you cannot use memcache to reliabliy increment the counter - it's just a cache and thus there's no such thing like a transaction (AFAIK) the memcache only makes sure, that a single put and get are atomic
the problem arises in this case: 2 users call the same servelt, but these 2 calls will not be processed on the same machine, but on 2 different machines. timeline: * request A calls cache.getKey() and the value is e.g. 10 * request B calls cache.getKey() and the value is still e.g. 10 * request A the next instruction is to increment the counter to 11 (this now is the value of the local variable on machine A) * request B the next instruction is to increment the counter to 11 (this now is the value of the local variable on machine B) * after that both requests will call cache.put() with the value of 11 (the sequence does not matter) and now you've lost one count in fact it does not really matter if that happens on 2 machines or 2 processes or threads - the problem is always the same if this is just a hitcounter, then it's may be ok to use this, because it does not really matter, if you sometimes lose a count. on the other hand, if you use the datastore and make these changes in a transaction, the datastore will make sure, that all operations in this transaction are executed atomicly: so: it does not matter, if A or B are processed first: each transaction will get the value, increment it and save it the datastore makes sure, that no other transaction can change the value during this operation hope this makes it clear (and I also hope it's correct) On Oct 28, 2:39 am, dburns <[email protected]> wrote: > I'm trying to make sure I understand issues surrounding multiple > concurrent users. Take the example > fromhttp://code.google.com/appengine/docs/python/datastore/transactions.h... > > def increment_counter(key, amount): > obj = db.get(key) > obj.counter += amount > obj.put() > > The text makes it clear that a naive implementation like that is > susceptible to a collision between two concurrent users (meaning two > separate instances of the python script running at the same time). > The second put() clobbers the results of the first one. > > I'm unclear on how memcache fits in. If those two concurrent > instances both instead call cached_obj=memcache.get("my_key") and get > a result back from some earlier memcache.add("my_key", my_object), are > they susceptible to the same sort of collision if each instance does > something like cached_obj.counter += amount? > > I kind of thought memcache worked seamlessly across instances, but I > don't see how in this case. > > Thanks for any clarification you can provide. --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "Google App Engine" group. To post to this group, send email to [email protected] To unsubscribe from this group, send email to [email protected] For more options, visit this group at http://groups.google.com/group/google-appengine?hl=en -~----------~----~----~----~------~----~------~--~---
