Gladius Penetration Test Report

Engagement: pt-2026-05-29-001
Scope: 172.18.0.26, gladius-target  ·  Intensity: full
Completed: 2026-05-29T21:35:00Z
10 findings 4 critical 3 high 6 attack paths 1 live hosts

Executive Summary

What we tested: We conducted a full authorized penetration test of a single Linux host (Metasploitable2) in the homelab Docker network, accessible as 'gladius-target' at 172.18.0.26. The test covered all phases from passive reconnaissance through active port scanning, service enumeration, credential checking, and CVE correlation against every identified service version.

What we found: The host is critically compromised by design, with 10 confirmed findings including 4 rated CRITICAL. Most alarmingly, port 1524 exposes an instant root shell to anyone on the network with no password required — total system takeover in under five seconds. Additionally, three separate service backdoors (vsftpd CVE-2011-2523, UnrealIRCd CVE-2010-2075, Samba CVE-2007-2447) provide independent remote root access paths, and SSH accepts the trivially guessable password 'user:user'. In total, six distinct attack paths lead to full system compromise.

What this means: Any device on the same network as this host should be considered potentially compromised. While Metasploitable2 is intentionally vulnerable for training purposes, it must never be reachable from untrusted networks or from other production systems. The findings document a real-world attack surface that mirrors vulnerabilities seen on neglected production Linux systems — making this an excellent training target, but a catastrophic risk if accidentally exposed.

Technical Summary

Attack surface: Single Linux host (Ubuntu/Metasploitable2) exposing 19 TCP services including FTP, SSH, Telnet, SMB, MySQL, PostgreSQL, IRC, NFS, Java RMI, and Apache Tomcat. Multiple services contain known trojaned/backdoored binaries from 2009–2011.

Critical findings:


Exploitable paths:

Tool coverage: port_scan, smb_enum, snmp_walk (negative — SNMP not running), web_fingerprint, nikto_scan, dir_enum, default_creds_check (SSH/FTP/Telnet), cve_correlation across 8 service versions. Port 80 unreachable from scanner; SMTP SSLv2 probe failed (no TLS on port 25).

Attack Paths

CRITICAL

INSTANT ROOT: An attacker connects to port 1524 with any TCP client (e.g. 'nc 172.18.0.26 1524'). The ingreslock backdoor immediately returns an interactive root shell with no authentication, credentials, or exploit code required. Full system compromise in under 5 seconds.

Ingreslock Root Backdoor on Port 1524 — Instant Unauthenticated Root Shell
CRITICAL

FTP BACKDOOR ROOT: An attacker connects to vsftpd on port 21, sends any username ending in ':)' (smiley face) to trigger CVE-2011-2523, then connects to port 6200 which now has a root bind-shell. Two connections, zero credentials, full root.

Anonymous FTP Login Allowed (vsftpd port 21) → vsftpd 2.3.4 Backdoor — Remote Root Shell via CVE-2011-2523
CRITICAL

SMB USERMAP RCE: An attacker connects to Samba on port 445 and sends a crafted username containing a shell command (leveraging CVE-2007-2447 usermap_script). Samba passes the unsanitized username to /bin/sh, executing the command as root. This is a well-known Metasploit module (exploit/multi/samba/usermap_script).

Samba 3.0.20 — SMB Null Session + Readable Share + RCE (CVE-2007-2447)
CRITICAL

SSH FOOTHOLD + LOCAL PRIVESC: An attacker SSHs in as user:user (F-04), gaining an unprivileged shell. From there, multiple local privilege escalation paths exist (kernel exploits, SUID binaries, sudo misconfig) to achieve root. This path is reliable even if other backdoors are patched.

SSH Default Credentials Accepted — user:user
HIGH

IRC BACKDOOR RCE: An attacker connects to UnrealIRCd on port 6667 and sends a message beginning with 'AB;' followed by an OS command (CVE-2010-2075). The trojaned binary executes the command on the server, typically as the IRC process user. A reverse shell payload achieves persistent access.

UnrealIRCd 3.2.8.1 Backdoor — Remote Code Execution via CVE-2010-2075
HIGH

TOMCAT MANAGER WAR DEPLOY: An attacker accesses /manager/html on port 8180, authenticates with default credentials (tomcat:tomcat), uploads a malicious WAR file containing a JSP web shell, and browses to the deployed path to execute OS commands within the Tomcat JVM process context.

