I've been searching for a while now but can't find an answere. I know that cherrypy creates a new thread for handling requests (GET, PUT, POST, DELETE etc).
Now i fetch the parameters like this:
...
@cherrypy.tools.json_in()
@cherrypy.tools.json_out()
def POST(self):
Forum.lock_post.acquire()
conn = self.io.psqlConnect(self.dict_psql)
cur = conn.cursor(cursor_factory = psycopg2.extras.RealDictCursor)
params = cherrypy.request.json
...
return some_dict
As you can see im locking the thread to avoid race condition on the variable params. But is this really necessary? I'm asking cos if i do it like this all the other requests on POST will have to wait. Is there any better solution without locking the whole POST? I'm using params several times along the code.
First a clarification, CherryPy doesn't create a new thread for each requests, it has a predetermined pool of threads (10 by default), from which indeed one thread can be used to handle a single request at a time.
As for if you should lock cherrypy.request.json. You really don't, there is a concept called "thread locals" on which you can have multiple references to different objects depending on which thread is accessing such object. (python docs).
Having said that... you should make sure that the code that you write doesn't interfere with the state of the other threads (you can use the cherrypy.thread_data as a quick fix).
Take a look into the cherrypy plugin architecture, if you want a resource to be shared among threads usually a plugin is the way to: http://docs.cherrypy.org/en/latest/extend.html#plugins