when i last tested it it was with a very complex custom auth +
validators. there are probably ways to trim it down. one thing that i
did was only init mail and attach mail to auth if i was doing something
that required mail.
cfh
On 8/7/12 6:45 , Alexei Vinidiktov wrote:
The app I'm developing is highly dependent on auth.
Does it mean that any request that uses auth won't be faster than 200ms on
GAE?
On Tue, Aug 7, 2012 at 7:30 AM, howesc wrote:
are you using auth? auth takes around 200ms to load i think on GAE. i
avoid initializing auth except for the controllers that use it, as well as
having my model definitions in modules.
cfh
On Monday, August 6, 2012 3:57:51 PM UTC-7, Felipe Meirelles wrote:
Actualy Django uses the datastore too. I'm using the memcache for all
things its possible (it already reduce the old django version database
access on 90% in this new web2py version).
I think the overhead is realy from loading all models, grids and menus
I've defined for my app. I'll try a Model less aproch, but tying to right a
less complex code, using most of the web2py api as possible.
As soon as I finish, I'll post here (:
Thank you all.
On Monday, August 6, 2012 7:50:59 PM UTC-3, Massimo Di Pierro wrote:
You should check the {{=response.toolbar()}} because there may be more
DB io than you think in the models.
You should also try remove the setting of model attributes (requires=,
models=, ...) and move them in the controllers that need them.
Also mind that web2py on GAE has sessions turned on by default and they
are stored and retrieved from datastore.
I do not know how Django stores sessions on GAE by default.
massimo
On Monday, 6 August 2012 14:19:35 UTC-5, Felipe Meirelles wrote:
Without any model I have a huge drop on the cpu use, from around 300ms
to around 60ms. Still higher than with Django, but its acceptable by the
concept of the framework.
Ill make some debug on my models and update the topic.
On Monday, August 6, 2012 4:05:56 PM UTC-3, rochacbruno wrote:
What do you have in models?
All model files runs alphabetically for each request, so we need to
know what are you doing on model files.
Can you try to test with an empty brand new app, remove all files from
models folder and try your simple controller
import logging
def test():
logging.info("Just making a performace test")
Makes difference?
On Mon, Aug 6, 2012 at 3:59 PM, Felipe Meirelles <
[email protected]**> wrote:
Even the simplest controller give me a high cpu usage:
import logging
def test():
logging.info("Just making a performace test")
ms=559 cpu_ms=612 api_cpu_ms=245
On Monday, August 6, 2012 3:37:38 PM UTC-3, Felipe Meirelles wrote:
Just compiled the app and deployed again, seems to make no
diference. App engine uses the bytecode generated on web2py? Did I need to
change some config at app.yaml?
On Monday, August 6, 2012 3:21:29 PM UTC-3, Felipe Meirelles wrote:
Pre compiled by the admin interface? Not yet, is there some impove
with the precompiled code?
On Mon, Aug 6, 2012 at 3:16 PM, Derek <[email protected]> wrote:
Have you compiled your app?
On Monday, August 6, 2012 10:46:33 AM UTC-7, Felipe Meirelles
wrote:
Hi,
I'm using web2py in a project that stands on top of a REST api.
The project is write intensive (around 110k requests a day, with all the
requests making at least 2 updates to the datastore).
All my requests seems to uso at least 200 cpu_ms (even dummy
requests that only places a item on the queue to be processeced later). Is
this behaviour right? I used to get around 20 cpu_ms while using django
with the same application (I'm currently re-writing it on web2py).
Is there some way to optimize web2py for a appengine deploy (I
already followed all the documentation)
Thanks!
--
--
Att,
Felipe Meirelles.
--
--
--