Apache Tomcat 5.5 Manager Exposed — Default Credentials + WAR Deployment Risk

Remediation Plan

⚠ 10 of 10 findings have no remediation recorded.
CRITICAL 4 finding(s) across 1 host(s)
↳ 172.18.0.26 (4)
Ingreslock Root Backdoor on Port 1524 — Instant Unauthenticated Root Shell F-01
⚠ Agent didn't populate remediation — see references below to look it up.
References: MITRE T1190
vsftpd 2.3.4 Backdoor — Remote Root Shell via CVE-2011-2523 F-02
⚠ Agent didn't populate remediation — see references below to look it up.
References: NVD CVE-2011-2523  ·  MITRE T1190
Samba 3.0.20 — SMB Null Session + Readable Share + RCE (CVE-2007-2447) F-03
⚠ Agent didn't populate remediation — see references below to look it up.
References: NVD CVE-2007-2447  ·  MITRE T1021.002
SSH Default Credentials Accepted — user:user F-04
⚠ Agent didn't populate remediation — see references below to look it up.
References: MITRE T1078.003
HIGH 3 finding(s) across 1 host(s)
↳ 172.18.0.26 (3)
UnrealIRCd 3.2.8.1 Backdoor — Remote Code Execution via CVE-2010-2075 F-05
⚠ Agent didn't populate remediation — see references below to look it up.
References: NVD CVE-2010-2075  ·  MITRE T1190
Apache Tomcat 5.5 Manager Exposed — Default Credentials + WAR Deployment Risk F-06
⚠ Agent didn't populate remediation — see references below to look it up.
References: MITRE T1190
ProFTPD 1.3.1 — SQL Injection via CVE-2009-0542 F-07
⚠ Agent didn't populate remediation — see references below to look it up.
References: NVD CVE-2009-0542  ·  MITRE T1190
MEDIUM 3 finding(s) across 1 host(s)
↳ 172.18.0.26 (3)
Anonymous FTP Login Allowed (vsftpd port 21) F-08
⚠ Agent didn't populate remediation — see references below to look it up.
References: MITRE T1078
Legacy r-services Exposed — rexec (512), rlogin (513) F-09
⚠ Agent didn't populate remediation — see references below to look it up.
References: MITRE T1021
NFS Service Exposed via RPC (port 2049) — Potential Unauthenticated Filesystem Mount F-10
⚠ Agent didn't populate remediation — see references below to look it up.
References: MITRE T1135

Findings

CRITICAL Ingreslock Root Backdoor on Port 1524 — Instant Unauthenticated Root Shell F-01
Host: 172.18.0.26  ·  Phase: active_recon  ·  Tool: port_scan  ·  ATT&CK: T1190

Port 1524 is running a root-level backdoor shell. The nmap service probe returned an interactive root shell prompt (root@d9d6daa95c1b:/#) with zero authentication required. Any attacker on the network can connect to this port with a basic TCP client (e.g. netcat) and receive a fully interactive root shell — total system compromise in a single step, with no credentials, no exploit code, and no privileges to escalate.

CRITICAL vsftpd 2.3.4 Backdoor — Remote Root Shell via CVE-2011-2523 F-02
Host: 172.18.0.26  ·  Phase: correlation  ·  Tool: cve_correlation  ·  ATT&CK: T1190  ·  CVE: CVE-2011-2523

The vsftpd 2.3.4 FTP daemon running on port 21 contains a deliberately introduced backdoor (CVE-2011-2523, CVSS 9.8). Sending a smiley face ':)' in the FTP username field during authentication triggers the backdoor, which binds a root shell on port 6200/tcp. This provides unauthenticated remote code execution as root. The affected binary was distributed via official mirrors between June 30 and July 3, 2011.

CRITICAL Samba 3.0.20 — SMB Null Session + Readable Share + RCE (CVE-2007-2447) F-03
Host: 172.18.0.26  ·  Phase: service_enum  ·  Tool: smb_enum  ·  ATT&CK: T1021.002  ·  CVE: CVE-2007-2447

Anonymous (null session) SMB login succeeds without any credentials — meaning anyone on the network can enumerate shares and read the 'tmp' share. Beyond data exposure, Samba 3.0.20 is vulnerable to CVE-2007-2447 (usermap_script), which allows an attacker to inject arbitrary shell commands via a specially crafted username during authentication, achieving remote code execution as root. This is one of Metasploitable2's most commonly exploited paths.

