i see similar behavior on python2.5, i just did my test on 2.7 as i'm
trying to move my clients that way anyway.
i'm going to do some more digging this afternoon.
On 2/19/12 11:53 , Jonathan Lundell wrote:
On Feb 19, 2012, at 11:41 AM, Massimo Di Pierro wrote:
I am very puzzled by this. I was under he impression nothing persisted
on GAE except for stuff in datastore. web2py for GAE was built under
that assumption. It appears to be not true. If you can find anything
else to narrow down the problem, let us know.
That's not the case with GAE + 2.7 + multithreading; it appears to run as a
persistent process, with requests as threads.
On Feb 19, 11:57 am, howesc<how...@umich.edu> wrote:
so i uploaded welcome from 1.99.4 to GAE, running python2.7 multi-threaded,
and hit all the URLS in welcome that i coule (which is not too many), and
here is the play-by-play. The numbers on the left are instance memory size
in MB
45.3 - first page
45.7 - same page, diff URL
45.7 - refresh
45.7 - refresh
45.8 - login page
46.1 - refresh
46.1 - refresh
46.4 - submit invalid login
46.4 - refresh
46.4 - submit invalid login
46.4 - lost password
46.5 - submit invalid email
46.8 - refresh (re-post form)
46.8 - refresh (re-post form)
46.8 - submit invalid email
46.8 - back to login
46.8 - refresh
46.9 - register
46.9 - submit empty form
46.9 - submit, password too short
47.3 - complete register, redirect to home
47.3 - profile
47.4 - save profile (no edits)
47.4 - password
47.4 - submit empty form
47.7 - change password
47.7 - logout
47.7 - register
47.9 - register new user, redirect to home
47.9 - profile
48.1 - update profile
48.1 - password
48.2 - change password
48.3 - logout
so we ate up 3 MB by creating users logging in and logging out. i suspect
that if i did this for another handful of users i'd get to the 128MB cap
and have my instance killed. any tips on how i can inspect memory to see
what is being kept in between requests?
