Skip to main content
Mallory
MediumPublic exploitVulnerability

CVE-2021-44832: Remote Code Execution via JDBC Appender in Apache Log4j2

Published Dec 28, 2021 · Updated 13 days ago
Trending this weekNot trendingOutside this week’s top 100
Mention activity · 30 daysNo mention activity in the last 30 days.
CVSS6.6Medium · v3.1
EPSS97.9%100th percentile
CISA KEVNoNot listed
In the wildReported3 exploitation records in Mallory
Vendors
Cisco SystemsOracle CorporationDebian ProjectFedora ProjectApache Software Foundation
IdentifiersCVE-2021-44832CWE-20 · Improper Input Validation

CVE-2021-44832 is a configuration-dependent remote code execution vulnerability in Apache Log4j2's JDBC Appender. It affects versions 2.0-beta7 through 2.17.0, excluding fixed releases 2.3.2 and 2.12.4. An attacker with permission to modify the logging configuration can configure a JDBC Appender whose data source references a JNDI LDAP URI pointing to an attacker-controlled server. Insufficient restriction of JNDI data source protocols can then permit execution of attacker-controlled code. Unlike the original Log4Shell vulnerability, exploitation requires control over logging configuration rather than merely attacker-controlled logged messages.

Do you run this?See affected versions in your environment and whether adversaries are exploiting this vulnerability against assets like yours.
Check your exposure
Analyst brief

Impact, mitigation & remediation

What it means and what to do now: patch path, interim mitigations, and the assume-compromise checklist.

Impact

What an attacker gets, and what they’ve been doing with it.

Successful exploitation permits arbitrary code execution in the affected application's execution context, potentially compromising confidentiality, integrity, and availability. The resulting access is bounded by the application's permissions; administrative or system-level privileges are not inherently granted by this vulnerability.

Mitigation

If you can’t patch tonight, do this now.

Restrict logging-configuration modification to trusted administrators and prevent unauthorized creation or replacement of configuration files in application directories. Where feasible, disable the JDBC Appender or remove configurations using remote JNDI data sources. Restrict outbound LDAP connectivity to reduce access to attacker-controlled servers. Message-lookup disabling options intended for other Log4j vulnerabilities do not address this configuration-based attack path.

Remediation

Patch, then assume compromise.

Upgrade Log4j Core to 2.17.1 for Java 8 or later, 2.12.4 for Java 7, or 2.3.2 for Java 6, or an appropriate later fixed release. These fixes restrict JNDI data source names to the java protocol. Update bundled copies through the affected application's vendor-supported patches. If Log4j API is used, keep it aligned with the Log4j Core version for compatibility.

Public exploits

Exploits

2 exploits validated by Mallory's analysis. Hidden: 4 rejected by Mallory's analysis as not a working exploit.

