§ Manual Installation

Flynn can be installed using our install script on Ubuntu 16.04 and 14.04 amd64.

We recommend starting with a clean Ubuntu installation on machines with at least 2GB of RAM, 40GB of storage, and two CPU cores each. It’s possible to run Flynn on servers with lower specs, but we don’t recommend it.

Before we get going with the installation, please note that if you plan on running a multi-node cluster, you should boot at least 3 nodes to provide high availability.

Note: If you are installing on a provider that uses a customized kernel by default, you may need to use the Ubuntu-supplied distribution kernel instead of the custom kernel for ZFS filesystem support. On Linode, use this guide for instructions on how to switch.

§ Installation

Download and run the Flynn install script:

$ sudo bash < <(curl -fsSL https://dl.flynn.io/install-flynn)

If you would rather take a look at the contents of the script before running it as root, download and run it in two separate steps:

$ curl -fsSL -o /tmp/install-flynn https://dl.flynn.io/install-flynn
... take a look at the contents of /tmp/install-flynn ...
$ sudo bash /tmp/install-flynn

To install a specific channel or version, you can use the --channel and --version flags.

Running the installer script will:

  1. Install Flynn’s runtime dependencies
  2. Download, verify and install the flynn-host binary
  3. Download and verify filesystem images for each of the Flynn components
  4. Install an Upstart job for controlling the flynn-host daemon

Some of the filesystem images are quite large (several hundred megabytes) so step 3 could take a while depending on your Internet connection.

§ Rinse and Repeat

You should install Flynn as above on every host that you want to be in the Flynn cluster.

§ Set Up Nodes

First, ensure that all network traffic is allowed between all nodes in the cluster (specifically all UDP and TCP packets). The following ports should also be open externally on the firewalls for all nodes in the cluster:

Note: A firewall with this configuration is required to prevent external access to internal management APIs.

The next step is to configure a Layer 0 cluster by starting the flynn-host daemon on all nodes. The daemon uses Raft for leader election, and it needs to be aware of all of the other nodes for it to function correctly.

If you are starting more than one node, the cluster should be configured using a discovery token. flynn-host init is a tool that handles generating and configuring the token.

On the first node, create a new token with the --init-discovery flag. The minimum multi-node cluster size is three, and this command does not need to be run if you are only starting a single node.

$ sudo flynn-host init --init-discovery

On the rest of the nodes, configure the generated discovery token:

$ sudo flynn-host init --discovery https://discovery.flynn.io/clusters/53e8402e-030f-4861-95ba-d5b5a91b5902

§ Start Flynn

Now, start the daemon and check that it has started:

Ubuntu 14.04

$ sudo start flynn-host
$ sudo status flynn-host

Ubuntu 16.04

$ sudo systemctl start flynn-host
$ sudo systemctl status flynn-host

If the status is stop/waiting, the daemon has failed to start. Check the log file (/var/log/flynn/flynn-host.log) for any errors and try starting the daemon again.

§ Bootstrap Flynn

After you have running flynn-host instances, you can now bootstrap the cluster with flynn-host bootstrap. You’ll need a domain name with DNS A records pointing to every node IP address and a second, wildcard domain CNAME to the cluster domain.


demo.localflynn.com.    A
demo.localflynn.com.    A
demo.localflynn.com.    A
*.demo.localflynn.com.  CNAME  demo.localflynn.com.

Set CLUSTER_DOMAIN to the main domain name and start the bootstrap process, specifying the number of hosts that are expected to be present and the discovery token if you created one.

$ sudo \
    CLUSTER_DOMAIN=demo.localflynn.com \
    flynn-host bootstrap \
    --min-hosts 3 \
    --discovery https://discovery.flynn.io/clusters/53e8402e-030f-4861-95ba-d5b5a91b5902

Note: You only need to run this on a single node in the cluster. It will schedule jobs on nodes across the cluster as required.

The bootstrapper will get all of necessary services running within Flynn. The final log line will contain configuration that may be used with the command-line interface.

If run into a problem while following these instructions, ensure that network traffic is flowing unimpeded through the flannel.1, flynnbr0, and veth* network interfaces and then open a GitHub issue describing the problem.

Now that you have Flynn installed and running, head over to the Flynn Basics page for a tutorial on deploying an application to Flynn.

§ Release Mailing List

If you’d like to receive email about each month’s stable release and security updates, subscribe here: