After a bit more of searching I've finally figured out what I wanted is "atomic updates".
Using the F() expression seems to be the best solution without much code overhaul: http://stackoverflow.com/questions/280075/atomic-operations-in-django http://efreedom.com/Question/1-1598932/Atomic-Increment-Counter-Django Thanks JeeyoungKim for pointing to the right direction. On 11月23日, 午後5:52, Iqbal Abdullah <iq...@marimore.co.jp> wrote: > Since we only run the application on a single server we're using file > based locking; fail_to_get_lock() and release_lock() tries to create a > file in the tmp directory with the user's unique identification data, > and release_lock() deletes it. > > What I don't understand is, even if the save() method is not thread > safe, am I correct to assume that everytime I call something like > object = MyModel.object.get(id=1) I will get the latest data for > object? > > On 11月20日, 午後2:18, JeeyoungKim <jeeyou...@gmail.com> wrote: > > > I'm wondering, how are fail_to_get_lock and release_lock implemented? > > It would be some kind of multiprocess lock, which seems overly > > complicated.. > > > using save() in this case is cumbersome, because save() isnot > > threadsafe... you'll have to try to maintain your own lock. > > > If you only want to implement increments and decrements, you can use > > F() object and update() methods to update the database. > > > I have an example, here > > >http://pastebin.com/qp4ExWC2 > > > On Nov 19, 4:30 pm, Iqbal Abdullah <iq...@marimore.co.jp> wrote: > > > > Hi Steve, > > > > Ops, yes, I forgot to include the object.save() in the pseudo code > > > above. Yes, we're actually saving it: > > > > 1 object = MyModel.object.get(id=1) > > > 2 print object.value # starting value is 5 > > > 3 while object.fail_to_get_lock(): > > > 4 sleep(5) > > > 5 object = MyModel.object.get(id=1) # Re-get the object so we can > > > have the latest state > > > 6 object.value = object.value - 1 > > > 7 object.save() > > > 8 print object.value # returns 4 > > > 9 object.release_lock() > > > > On 11月20日, 午前7:39, Steve Holden <holden...@gmail.com> wrote: > > > > > On 11/19/2010 5:35 PM, Iqbal Abdullah wrote: > > > > > > Hi, > > > > > > This might be a gotcha on themodelsside, but I would like > > > > > clarification and guidance on how to write the code better concerning > > > > >multipleprocess accessing the same data viamodels. > > > > > > I have a backend script that runs the following code inmultiple > > > > >processes: > > > > > > 1 object = MyModel.object.get(id=1) > > > > > 2 print object.value # starting value is 5 > > > > > 3 while object.fail_to_get_lock(): > > > > > 4 sleep(5) > > > > > 5 object = MyModel.object.get(id=1) # Re-get the object so we can > > > > > have the latest state > > > > > 6 object.value = object.value - 1 > > > > > 7 print object.value # returns 4 > > > > > 8 object.release_lock() > > > > > > If the above code fails to get the lock because another process is > > > > > running the code, it goes to sleep until the other process finishes. > > > > > The other process will also be decrementing object.value, so if we > > > > > have 2processesrunning the above script at the same time, I would > > > > > expect the later process to return line 7 as 3 andnot4. > > > > > Shouldn't some saving occur for that to be true? > > > > > regards > > > > Steve > > > > -- > > > > DjangoCon US 2010 September 7-9http://djangocon.us/ -- You received this message because you are subscribed to the Google Groups "Django users" group. To post to this group, send email to django-us...@googlegroups.com. To unsubscribe from this group, send email to django-users+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/django-users?hl=en.
