Proper way to use Peewee ORM Pooled Connection

Question

I'm using the Peewee ORM in my Python WSGI web app. I've been having issues with the MySQL Server "going away", so I switched to the PooledMySQLDatabase. I will still having issues with the server going away after a few hours..

According to the Docs, for best performance I must manage the connection(s) myself, by calling .connect() and .close(). I was not doing this at all.

I've added 2 decorators that get called before and after every request, but is this correct?

@pre_save()
def pre_save_handler(sender,instance,created):
    db.connect()
    logger.debug('Attempting to save %s' % instance)

@post_save()
def post_save_handler(sender, instance, created):
    db.close()
    logger.debug('%s was just saved' % instance)

Are there any pitfalls doing it this way? Should I move the .connect() and .close() into my application?

coleifer · Accepted Answer · 2015-01-14T16:32:52.320

1

Most WSGI frameworks provide some sort of hook for running code before and after a request. I suggest using those hooks to connect/close.

Check out these docs: http://docs.peewee-orm.com/en/latest/peewee/database.html#adding-request-hooks

edited Jan 14 '15 at 16:32

answered Jan 13 '15 at 16:33

coleifer

24,887
6
60
75

Thank you for the doc pointer. I'm actually using http://webpy.org/ , which has hooks -http://webpy.org/cookbook/application_processors . Through my testing it seems that peewee will call .connect() for me if I don't. Is it safe to assume peewee will do this in all situations, or should I explicitly call .connect() AND .close() – mox1 Jan 14 '15 at 17:53

score 0 · Answer 2 · answered Jan 15 '15 at 15:57

After thinking about this for a day or two, I realized this is 100% the wrong way to do it. This won't work at all for SELECT queries. These functions only get called on database stores! I've added a pre and post processor to my web app and it is now functioning correctly.

Proper way to use Peewee ORM Pooled Connection

2 Answers2