I've noticed the Engine can end up with a runaway exception when using MySQL if the MySQL timeout is exceeded (typically 8 hours of no activity by default). This seems to be related to marking Engine as a singleton with injected resources trying to use an old mysql connection that is no longer valid. The engine should be smarter about lost connections and stop trying after a certain number of attempts, maybe using an exponential backoff. This could also probably happen if the database goes down, but DataWolf is still up. More testing needs to be done here because the runaway exception grows the size of the log file very quickly.

Temporary Workaround: Set timeout and interactive_timeout to 1 year (31536000).