异步在Tornado中如何工作？

Question

All examples in the Tornado documents show how we can make further HTTP requests asynchronously using Tornado.

http_client = httpclient.AsyncHTTPClient()
http_client.fetch("http://www.google.com/", handle_request)

While I want to ensure that my tornado doesn't wait for response from database (get command of couchbase) to send next.

Check my other bug to understand why I want async call to couchbase.

Answer 1

As far as I know Couchbase driver is not an async driver so this wont work. Make the db call synchronous and optimize it to be as fast as possible.

The (scary) alternative is to take couchbase's driver and patch it with something like gevent to create your own async driver. I do not recommend doing this. If you really want to go down this route look at what this guy did with Motor - He made the mongo driver async.

Answer 2

See the Tornado wiki on github and this similar question on StackOverflow . Just do your database calls sync. If your database or your calls to the database/the database driver is your bottleneck your website will not be faster even if tornado is not blocked. Doing asynchronous calls may be more senseful when requesting ressources not under your control as other webservers or ressources which are not vital to most parts of your website like file I/Os.

Answer 3

Old thread, but I should probably mention that couchbase has support for twisted . See Twisted API for Couchbase not working with Python Tornado and http://pythonhosted.org/couchbase/api/txcouchbase.html