Linux Process Hunting
Now that we have detected there is a malicious process running on one of our hosts, let's hop onto the box and start looking through the processes.
Let's start looking at all the ways we can detect this and find out more information, without having to use any networking commands.
In this scenario, let's say our network analysts reported that weird traffic makes them believe the system has been compromised in the last week.
Running the "history" command and seeing what pops up is a solid first move once you get on any box. However, because the attacker cleared out their history, we shouldn't see anything there, which should be an alert to us itself!
find command we can check for files made in the last week with
* find / -mtime -o -ctime -7. To get even more specific, like if we thought we were recently compromised, we can make that 7 a 1. You should see a new file or two. Check the user of that file with
ls -l. If that user seems suspicious, let's check out their most recent commands run.
su over to the new user you discovered and check their history, maybe the attacker forgot to hide that as well.
Now that we know there is definitely a new user added and what looks like crontabs, let's start by looking for common persistence mechanisms from the command line. First, we can check cronjobs using the
crontab command, or
crontab -u to check other users.
We can also look for SETUID or SETGID binaries with
find / -perm -4000 -user root -type f and
find / -perm -6000 -type f. Modify those commands as you see fit.
Another great tool to use is going through the logs stored in /var/log. In the real world, many organizations collect all of these logs from their computers and store them in a centralized location where analysts can look through them to find evil, or after evil is detected, determine what else was done. With that said, just because an organization is collecting does not mean they are detecting, just like with packet captures.
There are a ton of different logs in there, but the most useful is named "syslog". You can look through that and find information about what has happened, especially if you search through it to find where the attacker began running commands.
Now that we've discovered that there is definitely malware on the box, in the real world an analyst would log onto the potentially compromised host and begin doing incident response. For us though, let's wrap this up.
For other commands useful to look for malicious processes check out this cheatsheet
Linux Rootkit Hunting
To go one step further, if you fear that you might have a rootkit, installing and running the
chrootkit tools is an excellent idea. These tools attempt to look for signatures of known rootkits or discrepancies in the output of commands which might indicate the presence of something modifying the operating system.... with that said, many rootkits have specific anti-detection defenses built-in, so it is a cat and mouse game.
- Write up a minimal report on the malicious actions taken by the attacker, as well as information about the backdoor. Include information from the command outputs to provide detail.
Congratulations on your first successful hunt! If you are looking to create a nice addition to your professional portfolio, I recommend combining your last two assignments, making it look pretty, and posting it somewhere. People will be very interested in it.