Irregular Traffic Log Entries

Are pages being cached and partially reloaded?

[ThirdWheel]

I am having some issues with my traffic logs. Note that I am not considering this an AdSense issue, simply that AdSense stats have alerted me to the fact I have a problem.

From early May I have been having wild numbers of AdSense impressions listed in my account, which are much greater than the traffic I receive. Every time this happens I go through the raw logs in the hope of finding the cause, but usually there is nothing there. Then on one day, when one channel claimed nearly 17000 impressions (instead of the normal <1000), the logs contained >16000 requests for an image file from one page within that channel. Requests were made at 1 second (or less) intervals, each receiving a 304 response. All requests from the same IP (residential) and with a generic UA. This IP made no other requests, just this one image (that precedes the AdSense code on the page).

In order to try and locate future sources, with the help of WW I added headers that should stop caching (although I am unclear if they are always honoured). I also used the old trick of adding a small piece of javascript that points to an external js file, each time generating a dynamic name by appending the js filename with the timecode, so that technically it cannot be cached.

However, these measures have made little difference. I am still getting large AdSense impressions, and on the odd occasion, an accompanying number of demads for an image file. If the image file is large then the response is usually 206. Each time logs a different IP and UA.

It would seem that my pages are being partially downloaded from a cache, and that only certain elements are requested, regardless of any changes I make to stop caching. However, I have never heard of that happening before, and am not even sure if that is possible. Thanks to WW I have countered AVG, but in my experience that doesn't execute javascript, so clearly something else is at work.

Does anyone have any ideas?

[eelixduppy]

For the image file, my guess would be that someone is using that image somewhere on the web and linking it to your site. Each time the page is requested where they used it, it requests the image from your server.

Is it always the same image? If so, how about changing the name of that image to something else.

[ThirdWheel]

Thanks for the response. I considered hotlinking, but for every occurence it is a different image. Also, the burst from almost nothing to 16000 requests coincides with an identical rise for the associated AS channel.

I have AS's "allowed sites" enabled - I've not heard of any problems with that feature, so presumably the code can only be run on my site.

I'll see if I can dig a little deeper, can't help thinking I must be missing a piece of the puzzle ...