View more in app →
log4shell-corazaThis repository is a defensive Log4Shell lab rather than a standalone offensive exploit drop, but it does contain working exploit components. The structure combines: (1) a vulnerable target container (`log4shell-app`, based on a public Spring Boot Log4j 2.14.1 image), (2) an nginx reverse proxy with the Coraza WAF and OWASP CRS plus custom Log4Shell rules, (3) an attacker container that builds and packages multiple JNDI exploitation toolkits, and (4) Suricata/Wazuh monitoring for network and SIEM visibility. The main exploit capability lives under `attacker/`. `attacker/exploit-classes/Exploit.java` is a malicious Java class intended to be loaded through a JNDI/LDAP reference. Its static initializer executes immediately on class load and performs proof-of-exploitation actions: OS-specific command execution to write a marker file, creation of `/tmp/log4shell-exploited.txt`, and an optional reverse shell routine that connects to `172.20.0.2:4444`. There is also a stubbed environment-information collection method. This makes the repo more than a detector: it includes an actual RCE payload class. `attacker/scripts/start-ldap-server.sh` operationalizes exploitation by launching one of three JNDI infrastructures: marshalsec, JNDI-Exploit-Kit, or rogue-jndi. These are configured to serve the malicious class or execute simple commands/callbacks. `attacker/scripts/test-payloads.sh` is a payload spray/test harness that sends many Log4Shell payload variants to the nginx proxy using common injection points such as `User-Agent` and `X-Forwarded-For`. It includes LDAP, RMI, DNS, nested-expression, env/sys fallback, Unicode, whitespace, and colon-dash obfuscation variants. This script is primarily for validation of WAF/IDS coverage, but it also demonstrates realistic attack strings. The rest of the repository is focused on detection and containment. `coraza/config/log4shell-rules.conf` defines custom Coraza rules for standard and obfuscated JNDI patterns, alternate protocols, JSON/XML body inspection, and anomaly-score blocking. `nginx/` contains a multi-stage Docker build that compiles libcoraza and the coraza-nginx module, installs OWASP CRS, and exposes HTTP/HTTPS reverse proxying to the vulnerable app. `suricata/rules/log4shell.rules` adds network signatures for inbound JNDI payloads and outbound LDAP/RMI/DNS/HTTP callbacks that would indicate successful exploitation. Overall purpose: provide an isolated lab to demonstrate the full Log4Shell attack chain and, more importantly, how nginx+Coraza and Suricata can detect/block it. Because the repository includes a real malicious Java payload and scripts to stand up LDAP/JNDI exploit servers, it should be classified as a valid exploit-capable lab with operational payloads, not merely documentation or a pure detection script.tieupham267 · Disclosed Apr 28, 2026 · markdown · dockerfile · web · networkOperationallog4shell-chainThis repository contains standalone, effect-only proof-of-concept reproducers for the Log4j vulnerability family, not a working RCE delivery framework. Its 32 files include three Java main classes, one Python callback recorder, eight Dockerfiles, two GitHub Actions workflows, three case manifests, three Maven build files, four Log4j XML configurations, documentation and ignore rules. The code-file count includes the four application sources, eight Dockerfiles and two executable workflows; Bash occurs inside workflow steps. The original repository URL, analyzed ref and archive size were not supplied, so their metadata fields use empty/zero placeholders. Case A compares Log4j 2.14.1, 2.15.0 and 2.16.0 using logged LDAP JNDI strings, with a non-default ThreadContext layout intended to demonstrate CVE-2021-45046. Case B compares 2.16.0 and 2.17.0 using a recursive context value and stack-exhaustion canary. Case C compares 2.17.0 and 2.17.1 using an attacker-controlled JDBC appender configuration. These versions include negative controls and should not all be interpreted as vulnerable to every listed CVE. The Java drivers exercise their own bundled dependencies rather than attacking a supplied remote application endpoint. The Python collaborator listens on TCP 1389 by default, receives up to 4096 bytes with a two-second timeout, closes each connection without replying, and records peer address, timestamp and a 256-byte preview. It serves neither Java classes nor LDAP referrals. Initial LDAP traffic may be a bind request without the token-bearing search DN, so a connection record does not guarantee token correlation or successful RCE. Important correctness limitations prevent accepting the manifests' expected 'proven' verdicts as evidence. Case B uses a plain message containing ${ctx:x} and a %msg layout on Log4j 2.16.0, where message lookups are removed; the supplied configuration therefore lacks an evident path to the claimed recursion. Its external watchdog is also not supplied. Case A's exact 2.15.0 bypass and layout evaluation timing remain unvalidated. Case C's JDBC/JNDI initialization, version-specific enablement and dependency wiring also need validation. A dark callback can reflect configuration failure rather than a successful security fix. NOTES.md and the README explicitly state that builds and live trials remain outstanding, despite other prose describing platform-generated proofs. The cases/*.yaml files are declarative inputs for a separate Tegron platform service; no local prove/assert CLI, proof artifacts or platform implementation is included. prove.yml builds images and prints expected chains but does not execute the trials, and its build steps tolerate failure. A separate Ferralon workflow invokes an external pinned scanner with console linking enabled and explicitly configures /runs and /ingest API destinations. Those destinations are CI integration endpoints, not exploit infrastructure; comments claiming that the workflow names no Ferralon host contradict its actual inputs. No reverse shell, persistence, credential theft or malicious code-loading payload is present.ferralon-demo · Disclosed Jun 23, 2026 · java · python · network · local · filePoC
Exposure surface

Affected products & vendors

Products and vendors Mallory has correlated with this vulnerability. Open in Mallory to drill down to specific CPE configurations and version ranges.

