I was wondering if there was a way that I could make my red hat system look like windows xp or even something else if scanned from the Internet. I have searched the web & the only thing that I found that would let me do this was an outdated kernel patch. I thought this might be a somewhat good idea for my firewall so that if someone tries to hack it they would spend enough time trying to exploit the wrong type of system & finally give up. I understand the idea of security through obscurity & realize that this isn't a failsafe method, but I would like to add another layer of security to my linux box as I'm switching to dsl soon.

Thanks for any help or info

If your box has any externally accessible services then it is usually fairly trivial to come to the conclusion that it is not Windows. If it does not, then they cannot really do anything anyway - disregarding a vulnerability in Linux's TCP/IP stack. This really is not worth your time .

Thanks for the reply. I was curious about this & believe (hope) that my box is fairly secure as I don't have any outside services running & have used several tools to help lock everything down.

If you have no services running they can connect to then they can't tell what you are running anyway. There aren't a whole lot of remote exploits around for linux anyway. There were a few involving old versions of apache but if apache isn't running as root then they worst they could do is hose your websites. SSH < 2.0 is exploitable but very few people run that anymore, there has been some security holes in OpenSSH but many of them were found and fixed without a successful exploit ever produced (they were theoretical holes, or at least things the developers thought weren't great for security).

In general as long as you don't have insecure stuff listening on external ports, and you run things like web servers and ftp servers under a user other then root you are pretty safe. There are things you can do with IP tables to make your machine practically disappear that other posts in this forum discuss.

This is not necessarily true. A lot of system exploitation relies on the abundance of local privilege elevating vulnerabilities coupled with any usable account on the target system. You not only should be looking at running processes as users other than the superuser, but the amount of the interaction these non-privileged users can have with the environment. For instance, it is bad practice to run all non-privileged processes as the user `nobody' (once a common thing), as this allows a compromise on (for example) the web server to possibly interact with an internal process running as `nobody'. The main thing you will want to do is stop the user from gaining a shell prompt, for this you will want to look into things like chrooting, spending hours with DAC or minutes with MAC - none of these methods are foolproof, but they help.