This project has moved to Codeberg. You can find the latest version at https://codeberg.org/kampka/nix-flake-crowdsec Please make sure to update your dependency to receive the latests updates.
This repository contains a Nix flake for running Crowdsec on NixOS.
CrowdSec is a security tool designed to protect servers, services, and applications by analyzing user behavior and network traffic to detect and block potential attacks. It operates similarly to Fail2Ban but with a few key differences:
CrowdSec leverages the power of its community by sharing information about attacks among users. When one user detects a new threat, the details are shared across the network, allowing others to protect themselves against this threat, effectively creating a collective intelligence about emerging threats.
In simple terms, think of CrowdSec as a neighborhood watch program for the internet, where everyone contributes to and benefits from a shared pool of intelligence about potential threats.
To setup the security engine, import the module and activate the service.
{
inputs = {
crowdsec = {
url = "github:kampka/nix-flake-crowdsec";
inputs.nixpkgs.follows = "nixpkgs";
};
};
outputs = flakes @ {
self,
nixpkgs,
crowdsec,
...
}: {
nixosConfiguration.<your-hostname> = nixpkgs.lib.nixosSystem {
# ...
modules = [
# ...
crowdsec.nixosModules.crowdsec
({ pkgs, lib, ... }: {
services.crowdsec = {
enable = true;
enrollKeyFile = "/path/to/enroll-key";
settings = {
api.server = {
listen_uri = "127.0.0.1:8080";
};
};
};
})
];
};
};
}
In case you are setting up a central security engine, adjust the listen_uri
to be reachable by your bouncers.
To enroll your crowdsec engine into the central API, you need to obtain an enrollment key from the central app dashboard.
Enrolling your engine will give it access to community or commercial blocklist and decisions, depending on your plan.
Enrollment is optional, if you do not want to enroll your engine and just at on your own logs / events, simply omit the enrollKeyFile
from the settings.
For additional configuration options, please consult the (Crowdsec documentation)[https://docs.crowdsec.net/docs/configuration/crowdsec_configuration/].
This flake ships the Crowdsec firewall bouncer. It will block traffic from blacklisted IPs on the firewall level.
At the time of writing, only iptables
support has proper defaults and testing.
If you are using nftables
(networking.nftables.enable = true
), you need to supply bouncer configuration yourself (PRs welcome).
Please consult the bouncer documentation for directions.
{
inputs = {
crowdsec = {
url = "github:kampka/nix-flake-crowdsec";
inputs.nixpkgs.follows = "nixpkgs";
};
};
outputs = flakes @ {
self,
nixpkgs,
crowdsec,
...
}: {
nixosConfiguration.<your-hostname> = nixpkgs.lib.nixosSystem {
# ...
modules = [
# ...
crowdsec.nixosModules.crowdsec-firewall-bouncer;
({ pkgs, lib, ... }: {
nixpkgs.overlays = [crowdsec.overlays.default];
services.crowdsec-firewall-bouncer = {
enable = true;
settings = {
api_key = "<api-key>";
api_url = "http://localhost:8080";
};
};
})
];
};
};
}
In order to connect to your security engine, you need to add your bouncer to the security engine.
You can either use a pre-generated key or have the security engine generate one for you.
Depending on your security requirements and secrets management, this process is scriptable through an ExecStartPre
script of the engine, eg.
{
systemd.services.crowdsec.serviceConfig = {
ExecStartPre = let
script = pkgs.writeScriptBin "register-bouncer" ''
#!${pkgs.runtimeShell}
set -eu
set -o pipefail
if ! cscli bouncers list | grep -q "my-bouncer"; then
cscli bouncers add "my-bouncer" --key "<api-key>"
fi
'';
in ["${script}/bin/register-bouncer"];
};
}
This scenario is probably the most common scenario when getting started with Crowdsec. It configures the engine to parse logs from the systems local systemd-journal for failed SSH authentications and blocks IPs trying to brute-force the SSH key.
{
services.crowdsec = let
yaml = (pkgs.formats.yaml {}).generate;
acquisitions_file = yaml "acquisitions.yaml" {
source = "journalctl";
journalctl_filter = ["_SYSTEMD_UNIT=sshd.service"];
labels.type = "syslog";
};
in {
enable = true;
allowLocalJournalAccess = true;
settings = {
crowdsec_service.acquisition_path = acquisitions_file;
};
};
}
Then, install a scenario to act on your logs. The crowdsecurity/linux collection provides a good base collection to get started.
cscli collections install crowdsecurity/linux
This can be automated using eg. ExecStartPre
scripts. See some example in this README for inspiration.