CRITICAL SSH Default Credentials Accepted — user:user F-04
Host: 172.18.0.26  ·  Phase: safe_checks  ·  Tool: default_creds_check  ·  ATT&CK: T1078.003

The SSH daemon on port 22 accepted the credential pair user/user, which is a default credential pair from the built-in 12-pair test list. Any attacker who attempts this trivially guessable combination gains an authenticated remote shell session. While the initial access level is an unprivileged user account, the age of the kernel and local privilege escalation vulnerabilities present on this system make full root escalation straightforward from this foothold.

HIGH UnrealIRCd 3.2.8.1 Backdoor — Remote Code Execution via CVE-2010-2075 F-05
Host: 172.18.0.26  ·  Phase: correlation  ·  Tool: cve_correlation  ·  ATT&CK: T1190  ·  CVE: CVE-2010-2075

The UnrealIRCd IRC server on port 6667 is version 3.2.8.1, which was distributed with a trojaned binary containing a backdoor (CVE-2010-2075, CVSS 7.5). The DEBUG3_DOLOG_SYSTEM macro was modified to execute arbitrary OS commands when an attacker sends a message prefixed with 'AB;'. This provides unauthenticated remote code execution on the server, typically running as the IRC server process user.

HIGH Apache Tomcat 5.5 Manager Exposed — Default Credentials + WAR Deployment Risk F-06
Host: 172.18.0.26  ·  Phase: safe_checks  ·  Tool: nikto_scan  ·  ATT&CK: T1190

The Apache Tomcat Manager web interface (/manager/html) on port 8180 is reachable and protected only by HTTP Basic Authentication. Nikto confirmed that HTTP PUT and DELETE methods are enabled, and WebDAV is active at /webdav/. If default credentials (commonly tomcat:tomcat) are valid, an attacker can deploy a malicious WAR (web application archive) file through the Manager, resulting in remote code execution within the Tomcat process context. The Manager should never be exposed without IP restriction and strong credentials.

HIGH ProFTPD 1.3.1 — SQL Injection via CVE-2009-0542 F-07
Host: 172.18.0.26  ·  Phase: correlation  ·  Tool: cve_correlation  ·  ATT&CK: T1190  ·  CVE: CVE-2009-0542

ProFTPD 1.3.1 on port 2121 is vulnerable to SQL injection via CVE-2009-0542 (CVSS 7.5). A percent character (%) in the FTP username is not properly sanitized before being passed to mod_sql's SQL query, introducing a single quote that breaks the SQL syntax. An attacker can craft a username to bypass FTP authentication entirely or, if mod_sql_mysql/postgres is configured, extract or modify database credentials.

MEDIUM Anonymous FTP Login Allowed (vsftpd port 21) F-08
Host: 172.18.0.26  ·  Phase: active_recon  ·  Tool: port_scan  ·  ATT&CK: T1078

The vsftpd FTP server on port 21 accepts anonymous logins without any credentials. The nmap scan confirmed 'Anonymous FTP login allowed (FTP code 230)' and logged in as the 'ftp' user. This allows any unauthenticated user to browse and potentially download files accessible to the FTP anonymous account, which may include sensitive configuration files or data depending on how the FTP root is configured.

MEDIUM Legacy r-services Exposed — rexec (512), rlogin (513) F-09
Host: 172.18.0.26  ·  Phase: active_recon  ·  Tool: port_scan  ·  ATT&CK: T1021

Ports 512 (rexec) and 513 (rlogin) are open and represent ancient Unix remote execution services from the pre-SSH era. These services transmit all data including credentials in cleartext and rely on weak IP-based trust authentication rather than cryptographic keys or passwords. An attacker on the same network can perform IP spoofing or sniff credentials in transit, potentially gaining remote command execution without valid passwords.

MEDIUM NFS Service Exposed via RPC (port 2049) — Potential Unauthenticated Filesystem Mount F-10
Host: 172.18.0.26  ·  Phase: active_recon  ·  Tool: port_scan  ·  ATT&CK: T1135

The rpcbind service on port 111 advertises an NFS mountd service on port 50419/tcp and 2049/tcp (NFS). If NFS exports are configured without proper host restrictions (using '*' or broad CIDR ranges), any host on the network can mount exported filesystem paths without authentication. This could expose the entire filesystem or sensitive directories such as /home or /etc to read/write access.

