The recommended way to run Falco on CoreOS is inside of its own Docker container using the install commands in the Docker section. This method allows full visibility into all containers on the host OS.
This method is automatically updated, includes some nice features such as automatic setup and bash completion, and is a generic approach that can be used on other distributions outside CoreOS as well.
However, some users may prefer to run Falco in the CoreOS toolbox. While not the recommended method, this can be achieved by installing Falco inside the toolbox using the normal installation method, and then manually running the
falcoctl driver tool:
toolbox --bind=/dev --bind=/var/run/docker.sock
curl -s https://falco.org/script/install | bash
falcoctl driver install
Google Kubernetes Engine (GKE) uses Container-Optimized OS (COS) as the default operating system for its worker node pools. COS is a security-enhanced operating system that limits access to certain parts of the underlying OS. Because of this security constraint, Falco cannot insert its kernel module to process events for system calls. However, COS provides the ability to leverage eBPF (extended Berkeley Packet Filter) to supply the stream of system calls to the Falco engine.
Falco can use eBPF with minimal configuration changes. To do so, set the
engine.kind configuration key to
ebpf in the Falco config file.
eBPF is currently supported only on GKE and COS, however here we provide installation details for a wider set of platforms
If you want to specify an alternative path for the probe file, you can also set
engine.kind.probe to the path of an existing eBPF probe.
When using the official container images, setting this environment variable will trigger the
falcoctl driver tool to download the kernel headers for the appropriate version of COS, and then compile the appropriate eBPF probe. In all the other environments you can call the
falcoctl driver tool yourself to obtain it in this way:
sudo falcoctl driver install --type ebpf
To execute the script above successfully, you will need
If you are installing Falco from packages, you can start and enable
falco-bpf.service systemd unit,
that takes care of forcing the eBPF driver for you.
If you are installing Falco with Helm, you will need to set the
driver.kind option to
helm install falco falcosecurity/falco --set driver.kind=ebpf
K3s is a lightweight, CNCF certified Kubernetes distribution. It has embedded components like etcd (datastore), CoreDNS, traefik ingress controller and etc. to simplify Kubernetes installation or upgrade.
If you are using K3s with containerd, you should set the CRI settings because the socket path is different from the default setting configured in Falco.
If you install Falco on host machine
- Append the parameter
--cri /run/k3s/containerd/containerd.sockwhen starting the Falco binary.
- Append the parameter
If you install Falco inside K3s with Helm
- Append below options when install with Helm:
--set collectors.containerd.enabled=true --set collectors.containerd.socket=/run/k3s/containerd/containerd.sock
Was this page helpful?
Let us know! You feedback will help us to improve the content and to stay in touch with our users.
Glad to hear it! Please tell us how we can improve.
Sorry to hear that. Please tell us how we can improve.