> Based on comments in a separate thread, GeoDango may have issues with > multithreaded configuration. Which Apache MPM was being used for each > Apache version, prefork or worker? The worker MPM uses threads and so > that could be the culprit.
Yes -- the libraries GeoDjango uses, GEOS and GDAL, are not thread safe. Thus, it is highly recommended to use the prefork version of Apache. > In a Python Shell, I am able to perform queries on my spatial models > successfully. In a web browser, I get a plain white screen if any > code is executed that works with spatially enabled models. I created an Ubuntu 8.10 VM to try and test out this problem. However, the only way that I could reproduce your exact problem (segfault upon HTTP request) is when using the mpm-worker (threaded) Apache. When using the prefork, however, I was able to get both mod_python and mod_wsgi to work with a simple demonstration app [1]. I think mpm-worker is the default when you do `apt-get install apache2`, so make sure you have it removed, e.g., `apt-get remove apache2-mpm-worker; apt-get install apache2-mpm-prefork`. BUT, I found another issue -- while I could login to and browse the admin interface, whenever I tried to view a geographic model an exception would get raised (but no segfault) deep in the admin widgets [2], crashing the app. While this crash occurs in 8.10 it does _not_ happen in my 8.04 VM. Thus, this leads me to believe that it may be a manifestation of the same troubles you're experiencing. Moreover, this admin crash happens with _both_ mod_python and mod_wsgi in 8.10 (mod_wsgi configured with `threads=1`). Needless to say, this behavior has me perplexed at the moment, and due to my finals I'm not going to have a lot of time to dig in deeper until next month. Perhaps there's a clash of the libraries that are linked to Apache and the ones used by the packaged versions of GEOS/ GDAL, or maybe it's caused by Ubuntu's AppArmor (confined to just CUPS in 8.04) -- but these are just potential possibilities. > Justin was good enough to run a test case I created and was unable to > duplicate the error. I'm wondering now about the exact versions of > Apache and other components that might be causing the problem. I still have the CentOS 5.2 VM, and I'll test it out again, but I'm not sure yet that this could be the same issue as you reported in October. It's a different distribution that used significantly older versions (that also worked in my tests). [1] http://geodjango.org/hg/world [2] http://code.djangoproject.com/browser/django/trunk/django/contrib/gis/admin/widgets.py#L43 --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "Django users" group. To post to this group, send email to django-users@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/django-users?hl=en -~----------~----~----~----~------~----~------~--~---
