0

OS is Debian. Imagine I have my own apt repo set up inside a private network.

This repo is set up to provide a single package to other servers on this network.

I can easily create a signing key for my repo, but this scenario got me thinking: As far as I understand, the key is needed so that whatever system downloads the package knows the package wasn't tampered with.

Could a MITM attack happen between a server and my repo? If so, how would that happen? Seems impossible since both the repo and the server are inside a private network, but I would like to understand this better.

2 Answers 2

0

Yes, man-in-the-middle attacks against package downloads can happen in local networks.

Even if a network is local, that doesn't necessarily mean it's completely isolated from the rest of the world. Some hosts or routers may have Internet access, or maybe there's a wireless access point. Anything that connects your network with the outside world potentially gives an attacker the ability to access your local network (without being physically present). For example, if just one host uses the Internet, there's a chance an attacker can gain network access by compromising this host.

The exact steps for a man-in-the-middle attack against package downloads then depend on 1) how the network is structured, 2) which parts of the network the attacker controls and 3) how the hosts access the repository. For example, a classical way of redirecting traffic meant for a particular IP address is ARP Spoofing. If this succeeds for an IP address of the repository, the attacker may be able to inject forged packages.

So checking package signatures is definitely necessary even in local networks. The risk of a man-in-the-middle attack may be lower in a local network compared to the Internet, but such attacks cannot be ruled out entirely.

0

Could a MITM attack happen between a server and my repo? If so, how would that happen?

Yes, it could. ARP poisoning is an easy and common way to execute a MitM attack on a private network.

If an attacker manages to get a root shell on any server on the same network segment as your server, it can impersonate the gateway and intercept any packet leaving the network.

On an overly simplified explanation, packets are send to MAC addresses, not IP addresses. The sender sends an ARP Request to the network asking for the physical address of the destination (its MAC Address) and the destination replies. The sender puts the pair IP:MAC on its internal table and uses this mapping on the future communications.

An attacker can just send a response without the answer (called Gratuitous ARP) to the victim and it will happily accept the mapping and use it later. The attacker can say he is the repository server and the server will ask it for the packages.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .