(This isn't strictly a search engine question, but since SEM's study their server logs more carefully than anyone on earth, I think someone here will know the answer.)

When I look at my server logs, I notice a very weird thing: lots and lots of requests for this non-existent URL:

www.gleeson.us/NONE

I mean lots. The URL '/NONE' gets hundreds of hits a day. Not just on gleeson.us either, but on other sites I maintain too. And it's always just like that, 'NONE' all caps, no file extension. My first thought was, it must be a hacker or DNS attack or something sinister. (But then, that's always my first thought. Hell, I remember the first time I saw a bunch of 'favicon.ico' requests, I assumed it was a virus, when it was just nice people bookmarking my site.)

Further study of the logs indicates that it is almost certainly not a hack, because the requests are always from real visitors to the site, and the referring URLs are always the real pages on my site. So, something on my pages is causing browsers to request 'NONE'. But danged if I can figure out what it could be.

My next thought was that it's Javascript's fault. On my home page I've got a script that preloads a few images for rollovers. It seemed reasonable that maybe on some machines the URL for the preload image somehow resolved to 'NONE' and the browser dutifully attempted to preload 'NONE'. But that ain't it either. I get the 'NONE' hits even when I remove all the Javascript.

So it's still a mystery. I haven't solved it yet, but I have put an empty text file named 'NONE.php' in my home directory. So now at least it's not swamping my error logs with 404 errors. But has anyone else run across this? Where the heck is this coming from?

> the requests are always from real visitors

How did you determine that? Looking at the referer is not enough as that is pretty easy to set to anything for a http request. Basically, I can design a small agent that goes to any of your pages and send a referrer string that tells you I came from another page on your site without actually being true.

I would look at the IP and agent name too. Are they the same for every request? Or, are the IPs from the same C-class?

It's sounds to me that (as you suggest) you either have some navigation problem or that someone is testing (bad) spiders at your site. In the last case you should be able to narrow down the IP(s) and just block them.

It's sounds to me that (as you suggest) you either have some navigation problem or that someone is testing (bad) spiders at your site. In the last case you should be able to narrow down the IP(s) and just block them.

I agree with this statement. Please let us know if you find any more information you can share.