Bug #1265

Caching breaks article editing due to InvalidAuthenticityToken error

Added by Joseph Chen 81 days ago. Updated 77 days ago.

Status :Assigned Start :09/11/2008
Priority :Urgent Due date :
Assigned to :neuro - % Done :

0%

Category :admin
Target version :5.1.4
Resolution :


Description

I am seeing an ActionController::InvalidAuthenticityToken in Admin/contentController#edit while editing an article with caching enabled in Development mode. Previously I had seen problems with caching in 5.0.3 and had disabled caching altogether. Now I would like to test the caching, but would like to do it in Development mode before pushing out to Production in case the caching is not working as expected. When I enable caching in development.rb and then try to edit an article, the InvalidAuthenticityToken error is raised. The stack trace shows that the problem comes from
action_controller/request_forgery_protection.rb:79:in `verify_authenticity_token'
Also the request parameters shows that an auth token is being sent
"authenticity_token"=>"c16566fd52b4f1f15bf83f010467d00c58dfafe6",

BTW, we are using Rails 2.0.2, and we have the Typo caching mode set to semi-static.

History

09/11/2008 06:42 PM - Joseph Chen

Also, we are using Typo 5.1.3 (I forgot to set the Target Version).

09/13/2008 03:08 PM - neuro -

  • Status changed from New to Assigned
  • Assigned to set to neuro -
  • Target version set to 5.1.4

OK, please tell me how you tried to edit it please, it's important for me. Was it from the admin, or by clicking form womewhere else ?

09/16/2008 07:22 AM - Joseph Chen

It looks like this may not be a bug after all but rather a problem on our end. For some reason, even with the caching disabled, our Typo is still throwing the InvalidAuthenticityToken error when editing an article using the Admin. However, this only occurs on the dev machine and not the production machine. When the protect_from_forgery line from application.rb is commented out, then we don't see the above errors, even with caching turned on.

It is now our conclusion that this must have been caused by some sort of problem specific to our environment, so you can go ahead and close this bug as being not a bug.

Also available in: Atom PDF