For a few months, while the front-end part of this site seemed to be functioning just fine, I have been having some significant issues on the backend. These issues would mostly commonly result in timeouts on Administrative functions. For instance, I would make a new post, and the process would time out. The post would go live, but some of the other related functions would not carry out and I’d have to do them manually. Or I’d go to approve a comment and would get a timeout. Things like that.
I was not overly concerned because the site was working fine for users. It was just the admin functions that seemed problematic. I spent several months trying a variety of things here and there, and a few days ago, it looks like I finally nailed the problem and corrected it.
One of the ways that you troubleshoot these types of issues in WordPress is to disable your plug-ins and then re-enable them, one at a time, to determine which, if any, might be the culprit. I did this, with mixed results. For instance, it seemed that WordPress JetPack caused some of the problem some of the time. I disabled it and things got a little better, but not entirely.
I remove all the plugins I did not need to keep the site functioning the way I wanted it, but still the problem remained. In fact, the only time it seemed to go away was when I purged the cache for the W3 Total Cache plugin. The problem would seem to go away for a little while, and then return. It got me thinking: could the caching plugin actually be the one that was causing trouble?
On Sunday afternoon, I disabled the plugin, noting on Twitter that people might see some strange behavior, just in case:
Doing some testing on the blog to track down a memory-hog plug-in. You _may_ see some strange behavior. Then again, what else is new?
— Jamie Todd Rubin (@jamietr) September 15, 2013
As soon as I disabled the W3 Total Cache plugin, the timeouts stopped cold. I was cautious, of course, thinking they might return in a few hours. But they didn’t. It has been three days since I disabled the plugin and all of the administrative functions are working normally once again, without timeouts, and without the need for me to jump through extra hoops to get posts out. It is a wonderful feeling, since this problem has been ongoing for more than 6 months, to finally have it solved.
The question remains: why does the W3 Total Cache cause problems? I suspect it is memory-related. The PHP processes being spawned from the site when W3 Total Cache was installed were in excess of 90 MB each! It could be that I misconfigured W3 Total Cache, and thus the problem was really my fault in the first place. Well, I don’t have the time to delve into it, but I don’t really need to either. The back-end is functioning normally again, and the front-end seemed unaffected by the removal of the plug-in.
All is well, and I am happy!
I gave up on W3TotalCache years ago; I used a few different caching programs for brief periods after that, but now I just use CloudFlare and that works fine (and better, really.) For one of my clients we set up Railgun and those sites blaze but the way I host my sites doesn’t allow me to use Railgun right now.