> Hi, > > My models are relatively big in file size (> 200 kb in total). > > Is ALL this code being loaded into memory for each web request over and > over > again, and without reusing the same files (that do not change)?. > > I'm noticing a consistent increase in memory % in each of the uwsgi-python > processes/workers for simple web requests. > > This makes me think that web2py (or uwsgi-python) is loading the code for > all models into memory for each web request without any caching / reusing, > and without releasing any memory from previous requests, but only > consuming > more and more memory each time. > > This means that I run out of memory after a very few requests ! > > Am I not correct above?, am I doing something wrong?, is this actually > expected?, does this occur regardless of the web server or wsgi > interface?. > > On a related note, should I move everything to modules (instead of > models)?, > will this fix the memory issues?. > > How can I solve this show-stopper problem?. > >
Honestly, (without knowing your app) i do no think this is a problem related to web2py (and for sure it is not a uWSGI as its policy is: NO LEAK). Probably something in your app is leaking memory, but in the mean time you can "patch" the problem with something like --reload-on-rss <n> where n is the number of megs after which a worker is restarted (freeing its memory). I think 200 (megs) should be enough to maintain your webserver healthy. -- Roberto De Ioris http://unbit.it
