Falcon Engine safe on multisite?

No we have not tested this as part of our testing suite. Do you have a test server you can run this on? I’d be happy to work with you if you find any issues.

Regards,

Mark.

Hi Mark!

I’ve been hoping to find the time to do that testing, but things have been pretty crazy lately. Since it’s been a while since Falcon Engine came out, I wondered if anyone had found the time to test it on multi site.

Can you briefly comment on how you expect Falcon Engine would deal with rewrite rules in a multisite situation?

Hopefully, I’ll get a test network up soon.

Bet can you post your current rules without posting any private/sensitive info?

Thanks.

Hi Mark!

It’s not the current .htaccess rules that might present a conflict. It’s the way WordPress multisite uses domain mapping.

One of my multi-sites is bhmbizsites.com – that’s the main site of the multisite. New sites are created within the network using a subdomain structure, so those sites might be: biz1.bhmbizsites.com and biz2.bhmbizsites.com.

But each of these subsites is a client site with a completely separate domain mapped to it. The domain mapping redirects the subdomain URLs to the separate domain. So the page “biz1.bhmbizsites.com/about” becomes “spiffydomain.com/about”. I’m pretty sure that this is a dynamic mapping process – if I turn off domain mapping for a particular subsite, it just goes back to the regular subdomain URLs.

So if I understand what WordFence is doing in Falcon Engine correctly – flattening the directory structure – what I’m not sure about is whether Falcon Engine would do this with the subdomain URLs or with the mapped URLs. And if it’s with the subdomain URLs, I’m guessing the speed increase would be offset by having to redirect to the mapped domain URL?

Anything you can tell me about how Falcon Engine might handle this?

We use the HTTP_HOST variable in .htaccess:

RewriteCond “%{DOCUMENT_ROOT}{$pathPrefix}/wp-content/wfcache/%{HTTP_HOST}_%1/%2~%3~%4~%5~%6_wfcache%{ENV:WRDFNC_HTTPS}.html%{ENV:WRDFNC_ENC}” -f

So the file we cache is stored in a path that is made up of the site hostname. As long as your rewrite rules appear before ours and the HTTP_HOST is set correctly, it should work.

If the path is not stored correctly, then your site won’t crash. Wordfence simply won’t serve up the cached file.

So you might try this. Then go to the /wp-content/wfcache directory and see which hostnames are being used to store cached files.

Also enable the option on the Performance Setup page to add an HTML comment to all cached pages. Then hit your site with a new incognito window in your browser and check the footer of the HTML source to see if the page was served from the cache.

Regards,

Mark.

I finally had time to do a little bit of testing with Falcon Engine on a test multi-site network.

I network activated Wordfence, then in the network admin dashboard, enabled Falcon Engine just like I would for a single install. Everything seemed to work okay, and my site load time went from 1.62s to 1.56s — not as much help as I might have expected, but the sites do seem a little faster to me, so maybe I didn’t do the loadtime tests right. (I’m new to this)

I tried network disabling Wordfence, then it enabling it in one sub-site of the network. The Wordfence menu did not appear in the sidebar in this instance.I guess this is because Wordfence is intended to be network activated in a multisite environment, right?

Your WF rewrite rules come before any of mine in the .htaccess. Don’t know if that will be an issue for some folks…

Finally, I looked in the /wp-content/wfcache directory, but I’m not seeing any cached files there. Does that mean that no caching is actually happening? (I hunted around in my sites, and there are no cached files in that directory for a couple of my single-site installs running WF-FE– is that normal?)

Hello Bet, I too just network enabled Falcon on a multi-site set-up (much kudos to WF for forcing me to backup my .htaccess file). It went smoothly. To address your question, I don’t think WF is intended as multi-site product only. That being said, I cannot get it to work as a sub-site only activation. Having switched WF back to network enabled I can see only 5% of my sites listed in /wp-content/wfcache & each site’s folder currently stores only one random page per site. Maybe it takes some time for the cache to be accumulated? (However, read the comment here about another user seeing only their home page cached). Strangely, having done a test using P3 – Plugin Performance Profiler (a site I tested 12 hours ago) has seen a 200% increase in PHP ticks, 300% increase in SQL queries, 30% increase in core load time & a 20% increase in site load time. None of these are good. I cannot tell if this is WF related, so I need to do some more investigating.

Just switched off Falcon & ran the P3 test again – no significant change in load times. Therefore, I should reasonably conclude that the overnight increase in ticks, queries & load times had nothing to do with the activation of Falcon … although I’m stumped as to what is happening.

Okay. Maybe it just takes a while for the cache to build – now I’m seeing cached pages (or at least directories for them) in wp-content/wfcache.

More to the point of the OP however, those cached pages seem to be using the mapped domains, and not the subdomain pages. For example, I see a directory for a page: acwprogram.com_staff, when the orginal, un-domain-mapped url for that page is: test3.bethannon.net/staff.

The site does seem more zippy, but the overall pingdom load speeds didn’t improve dramatically. (Maybe it’s a placebo effect?)

Mark, it seems like my initial tests on multisite presented no issues. Anything else I should look for or be aware of before I deploy this on a production multisite network?