We do what we must because we can" - GLaDOS