Examples contains Config Service data directories showcasing different network-bootable baremetal clusters. These examples work with the libvirt VMs created by scripts/libvirt and the Libvirt Guide.

Name	Description	Docs
etcd-small	Cluster with 1 etcd node, 4 proxies	reference
etcd-large	Cluster with 3 etcd nodes, 2 proxies	reference
kubernetes	Kubernetes cluster with 1 master, 1 worker, 1 dedicated etcd node	reference

Experimental

These examples are experimental and designed to demonstrate booting and configuring CoreOS clusters, especially locally. They have NOT been hardened for production yet.

Virtual or Physical Hardware

Create a network of virtual hardware on your Linux machine before running the Config Service with one of these examples.

Create 5 libvirt VM nodes, which should be enough for any of the examples. The scripts/libvirt script will create 5 VM nodes with known hardware attributes, on the docker0 bridge network.

# Fedora/RHEL
dnf install virt-manager

# create node1 ... node5
./scripts/libvirt create

The nodes can be conveniently managed together.

./scripts/libvirt reboot
./scripts/libvirt shutdown        # graceful
./scripts/libvirt poweroff        # non-graceful
./scripts/libvirt destroy

The Config Service can uses these examples to provision physical clusters, but you'll have to edit the static IPs to suit your network and hardware. See the Baremetal Guide.

Config Service

The Config service matches machines to boot configurations, ignition configs, and cloud configs. It optionally serves OS images and other assets.

Let's run the config service on the virtual network.

docker pull quay.io/coreos/bootcfg:latest

Run the command for the example you wish to use.

etcd-small Cluster

docker run -p 8080:8080 --name=bootcfg --rm -v $PWD/examples/etcd-small:/data:Z -v $PWD/assets:/assets:Z coreos/bootcfg:latest -address=0.0.0.0:8080 -log-level=debug

etcd-large Cluster

docker run -p 8080:8080 --name=bootcfg --rm -v $PWD/examples/etcd-large:/data:Z -v $PWD/assets:/assets:Z coreos/bootcfg:latest -address=0.0.0.0:8080 -log-level=debug

Kubernetes Cluster

docker run -p 8080:8080 --name=bootcfg --rm -v $PWD/examples/kubernetes:/data:Z -v $PWD/assets:/assets:Z coreos/bootcfg:latest -address=0.0.0.0:8080 -log-level=debug

The mounted data directory (e.g. -v $PWD/examples/etcd-small:/data:Z) depends on the example you wish to run.

Assets

The examples require the CoreOS stable PXE kernel and initrd images be served by the Config service. Run get-coreos to download those images to assets.

./scripts/get-coreos

The following examples require a few additional assets be downloaded or generated. Acquire those assets before continuing.

Kubernetes Cluster

Network Environment

Run an iPXE setup with DHCP and TFTP on the virtual network on your machine similar to what would be present on a real network. This allocates IP addresses to VM hosts, points PXE booting clients to the config service, and chainloads iPXE.

docker run --rm --cap-add=NET_ADMIN coreos/dnsmasq -d -q --dhcp-range=172.17.0.43,172.17.0.99 --enable-tftp --tftp-root=/var/lib/tftpboot --dhcp-userclass=set:ipxe,iPXE --dhcp-boot=tag:#ipxe,undionly.kpxe --dhcp-boot=tag:ipxe,http://172.17.0.2:8080/boot.ipxe

Boot

Reboot the nodes to PXE boot them into your new cluster!

./scripts/libvirt reboot
# if nodes are in a non-booted state
./scripts/libvirt poweroff

All examples use autologin so you can check whether your nodes were setup correctly, depending on the cluster example you chose.

If something goes wrong, see troubleshooting.

If everything works, congratulations! Stay tuned for developments.

README.md