RTRTR – An RPKI data proxy

RTRTR is currently in early development. Right now, it can read RPKI data from multiple RPKI Relying Party packages via RTR and provide it, also via RTR, to routers. The HTTP server provides a monitoring endpoint in plain text and Prometheus format.

Over time, we will add more functionality, such as transport using RTR over TLS, as well as plain and signed JSON over HTTPS.

Architecture

RTRTR is a very versatile tool. It comes with a number of components for different purposes that can be connected to serve multiple use cases. There are two classes of components: Units take filtering data from somewhere – this could be other units or external sources –, and produce and constantly update one new set of data. Targets take the data set from one particular unit and serve it to an external party.

Which components RTRTR will use and how they are connected is described in a config file. An example can be found in etc/rtrtr.conf. For the moment, this example file also serves as a manual for the available components and their configuration.

Quick Start

If you have already installed Routinator, this should all be somewhat familiar.

Assuming you have a newly installed Debian or Ubuntu machine, you will need to install the C toolchain and Rust. You can then install RTRTR using Cargo, Rust’s build tool.

apt install rsync build-essential
curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh
source ~/.cargo/env
cargo install --locked rtrtr

If you have an older version of Rust and RTRTR, you can update using

rustup update
cargo install --locked --force rtrtr

If you want to try the main branch from the repository instead of a release version, you can run

cargo install --git https://github.com/NLnetLabs/rtrtr.git --branch main

Once RTRTR is installed, you need to create a config file that suits your needs. The example in etc/rtrtr.conf may be a good way to start. The config file to use needs to be passed to RTRTR via the -c option:

rtrtr -c rtrtr.conf

Using Docker

To run RTRTR with Docker you will first need to create an rtrtr.conf file somewhere on your host computer and make that available to the Docker container when you run it. For example if your config file is in /etc/rtrtr.conf on the host computer:

docker run -v /etc/rtrtr.conf:/etc/rtrtr.conf nlnetlabs/rtrtr -c /etc/rtrtr.conf

RTRTR will need network access to fetch and publish data according to the configured units and targets respectively. Explaining Docker networking is beyond the scope of this README, however below are a couple of examples to get you started.

If you need an RTRTR unit to fetch data from a source port on the host you will also need to give the Docker container access to the host network. For example one way to do this is with --net=host:

docker run --net=host ...

(where ... represents the rest of the arguments to pass to Docker and RTRTR)

This will also cause any configured RTRTR target ports to be published on the host network interface.

If you're not using --net=host you will need to tell Docker to expoee the RTRTR target ports, either one by one using -p, or you can publish the default ports exposed by the Docker container (and at the same time remap them to high numbered ports) using -P. E.g.

docker run -p 8080:8080/tcp -p 9001:9001/tcp ...

Or:

docker run -P ...

You can verify which ports are exposed using the docker ps command which should show something like this:

CONTAINER ID   IMAGE             COMMAND                  CREATED          STATUS          PORTS                                              NAMES
146237ba9b4b   nlnetlabs/rtrtr   "/sbin/tini -- rtrtr…"   16 seconds ago   Up 14 seconds   0.0.0.0:49154->8080/tcp, 0.0.0.0:49153->9001/tcp   zealous_tesla

(the output in this example shows the high-numbered port mapping that occurs when using docker run -P)