SELinux and Apache Treason Uncloaked error!
 

Hi! I was wondering if someone could give me some insight on if setting up Apache within SELinux would stop all the treason uncloaked attempts on my server. Would SELinux essentially jail/sandbox the Apache process so that anything that accesses the daemon can't break out of it? If so wouldn't this not essentially deny all the Treason uncloaked attempts since they are not in the httpd_t context?

Ex:

TCP: Treason uncloaked! Peer 6.6.6.37:38/80 shrinks window 2718851748:2718853128. Repaired.


Thanks

I had to look up what that meant:
http://linux.derkeiler.com/Mailing-L.../msg03750.html

Thanks jschiwal, this is a strange bug, the only web servers that we see these errors on are the ones facing the Internet. We have two web servers facing the web for redundancy and they both are seeing the same "Treason Uncloaked" errors. All of our internal web servers do not have these errors. Which lead me to the thought of it being a breach/denial of service attempt.

It may be a type of denial of service attack or an attempt to crash the server. I guess the gist of it is that the client reduces the size of the window against protocol. Is this coming from a single IP or a number of them? You could blacklist that IP if it is.

The attacks come from several IP's I just listed one for an example. I'll look into scripting out a IPTables rule to ban these IP's. Perhaps a daily/hourly scan of the /var/log/dmesg file for any "Treason Uncloaked" attempts and deny those IP addresses.

You might consider subscribing to the kernel mailing list. From your message it looks like the kernel is doing something to protect itself, but I'm not certain. The links I found were a couple years old. There may be more work done in this area as well.

This more recent link indicates that the problem (for that user) wasn't with an attacker at all:
http://linux.derkeiler.com/Mailing-L.../msg04353.html
Check if your kernel precedes the patch. If your kernel is more recent then you probably need to assume it is a dos attack.

We've had this question before (and you should search LQ keysorsoze, because you've asked this before as well) and quite some time ago, proof this has been in the kernel for a long time. The message (of the informational level) is proof of the robustness of the Linux kernel: it determined there was a problem, corrected it automagically, all by itself. Refs:
http://www.linuxquestions.org/questi...tempts-572874/
http://www.linuxquestions.org/questi...ssages-526332/
http://www.linuxquestions.org/questi...loaked-112531/
http://www.linuxquestions.org/questi...loaked-127984/