Logging Failures

Logging & Alerting Failures (OWASP A09)

Purpose

Provide detection patterns for logging vulnerabilities including log injection, insufficient logging of security events, secrets in logs, and log tampering vulnerabilities.

OWASP Top 10 Mapping

Category: A09 - Security Logging & Alerting Failures

CWEs:

• CWE-117: Improper Output Neutralization for Logs
• CWE-223: Omission of Security-Relevant Information
• CWE-532: Insertion of Sensitive Information into Log File
• CWE-778: Insufficient Logging

When to Use

Activate this skill when:

• Reviewing logging implementations
• Checking for log injection vulnerabilities
• Auditing security event logging
• Looking for secrets exposed in logs
• Verifying audit trail completeness

---

Log Injection (CWE-117)

Overview

Log injection occurs when user-controlled input is written to logs without sanitization, allowing attackers to inject fake log entries or manipulate log output.

Detection Patterns

Python

# String concatenation in logs
grep -rniE "logging\.(info|debug|error|warn).*\+|logger\.(info|debug|error|warn).*\+" --include="*.py"

# f-string/format in logs with user input
grep -rniE "logging\.(info|debug|error|warn).*f['\"]|logger\..*\.format\(" --include="*.py"

Vulnerable:

# VULNERABLE: User input directly in log
logger.info(f"User login: {request.form['username']}")  # Can inject newlines

Secure:

# SAFE: Structured logging
logger.info("User login", extra={"username": sanitize(username)})

Java

# String concatenation in logs
grep -rniE "log\.(info|debug|error|warn).*\+|logger\.(info|debug|error|warn).*\+" --include="*.java"

# Format with user input
grep -rniE "String\.format.*log|log.*String\.format" --include="*.java"

Vulnerable:

// VULNERABLE: Direct concatenation
logger.info("User login: " + username);  // Can inject newlines

Secure:

// SAFE: Parameterized logging
logger.info("User login: {}", sanitize(username));

Go

# Printf-style logging with user input
grep -rniE "log\.Printf|log\.Print.*\+" --include="*.go"

# Zap/logrus with user input
grep -rniE "zap\.String.*request|logrus\.WithField.*request" --include="*.go"

Vulnerable:

// VULNERABLE: Direct formatting
log.Printf("User login: %s", userInput)  // Can inject newlines

Secure:

// SAFE: Structured logging
logger.Info("user login", zap.String("username", sanitize(username)))

TypeScript

# Console/logger with concatenation
grep -rniE "console\.(log|info|error|warn).*\+|logger\.(log|info|error|warn).*\+" --include="*.ts"

PHP

# error_log with user input
grep -rniE "error_log.*\$_|syslog.*\$_|log.*\$_(GET|POST|REQUEST)" --include="*.php"

Log Injection Payloads

# Inject fake log entry
username: legitimate_user\n[ERROR] Admin password changed by attacker

# Inject multiple lines
input: line1\n[INFO] Fake entry\n[DEBUG] More fake entries

# ANSI escape codes (terminal injection)
input: \x1b[2J\x1b[1;1H  # Clear terminal

---

Insufficient Logging (CWE-778)

Overview

Missing logs for security-critical events prevents detection of attacks and incident response.

Events That MUST Be Logged

Event Category	Specific Events
Authentication	Login success/failure, logout, password change
Authorization	Access denied, privilege escalation attempts
Input Validation	Rejected/suspicious input
Session	Session creation, destruction, timeout
Data Access	Sensitive data read/write/delete
Configuration	Settings changes, feature toggles
Errors	Exceptions, failures (without stack traces)

Detection Patterns

# Find auth functions without logging
grep -rniE "def (login|authenticate|authorize|check_permission)" --include="*.py" -A 10 | grep -v "log\."

# Find exception handlers without logging
grep -rniE "except.*:|catch\s*\(" --include="*.py" --include="*.java" -A 5 | grep -v "log\|logger"

# Find permission checks without logging
grep -rniE "has_permission|is_admin|check_role" --include="*.go" --include="*.py" --include="*.java" -A 5 | grep -v "log"

What to Check

# Authentication logging
grep -rniE "login|authenticate|logout" --include="*.go" --include="*.py" --include="*.java" --include="*.ts" --include="*.php" | xargs -I {} sh -c 'grep -l "log" {} || echo "MISSING: {}"'

# Failed access logging
grep -rniE "forbidden|unauthorized|access.*denied" --include="*.go" --include="*.py" --include="*.java" --include="*.ts" --include="*.php"

# Data modification logging
grep -rniE "delete|update|insert" --include="*.go" --include="*.py" --include="*.java" --include="*.ts" --include="*.php" | xargs -I {} sh -c 'grep -l "log\|audit" {} || echo "MISSING: {}"'

---

Secrets in Logs (CWE-532)

Overview

Logging sensitive data (passwords, tokens, API keys) exposes them to anyone with log access.

Detection Patterns

# Password in logs
grep -rniE "log.*(password|passwd|pwd|secret|token|api_key|apikey|credential)" --include="*.go" --include="*.py" --include="*.java" --include="*.ts" --include="*.php"

# Request body logging (may contain secrets)
grep -rniE "log.*request\.body|log.*req\.body|log.*getBody" --include="*.go" --include="*.py" --include="*.java" --include="*.ts" --include="*.php"

# Full object logging
grep -rniE "log.*user\)|log.*%v.*user|log.*JSON\.stringify" --include="*.go" --include="*.py" --include="*.java" --include="*.ts" --include="*.php"

Sensitive Fields to Never Log

• password, passwd, pwd
• token, access_token, refresh_token
• api_key, apikey, secret_key
• credit_card, ccn, cvv
• ssn, social_security
• private_key, secret
• Authorization headers
• Session IDs

Secure Pattern - Field Redaction

# Python - Redact sensitive fields
SENSITIVE_FIELDS = ['password', 'token', 'api_key', 'secret']

def redact_sensitive(data: dict) -> dict:
    return {k: '***REDACTED***' if k.lower() in SENSITIVE_FIELDS else v
            for k, v in data.items()}

logger.info("User data", extra=redact_sensitive(user_data))

// Go - Redact sensitive fields
func redactSensitive(data map[string]interface{}) map[string]interface{} {
    sensitive := []string{"password", "token", "api_key"}
    for _, key := range sensitive {
        if _, exists := data[key]; exists {
            data[key] = "***REDACTED***"
        }
    }
    return data
}

---

Log Tampering

Overview

If logs can be modified or deleted, attackers can cover their tracks.

Detection Patterns

# Predictable log file paths
grep -rniE "logfile\s*=|LOG_FILE\s*=|log_path\s*=" --include="*.go" --include="*.py" --include="*.java" --include="*.ts" --include="*.php" --include="*.env"

# User-controlled log paths
grep -rniE "log.*request\.(path|param|query)" --include="*.go" --include="*.py" --include="*.java" --include="*.ts" --include="*.php"

# Log rotation without integrity
grep -rniE "RotatingFileHandler|logrotate|rotate.*log" --include="*.go" --include="*.py" --include="*.java" --include="*.conf"

Secure Patterns

• Write logs to append-only storage
• Use centralized logging (SIEM)
• Implement log integrity verification (signing)
• Restrict log file permissions
• Use remote logging to prevent local tampering

---

Missing Alerting

Overview

Logs without alerts allow attacks to go unnoticed.

Events That Should Alert

Event	Alert Priority
Multiple failed logins	HIGH
Admin actions	MEDIUM
Privilege escalation	CRITICAL
Sensitive data access	HIGH
Configuration changes	MEDIUM
Error rate spike	HIGH
Unusual access patterns	MEDIUM

Detection Patterns

# Check for alerting configuration
grep -rniE "alert|notify|pagerduty|slack.*webhook|email.*notify" --include="*.go" --include="*.py" --include="*.java" --include="*.ts" --include="*.yaml" --include="*.json"

# Check for threshold-based alerts
grep -rniE "threshold|rate_limit|max_attempts|lockout" --include="*.go" --include="*.py" --include="*.java" --include="*.ts" --include="*.php"

---

Language-Specific Secure Logging

Python (structlog)

import structlog

logger = structlog.get_logger()

# Structured with automatic sanitization
logger.info(
    "user_login",
    username=sanitize(username),
    ip_address=request.remote_addr,
    success=True
)

Java (SLF4J + Logback)

// Use MDC for structured context
MDC.put("userId", sanitize(userId));
MDC.put("action", "login");
logger.info("User authentication successful");
MDC.clear();

Go (Zap)

logger.Info("user authentication",
    zap.String("user_id", sanitize(userID)),
    zap.String("action", "login"),
    zap.Bool("success", true),
)

TypeScript (Winston)

logger.info('user authentication', {
    userId: sanitize(userId),
    action: 'login',
    success: true,
    timestamp: new Date().toISOString()
});

---

Log Format Requirements

A secure log entry should include:

Field	Purpose
Timestamp	When (ISO 8601)
Level	Severity (INFO, WARN, ERROR)
Source	Where (service, file, function)
User	Who (user ID, not PII)
Action	What happened
Resource	What was accessed
Outcome	Success/failure
Request ID	Correlation
IP Address	Origin

Example Secure Log Format

{
  "timestamp": "2025-01-15T10:30:00Z",
  "level": "INFO",
  "service": "auth-service",
  "event": "user_login",
  "user_id": "usr_12345",
  "action": "authenticate",
  "outcome": "success",
  "request_id": "req_abc123",
  "ip_address": "192.168.1.1",
  "user_agent": "Mozilla/5.0..."
}

---

Verification Checklist

• Log injection prevented (sanitized input)
• Authentication events logged
• Authorization failures logged
• Data access logged
• Sensitive data redacted
• Logs written to secure location
• Log integrity protected
• Alerting configured for critical events
• Log retention policy defined

---

CWE References

CWE	Name	Example
CWE-117	Log Injection	Newlines in log messages
CWE-223	Omission of Info	Missing auth failure logs
CWE-532	Secrets in Logs	Passwords logged
CWE-778	Insufficient Logging	No audit trail