Kill Chain

passive recon
resolve_target
gladius-target
Resolved gladius-target to 172.18.0.26 on Docker net_core network.
active recon
port_scan← resolve_target
172.18.0.26
19 open ports found. Critical: port 1524 returns root shell prompt with no auth. vsftpd 2.3.4, Samba 3.0.20, UnrealIRCd 3.2.8.1, OpenSSH 4.7p1, Tomcat 5.5, ProFTPD 1.3.1, MySQL 5.0.51a all confirmed.
service enum
smb_enum← port_scan
172.18.0.26
Anonymous SMB null session login succeeded. 'tmp' share readable without credentials. Samba 3.0.20 confirmed — maps to CVE-2007-2447 usermap_script RCE.
web_fingerprint← port_scan
172.18.0.26:8180
Apache Tomcat 5.5 confirmed on port 8180. Manager at /manager/html returns 401 Basic Auth challenge.
safe checks
default_creds_check← smb_enum
172.18.0.26:22
SSH credential user:user accepted. Authenticated remote shell access confirmed. CRITICAL finding F-04.
nikto_scan← web_fingerprint
172.18.0.26:8180
Tomcat Manager /manager/html accessible. PUT/DELETE HTTP methods enabled. WebDAV at /webdav/ confirmed. Default creds WAR deployment path identified.
correlation
cve_correlation← default_creds_check
vsftpd 2.3.4
CVE-2011-2523 CVSS 9.8 confirmed — backdoor in vsftpd 2.3.4 opens root shell on port 6200 when smiley-face ':)' sent in username field.
cve_correlation← port_scan
UnrealIRCd 3.2.8.1
CVE-2010-2075 CVSS 7.5 confirmed — trojaned UnrealIRCd binary executes arbitrary OS commands via AB; prefix.
cve_correlation← port_scan
ProFTPD 1.3.1
CVE-2009-0542 CVSS 7.5 confirmed — SQL injection via percent character in FTP username allows auth bypass.
cve_correlation← port_scan
MySQL 5.0.51a
CVE-2009-4484 CVSS 7.5 identified — stack overflow in TaoCrypt/yaSSL allows RCE via SSL handshake.

MITRE ATT&CK Coverage

T1046 ↗
Network Service Discovery
Discovery
T1018
Remote System Discovery
Discovery
T1135 ↗
Network Share Discovery
Discovery
T1083
File and Directory Discovery
Discovery
T1110
Brute Force
Credential Access
T1110.001
Password Guessing
Credential Access
T1078 ↗
Valid Accounts
Initial Access
T1078.003 ↗
Local Accounts
Initial Access
T1190 ↗
Exploit Public-Facing Application
Initial Access
T1133
External Remote Services
Initial Access
T1021 ↗
Remote Services
Lateral Movement
T1021.002 ↗
SMB/Windows Admin Shares
Lateral Movement
T1021.004
SSH
Lateral Movement
T1021.006
Windows Remote Management
Lateral Movement
T1071
Application Layer Protocol
Command and Control
T1071.001
Web Protocols
Command and Control
T1213
Data from Information Repositories
Collection
T1592
Gather Victim Host Information
Reconnaissance
T1592.002
Software
Reconnaissance
T1595
Active Scanning
Reconnaissance
T1595.001
Scanning IP Blocks
Reconnaissance
T1595.002 ↗
Vulnerability Scanning
Reconnaissance
T1595.003
Wordlist Scanning
Reconnaissance
T1596
Search Open Technical Databases
Reconnaissance
T1596.001
DNS/Passive DNS
Reconnaissance
T1589
Gather Victim Identity Information
Reconnaissance
T1590
Gather Victim Network Information
Reconnaissance
T1590.001
Domain Properties
Reconnaissance
T1590.002
DNS
Reconnaissance
T1590.005
IP Addresses
Reconnaissance
T1591
Gather Victim Org Information
Reconnaissance
T1552
Unsecured Credentials
Credential Access
T1552.001
Credentials In Files
Credential Access
T1552.004
Private Keys
Credential Access
T1212
Exploitation for Credential Access
Credential Access
T1003
OS Credential Dumping
Credential Access
T1499
Endpoint Denial of Service
Impact
T1219
Remote Access Software
Command and Control
T1040
Network Sniffing
Credential Access
T1199
Trusted Relationship
Initial Access
T1059
Command and Scripting Interpreter
Execution
T1554
Compromise Client Software Binary
Persistence
T1574
Hijack Execution Flow
Persistence
T1530
Data from Cloud Storage
Collection
T1119
Automated Collection
Collection
T1592.004
Client Configurations
Reconnaissance
T1596.005
Scan Databases
Reconnaissance
T1589.001
Credentials
Reconnaissance
T1003.005
Cached Domain Credentials
Credential Access