Installation

We recommend installing faucet with apt for first time users and provide a Installing faucet for the first time tutorial which walks you through all the required steps for setting up faucet and gauge for the first time.

Once installed, see Configuration for documentation on how to configure faucet. Also, see Vendor-specific Documentation for documentation on how to configure your switch.

More advanced methods of installing faucet are also available here:

1. Installation using APT
2. Installation with Docker
3. Installation with Pip
4. Installing on Raspberry Pi
5. Intalling with Virtual Machine image

Installation using APT

We maintain a apt repo for installing faucet and its dependencies on Debian-based Linux distributions.

Here is a list of packages we supply:

Package	Description
python3-faucet	Install standalone faucet/gauge python3 library
faucet	Install python3 library, systemd service and default config files
gauge	Install python3 library, systemd service and default config files
faucet-all-in-one	Install faucet, gauge, prometheus and grafana. Easy to use and good for testing faucet for the first time.

Installation on Debian/Raspbian 8 (jessie)

Installing faucet on jessie requires jessie-backports.

First follow the official instructions on adding the backports repo to jessie.

sudo apt-get install curl apt-transport-https gnupg lsb-release
echo "deb https://packagecloud.io/faucetsdn/faucet/$(lsb_release -si | awk '{print tolower($0)}')/ $(lsb_release -sc) main" | sudo tee /etc/apt/sources.list.d/faucet.list
curl -L https://packagecloud.io/faucetsdn/faucet/gpgkey | sudo apt-key add -
sudo apt-get install -t jessie-backports python3-oslo.config libjs-jquery libjs-mustache
sudo apt-get update

Then to install all components for a fully functioning system on a single machine:

sudo apt-get install faucet-all-in-one

or you can install the individual components:

sudo apt-get install faucet
sudo apt-get install gauge

Installation on Debian/Raspbian 9+ and Ubuntu 16.04+

sudo apt-get install curl gnupg apt-transport-https lsb-release
echo "deb https://packagecloud.io/faucetsdn/faucet/$(lsb_release -si | awk '{print tolower($0)}')/ $(lsb_release -sc) main" | sudo tee /etc/apt/sources.list.d/faucet.list
curl -L https://packagecloud.io/faucetsdn/faucet/gpgkey | sudo apt-key add -
sudo apt-get update

Then to install all components for a fully functioning system on a single machine:

sudo apt-get install faucet-all-in-one

or you can install the individual components:

sudo apt-get install faucet
sudo apt-get install gauge

Installation with Docker

We provide official automated builds on Docker Hub so that you can easily run Faucet and it’s components in a self-contained environment without installing on the main host system.

Installing docker

We recommend installing Docker Community Edition (CE) according to the official docker engine installation guide.

Configuring dockers

First, we need to create some configuration files on our host to mount inside the docker containers to configure faucet and gauge:

sudo mkdir -p /etc/faucet
sudo vi /etc/faucet/faucet.yaml
sudo vi /etc/faucet/gauge.yaml

See the Configuration section for configuration options.

Starting dockers

We use Docker tags to differentiate between versions of Faucet. The latest tag will always point to the latest stable release of Faucet. All tagged versions of Faucet in git are also available to use, for example using the faucet/faucet:1.8.0 Docker will run the released version 1.8.0 of Faucet.

By default the Faucet and Gauge images are run as the faucet user under UID 0, GID 0. If you need to change that it can be overridden at runtime with the Docker flags: -e LOCAL_USER_ID and -e LOCAL_GROUP_ID.

To pull and run the latest version of Faucet:

mkdir -p /var/log/faucet/
docker pull faucet/faucet:latest
docker run -d \
    --name faucet \
    --restart=always \
    -v /etc/faucet/:/etc/faucet/ \
    -v /var/log/faucet/:/var/log/faucet/ \
    -p 6653:6653 \
    -p 9302:9302 \
    faucet/faucet

Port 6653 is used for OpenFlow, port 9302 is used for Prometheus - port 9302 may be omitted if you do not need Prometheus.

To pull and run the latest version of Gauge:

mkdir -p /var/log/faucet/gauge/
docker pull faucet/gauge:latest
docker run -d \
    --name gauge \
    --restart=always \
    -v /etc/faucet/:/etc/faucet/ \
    -v /var/log/faucet/:/var/log/faucet/ \
    -p 6654:6653 \
    -p 9303:9303 \
    faucet/gauge

Port 6654 is used for OpenFlow, port 9303 is used for Prometheus - port 9303 may be omitted if you do not need Prometheus.

Additional arguments

You may wish to run faucet under docker with additional arguments, for example: setting certificates for an encrypted control channel. This can be done by overriding the docker entrypoint like so:

docker run -d \
    --name faucet \
    --restart=always \
    -v /etc/faucet/:/etc/faucet/ \
    -v /etc/ryu/ssl/:/etc/ryu/ssl/ \
    -v /var/log/faucet/:/var/log/faucet/ \
    -p 6653:6653 \
    -p 9302:9302 \
    faucet/faucet \
    faucet \
    --ctl-privkey /etc/ryu/ssl/ctrlr.key \
    --ctl-cert /etc/ryu/ssl/ctrlr.cert  \
    --ca-certs /etc/ryu/ssl/sw.cert

You can get a list of all additional arguments faucet supports by running:

docker run -it faucet/faucet faucet --help

Docker compose

This is an example docker-compose file that can be used to set up gauge to talk to Prometheus and InfluxDB with a Grafana instance for dashboards and visualisations.

It can be run with:

docker-compose pull
docker-compose up

The time-series databases with the default settings will write to /opt/prometheus/ /opt/influxdb/shared/data/db you can edit these locations by modifying the docker-compose.yaml file.

On OSX, some of the default shared paths are not accessible, so to overwrite the location that volumes are written to on your host, export an environment varible name FAUCET_PREFIX and it will get prepended to the host paths. For example:

export FAUCET_PREFIX=/opt/faucet

When all the docker containers are running we will need to configure Grafana to talk to Prometheus and InfluxDB. First login to the Grafana web interface on port 3000 (e.g http://localhost:3000) using the default credentials of admin:admin.

Then add two data sources. Use the following settings for prometheus:

Name: Prometheus
Type: Prometheus
Url: http://prometheus:9090

And the following settings for InfluxDB:

Name: InfluxDB
Type: InfluxDB
Url: http://influxdb:8086
With Credentials: true
Database: faucet
User: faucet
Password: faucet

Check the connection using test connection.

From here you can add a new dashboard and a graphs for pulling data from the data sources. Hover over the + button on the left sidebar in the web interface and click Import.

We will import the following dashboards, just download the following links and upload them through the grafana dashboard import screen:

• Instrumentation
• Inventory
• Port Statistics

Installation with Pip

You can install the latest pip package, or you can install directly from git via pip.

Installing faucet

First, install some python dependencies:

apt-get install python3-dev python3-pip
pip3 install setuptools
pip3 install wheel

Then install the latest stable release of faucet from pypi, via pip:

pip3 install faucet

Or, install the latest development code from git, via pip:

pip3 install git+https://github.com/faucetsdn/faucet.git

Starting faucet manually

Faucet includes a start up script for starting Faucet and Gauge easily from the command line.

To run Faucet manually:

faucet --verbose

To run Gauge manually:

gauge --verbose

There are a number of options that you can supply the start up script for changing various options such as OpenFlow port and setting up an encrypted control channel. You can find a list of the additional arguments by running:

faucet --help

Starting faucet With systemd

Systemd can be used to start Faucet and Gauge at boot automatically:

$EDITOR /etc/systemd/system/faucet.service
$EDITOR /etc/systemd/system/gauge.service
systemctl daemon-reload
systemctl enable faucet.service
systemctl enable gauge.service
systemctl restart faucet
systemctl restart gauge

/etc/systemd/system/faucet.service should contain:

faucet.service

[Unit]
Description="Faucet OpenFlow switch controller"
After=network-online.target
Wants=network-online.target

[Service]
EnvironmentFile=/etc/default/faucet
User=faucet
Group=faucet
ExecStart=/usr/local/bin/faucet --ryu-config-file=${FAUCET_RYU_CONF} --ryu-ofp-tcp-listen-port=${FAUCET_LISTEN_PORT}
ExecReload=/bin/kill -HUP $MAINPID
Restart=always

[Install]
WantedBy=multi-user.target

/etc/systemd/system/gauge.service should contain:

gauge.service

[Unit]
Description="Gauge OpenFlow statistics controller"
After=network-online.target
Wants=network-online.target

[Service]
EnvironmentFile=/etc/default/gauge
User=faucet
Group=faucet
ExecStart=/usr/local/bin/gauge --ryu-config-file=${GAUGE_RYU_CONF} --ryu-ofp-tcp-listen-port=${GAUGE_LISTEN_PORT} --ryu-wsapi-host=${WSAPI_LISTEN_HOST} --ryu-app=ryu.app.ofctl_rest
Restart=always

[Install]
WantedBy=multi-user.target

Installing on Raspberry Pi

We provide a Raspberry Pi image running FAUCET which can be retrieved from the latest faucet release page on GitHub. Download the faucet_VERSION_raspbian-lite.zip file.

The image can then be copied onto an SD card following the same steps from the official Raspberry Pi installation guide.

Once you have booted up the Raspberry Pi and logged in using the default credentials you can follow through the Installing faucet for the first time tutorial starting from Configure prometheus to properly configure each component.

Note

It is strongly recommended to use a Raspberry Pi 3 or better.

Intalling with Virtual Machine image

We provide a VM image for running FAUCET for development and learning purposes. The VM comes pre-installed with FAUCET, GAUGE, prometheus and grafana.

Openstack’s diskimage-builder (DIB) is used to build the VM images in many formats (qcow2,tgz,squashfs,vhd,raw).

Downloading pre-built images

Pre-built images are available on github, see the latest faucet release page on GitHub and download the faucet-amd64-VERSION.qcow2 file.

Building the images

If you don’t want to use our pre-built images, you can build them yourself:

1. Install the latest disk-image-builder
2. Install a patched vhd-util
3. Run build-faucet-vm.sh

Security considerations

This VM is not secure by default, it includes no firewall and has a number of network services listening on all interfaces with weak passwords. It also includes a backdoor user (faucet) with weak credentials.

Services

The VM exposes a number of ports listening on all interfaces by default:

Service	Port
SSH	22
Faucet OpenFlow Channel	6653
Gauge OpenFlow Channel	6654
Grafana Web Interface	3000
Prometheus Web Interface	9090

Default Credentials

Service	Username	Password
VM TTY Console	faucet	faucet
SSH	faucet	faucet
Grafana Web Interface	admin	admin

Post-install steps

Grafana comes installed but unconfigured, you will need to login to the grafana web interface at http://VM_IP:3000 and configure a data source and some dashboards.

After logging in with the default credentials shown above, the first step is to add a prometheus data source, use the following settings then click Save & Test:

Name:   Prometheus
Type:   Prometheus
URL:    http://localhost:9090

Next we want to add some dashboards so that we can later view the metrics from faucet.

Hover over the + button on the left sidebar in the web interface and click Import.

We will import the following dashboards, just download the following links and upload them through the grafana dashboard import screen:

• Instrumentation
• Inventory
• Port Statistics

You will need to supply your own faucet.yaml and gauge.yaml configuration in the VM. There are samples provided at /etc/faucet/faucet.yaml and /etc/faucet/gauge.yaml.

Finally you will need to point one of the supported OpenFlow vendors at the controller VM, port 6653 is the Faucet OpenFlow control channel and 6654 is the Gauge OpennFlow control channel for monitoring.