I think there's a general way to clear out the SQLObject cache better than has been done in TG. I think it was added to a more recent version of SO, and I must admit to not being a regular SO user in the last year or so.
If somebody using SO could take a look at this ticket, I think it might have suggestions that lead to a resolution of this issue. http://trac.turbogears.org/ticket/85 Of course, I use SA and it's been resolved there... But if there's a good SO solution, we should try to get that figured out. --Mark Ramm On Mon, Mar 31, 2008 at 9:02 PM, GSP <[EMAIL PROTECTED]> wrote: > > Hello Graham, > > Based on the provided info and configuration, were you able to > identify a potential cause of the database transaction issues that I > observed.? > My work around was admittedly less than ideal and I am still curious > to know why the problem only occurs when using mod_wsgi. > > Thanks! > > > > On Mar 29, 1:18 am, vandevel <[EMAIL PROTECTED]> wrote: > > Thanks for the prompt response. I am aware of the problems with using > > SQLObject in a multi-process environment > > so mod_wsgi was configured to run in daemon mode. I'm not entirely > > familiar with apache configuration so I am guessing the daemon > > process is running as Apache user since I did not explicitly set that > > configuration option. > > > > I did discover an issue in the my code which lead to the error when > > mod_wsgi is enabled. I loaded a dictionary of SQLObject instances at > > module scope in a controller module in order to avoid hitting the > > database with each request. This dictionary of course persists across > > transactions so perhaps that indicates something. To resolve the > > issue, I simply copied the data into > > wrapper objects in order to disconnect the data from the SQLObject > > machinery. That did the trick, however, I am a curious to know if > > there is a more intelligent solution. It is worth pointing out that I > > did set cacheValues=False in the SQLObject class definition and that > > did not solve the problem. > > > > The apache configuration is as follows(purged of sensitive details of > > course): > > > > WSGIDaemonProcess site-1 threads=25 > > WSGIProcessGroup site-1 > > > > Alias /static /project_dir/static > > > > WSGIScriptAlias / /project_dir/project.wsgi > > > > NameVirtualHost *:80 > > <VirtualHost *:80> > > ServerAdmin [EMAIL PROTECTED] > > > > DocumentRoot /project_dir > > > > # Possible values include: debug, info, notice, warn, error, > > crit, > > # alert, emerg. > > LogLevel warn > > > > CustomLog /var/log/apache2/access.log combined > > ServerSignature On > > > > <Directory /project_dir> > > Order deny,allow > > Allow from all > > </Directory> > > </VirtualHost> > > > > Thanks for your help! > > > > On Mar 28, 10:48 pm, Graham Dumpleton <[EMAIL PROTECTED]> > > wrote: > > > > > On Mar 29, 10:17 am, vandevel <[EMAIL PROTECTED]> wrote: > > > > > > I am wondering if anyone has experienced database transaction problems > > > > usingmod_wsgi1.3. > > > > My turbogears project is based on TG 1.0.4.4 using SQLObject and I am > > > > observing this error when executing some controller methods: > > > > > > AssertionError: This transaction has already gone through ROLLBACK; > > > > begin another transaction. > > > > > > This only occurs when running the app behindmod_wsgi, however it is > > > > not a problem when running the app straight from cherrypy. > Clearly,mod_wsgiis doing something which causes the app to behave in this > > > > manner. > > > > > More information required. > > > > > Are you running your application in embedded mode of mod_wsgi or in > > > daemon mode? If in embedded mode, you did read about the problems with > > > SQLObject and multiprocess web servers in: > > > > > http://code.google.com/p/modwsgi/wiki/IntegrationWithTurboGears > > > > > If running in daemon mode, are you running it as Apache user, or did > > > you configure daemon process to run as you? Also, if in daemon mode, > > > how many processes did you set for the daemon process group. Like > > > above, if more than one and using multiprocess server, there can be > > > issues, although it shouldn't be what you are seeing. > > > > > Anyway, post what the mod_wsgi configuration you are using is from > > > your Apache configuration. > > > > > Grahamth > > > -- Mark Ramm-Christensen email: mark at compoundthinking dot com blog: www.compoundthinking.com/blog --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "TurboGears" group. To post to this group, send email to [email protected] To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/turbogears?hl=en -~----------~----~----~----~------~----~------~--~---
