Jim Gallacher wrote:
Gregory (Grisha) Trubetskoy wrote:
Is this reproducable? With a 1 in 31 chance it should be pretty easy
to reproduce... Just want to make sure that there isn't some
undrelying reason why this was done this way.
The supposition that there is a deadlock is untested and simply based on
inspecting the code. dharana was asking about using psp outside of
apache, and when I took a look at psp.py I noticed the potential
deadlock. I'll give it a whirl and see if I can lock up my server. :)
Confirmed and reproducable.
Using mpm-worker with 2 servers started, deadlock occurs within 10 to 50
requests based on 5 test runs
ab -n 100 http://localhost/mp/mptest.py.
Bad handler - deadlock
----------------------
def handler(def):
req.content_type = 'text/plain'
sess = Session.FileSession(req)
template = psp.PSP(req, filename='/var/www/mp/template.psp')
template.run({'what':'world'})
return apache.OK
Good handler - no deadlock when session is unlocked
---------------------------------------------------
def handler(def):
req.content_type = 'text/plain'
sess = Session.FileSession(req)
sess.unlock()
template = psp.PSP(req, filename='/var/www/mp/template.psp')
template.run({'what':'world'})
return apache.OK
template.psp
------------
psp test
<%= what %>
End test
Jim
