Hi,
I have some photos posted in facebook.com
Into each caption I've added a link to my server to let friends download them in larger sizes.
tail -f access_log only displays some of those accesses, I don't understand why.
If I reload large image page an input is recorded and displayed from access_log
What can be happend?
Thanks.
P.S. (Safari OS 10.6.3)

@marciano: Presumably you're running Apache web server on Ubuntu? And your friends are using the Safari web browser on OSX? (I'm taking guesses based on the limited info you've provided.)

This is going to be difficult to troubleshoot unless you can personally recreate the problem. Can you? Without that, you're relying on third-party info and may or may not be able to set up credible test cases.

On a WAG, there may be some http caching server sitting at any number of points between your Ubuntu host and your friends' web browsers. That's one possible explanation for why you're seeing only certain GET requests on your end.

Hello Anomie,
thanks for your interest.
My server is running Apache 2.2 on CentOS 5.3 on a remote dedicated server.
My test were performed from both Kubuntu and OSX.
I repeated the test (following facebook's links) on a pair of photos.
One them never left a log entry while the other always did.
All other sort of access were logged: dir access, page not found access, refreshes (in all cases)
I thought something about caching but make no sense for me such kind of file discrimination.
There were not 'friends', just me, this was a hidding 'business page'.
Logs only contained my IP.
Links in facebook are place as captions for each photo. They are also displayed below correspondent thumbnail in my 'wall' place.
Thank you

Attached Thumbnails

 

That's odd, but I still suspect something external to your CentOS host is the cause. To help prove the point, monitor all tcp 80 (http) packets on your CentOS host, a la:
Now experiment with accessing the mysterious photo file that is not generating requests in your Apache logs. Did you see any tcp requests to port 80?

Another thing: you said you tested from a Kubuntu box and from an OSX box. Out of curiosity, are those both using the same ISP?

tcpdump port 80 floods large amount of data. grep?

No, don't use grep on that. You can reduce noise if you're able to pinpoint the host that you expect the request to be coming from (i.e. the workstation you're using to test).

(Use your real IP, obviously.)

Well, behavior is not random today.
Yesterday, I had that problem with more than one file.
I remember I had followed the same link several times for certain file and always a log entry was recorded, not the same behavior as other files.
Today this is what is happening.
First time I follow facebook's link a log entry was recorded and tcp data was dumped.
Second+ time, no log entry, no data dumped.
Same for 5 or 6 files I've tested.
Thank you

Then it's highly likely that the photos are being cached at some point between your workstation and your Apache host. Could be as simple as your client web browser doing the caching. (i.e. Clear your browser cache and try the request again.) Could even be your ISP doing the caching...