Alex Rivera | Logout

What's a good approach to managing the db connection in a Google Cloud SQL (GAE) Python app?

Asked 2012-04-15T01:13:01.917
10

I'm just learning Google App Engine and am trying to figure out a good approach to managing my database connection to a Google Cloud SQL instance (if you haven't used GC-SQL, basically, it's MySQL in the cloud, with a few limitations).

I'm using the python (2.7) GAE environment with the webapp2 framework for handling requests. I know the FAQ says that it's recommended that a new connection to the DB be made with each request, but I don't know what the recommended way of closing the connection is. Each time I try to drop tables during development, the GC-SQL hangs and "show processlist" shows that there are a bunch of processes (likely because I'm not closing the DB) and that one of them is waiting for a lock (likely the process trying to drop the tables). This is annoying and forces me to restart the GC-SQL instance (like restarting mysql-server service, I imagine). There are also occasional DB hiccups that I believe are related to the fact that I'm not really closing my DB connection.

So, for example, should I have a destructor on my webapp2.Requesthandler subclass instance to disconnect from the DB? GAE objects seem to be cached sometimes, so that's also something to consider. I suppose I could just connect/query/disconnect for each query, but this seems suboptimal.

I know this is a vague question, but I'm hoping someone who's played in this area can thow some tips my way.

Thanks in advance!

Update: I tried implementing a wrapper around methods that need a cursor, using Shay's answer as a starting point. I'm getting GAE errors. Here's a new question specific to that: What are the connection limits for Google Cloud SQL from App Engine, and how to best reuse DB connections?

python mysql google-app-engine webapp2 google-cloud-sql
Edit
Report

1 Answer

10

Here is a complete example of the helloworld example app from the Getting Started Guide. It is based on snippets from Shay Erlichmen and JJC, but this version is threadsafe.

You can use it like this:

  @with_db_cursor(do_commit = True)
  def get(self, cursor):
        cursor.execute('SELECT guestName, content, entryID FROM entries')

app.yaml

application: helloworld
version: 1
runtime: python27
api_version: 1
threadsafe: true

handlers:
- url: /.*
  script: helloworld.app

helloworld.py

import cgi
import logging
import os
import threading
import webapp2

from google.appengine.api import rdbms

_INSTANCE_NAME = <name goes here>

def _db_connect():
  return rdbms.connect(instance=_INSTANCE_NAME, database='guestbook')

_mydata = threading.local()

def with_db_cursor(do_commit = False):
  """ Decorator for managing DB connection by wrapping around web calls.

  Stores connections and open cursor count in a threadlocal
  between calls.  Sets a cursor variable in the wrapped function. Optionally
  does a commit.  Closes the cursor when wrapped method returns, and closes
  the DB connection if there are no outstanding cursors.

  If the wrapped method has a keyword argument 'existing_cursor', whose value
  is non-False, this wrapper is bypassed, as it is assumed another cursor is
  already in force because of an alternate call stack.
  """
  def method_wrap(method):
    def wrap(self, *args, **kwargs):
      if kwargs.get('existing_cursor', False):
        # Bypass everything if method called with existing open cursor.
        return method(self, None, *args, **kwargs)

      if not hasattr(_mydata, 'conn') or not _mydata.conn:
        _myd
answered 2012-05-03T20:10:29.560

Your Answer