VendorProductType
Apache Software FoundationLog4japplication
Cisco SystemsCloudcenterapplication
Debian ProjectDebian Linuxoperating_system
Fedora ProjectFedoraoperating_system
Oracle CorporationCommunications Brm - Elastic Charging Engineapplication
Oracle CorporationCommunications Diameter Signaling Routerapplication
Oracle CorporationCommunications Interactive Session Recorderapplication
Oracle CorporationCommunications Offline Mediation Controllerapplication
Oracle CorporationFlexcube Private Bankingapplication
Oracle CorporationHealth Sciences Data Management Workbenchapplication
Oracle CorporationOracle Weblogic Serverapplication
Oracle CorporationPolicy Automationapplication
Oracle CorporationPolicy Automation For Mobile Devicesapplication
Oracle CorporationPrimavera Gatewayapplication
Oracle CorporationPrimavera P6 Enterprise Project Portfolio Managementapplication
Oracle CorporationPrimavera Unifierapplication
Oracle CorporationProduct Lifecycle Analyticsapplication
Oracle CorporationRetail Assortment Planningapplication
Oracle CorporationRetail Fiscal Managementapplication
Oracle CorporationRetail Order Brokerapplication
Oracle CorporationRetail Xstore Point Of Serviceapplication
Oracle CorporationSiebel Ui Frameworkapplication
FAQ

Frequently asked questions

Is CVE-2021-44832 exploited in the wild?
Yes. As of Nov 4, 2026, Mallory has reports of exploitation in the wild.
Is there a public exploit for CVE-2021-44832?
Yes. Mallory tracks 2 public exploits or proofs of concept for CVE-2021-44832, the earliest disclosed Apr 28, 2026.
Is there a patch for CVE-2021-44832?
Upgrade Log4j Core to 2.17.1 for Java 8 or later, 2.12.4 for Java 7, or 2.3.2 for Java 6, or an appropriate later fixed release. These fixes restrict JNDI data source names to the java protocol.
What products are affected by CVE-2021-44832?
Apache Software Foundation Log4j, Cisco Systems Cloudcenter, Oracle Corporation Communications Brm - Elastic Charging Engine, Oracle Corporation Communications Diameter Signaling Router, Oracle Corporation Communications Interactive Session Recorder, and at least 17 more.
How severe is CVE-2021-44832?
CVE-2021-44832 has a CVSS v3.1 score of 6.6 (Medium) and an EPSS score of 97.9% (100th percentile) as of Nov 4, 2026.
Activity feed

Recent activity

54 sources tracked across advisories, community write-ups, and news. New activity surfaces here as Mallory finds it.

View more in app →
+36 more in the timeline. Sign up to see them
Do you run this?See affected versions in your environment and whether adversaries are exploiting this vulnerability against assets like yours.
Check your exposure
Mallory platform

The version that knows your environment.

This page is what’s public. Mallory adds the parts that aren’t: which of your assets are affected, which adversaries are exploiting it right now, and which detections to deploy tonight.
Exposure mappingQuery your assets running an affected version, and investigate the blast radius.
Threat actor evidenceEvery report linking this CVE to a named adversary.
Associated malwareMalware families linked to this CVE, with evidence and IOCs.
Detection signaturesYARA, Sigma, Snort, and vendor rules, auto-deployed to your SIEM.
Vendor-by-vendor mappingCross-references every affected SKU, including bundled OEM variants.
Social activity1Community discussion across Reddit, Mastodon, and other social sources.