Need to refresh/F5 to show updates in 3.8.2
 

Hi - just cross posting here in case this is a server configuration problem ( original thread at http://www.vbulletin.com/forum/showthread.php?t=309938 ).

I've just installed a brand new install of vbulletin (3.8.2) and have a problem displaying updated content.

After doing something that adds new dynamic content (like adding a new post) and going back to a page that has already been viewed, the old page is loaded from cache and not the updated one. I can click around and constantly get the old content until I force the refresh of the page. I can replicate the problem myself and I have had about a dozen users of my site experiencing the same problem.

I can replicate the problem on my production web server as well as on my development server, both running on centos/apache2.2.3 and running php5.2.9 and 5.1.6 respectively. It makes no difference whether it's running with my plugins enabled (vbadvanced/passive video/usergroup as mod) or whether I disable them all (including setting DISABLE_HOOKS to true in config.php).

I've also check to ensure I can replicate using the standard theme, and in fact after selecting the standard theme I had to hit refresh to stop it rendering the old page/theme because of the problem. Have tried IE and Firefox and it happens in both, have ensured I've flushed cache/cookies/etc.

The only way I can stop the problem is to force "Add No-Cache HTTP Headers" to ON. This fixes the refresh problem, but increases the page serving going on and is contrary to almost every optimisation recommendation ever made for vbulletin.

I've seen a lot of threads describing this exact same problem with 3.8.2 but no solutions. Surely forcing no-cache on isn't seen as a reasonable solution.

Given that I can replicate this using different browsers, different physical machines, different versions of PHP, and using a brand new install of vbulletin with hooks turned off and the standard theme, it seems a problem that is definitely related to vbulletin.

Are there any experiences with this problem that people know relate to server configuration? I can replicate the problem on 2 different machines (one running prod from my host, one running under vmware in my own dev environment), so its not one box, but it could be a common configuration as they are both running the CentOS redhat derivative.

Any help much appreciated.