Hi all, First off, I'm having a great time using and learning about Storm, so thankyou to all involved!
I'm using Storm v 0.12 against a MySQL 5 DB in a large-ish project that is to be deployed using a mod_python / Apache prefork app server model (multiple stores in use in the app, many app servers involved, and many child worker processes configured for each app server) and have discovered that this is a recipe for quite a few database connections being held open simultaneously due to the way a Storm store keeps a database connection for re-use, and the default mysqld wait_time configuration setting of 8 hours. I'm currently exploring the possibilities for reducing the number of connections open simultaneously to the DB server, and was wondering if there is any best practice (or even sighting in the wild) for using a Storm Store in a request -> connect-to-db -> get results -> disconnect-from-db type model, although my current rough take on Storm's design so far makes this approach seem anathema to me frankly. I've looked at Store.close() / Connection.close(), and all of the rollback and disconnection checking contributed by James Henstridge, and remain stymied. As far as I can tell, one can close a Store down easily enough, which will wipe the underlying Connection's raw_connection, but there is no exposed way to start one back up. Am I missing something obvious, or am I possibly faced with reaching in through the Store and all the way down to the Store's Connection's Database's raw connection? Thanks! -Jer -- storm mailing list [email protected] Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/storm
