typhoon/docs/flatcar-linux/bare-metal.md

16 KiB

Bare-Metal

In this tutorial, we'll network boot and provision a Kubernetes v1.27.4 cluster on bare-metal with Flatcar Linux.

First, we'll deploy a Matchbox service and setup a network boot environment. Then, we'll declare a Kubernetes cluster using the Typhoon Terraform module and power on machines. On PXE boot, machines will install Container Linux to disk, reboot into the disk install, and provision themselves as Kubernetes controllers or workers via Ignition.

Controller hosts are provisioned to run an etcd-member peer and a kubelet service. Worker hosts run a kubelet service. Controller nodes run kube-apiserver, kube-scheduler, kube-controller-manager, and coredns while kube-proxy and calico (or flannel) run on every node. A generated kubeconfig provides kubectl access to the cluster.

Requirements

  • Machines with 2GB RAM, 30GB disk, PXE-enabled NIC, IPMI
  • PXE-enabled network boot environment (with HTTPS support)
  • Matchbox v0.6+ deployment with API enabled
  • Matchbox credentials client.crt, client.key, ca.crt
  • Terraform v0.13.0+

Machines

Collect a MAC address from each machine. For machines with multiple PXE-enabled NICs, pick one of the MAC addresses. MAC addresses will be used to match machines to profiles during network boot.

  • 52:54:00:a1:9c:ae (node1)
  • 52:54:00:b2:2f:86 (node2)
  • 52:54:00:c3:61:77 (node3)

Configure each machine to boot from the disk through IPMI or the BIOS menu.

ipmitool -H node1 -U USER -P PASS chassis bootdev disk options=persistent

During provisioning, you'll explicitly set the boot device to pxe for the next boot only. Machines will install (overwrite) the operating system to disk on PXE boot and reboot into the disk install.

!!! tip "" Ask your hardware vendor to provide MACs and preconfigure IPMI, if possible. With it, you can rack new servers, terraform apply with new info, and power on machines that network boot and provision into clusters.

DNS

Create a DNS A (or AAAA) record for each node's default interface. Create a record that resolves to each controller node (or re-use the node record if there's one controller).

  • node1.example.com (node1)
  • node2.example.com (node2)
  • node3.example.com (node3)
  • myk8s.example.com (node1)

Cluster nodes will be configured to refer to the control plane and themselves by these fully qualified names and they'll be used in generated TLS certificates.

Matchbox

Matchbox is an open-source app that matches network-booted bare-metal machines (based on labels like MAC, UUID, etc.) to profiles to automate cluster provisioning.

Install Matchbox on a Kubernetes cluster or dedicated server.

!!! tip Deploy Matchbox as service that can be accessed by all of your bare-metal machines globally. This provides a single endpoint to use Terraform to manage bare-metal clusters at different sites. Typhoon will never include secrets in provisioning user-data so you may even deploy matchbox publicly.

Matchbox provides a TLS client-authenticated API that clients, like Terraform, can use to manage machine matching and profiles. Think of it like a cloud provider API, but for creating bare-metal instances.

Generate TLS client credentials. Save the ca.crt, client.crt, and client.key where they can be referenced in Terraform configs.

mv ca.crt client.crt client.key ~/.config/matchbox/

Verify the matchbox read-only HTTP endpoints are accessible (port is configurable).

$ curl http://matchbox.example.com:8080
matchbox

Verify your TLS client certificate and key can be used to access the Matchbox API (port is configurable).

$ openssl s_client -connect matchbox.example.com:8081 \
  -CAfile ~/.config/matchbox/ca.crt \
  -cert ~/.config/matchbox/client.crt \
  -key ~/.config/matchbox/client.key

PXE Environment

Create an iPXE-enabled network boot environment. Configure PXE clients to chainload iPXE firmware compiled to support HTTPS downloads. Instruct iPXE clients to chainload from your Matchbox service's /boot.ipxe endpoint.

For networks already supporting iPXE clients, you can add a default.ipxe config.

# /var/www/html/ipxe/default.ipxe
chain http://matchbox.foo:8080/boot.ipxe

For networks with Ubiquiti Routers, you can configure the router itself to chainload machines to iPXE and Matchbox.

Read about the many ways to setup a compliant iPXE-enabled network. There is quite a bit of flexibility:

  • Continue using existing DHCP, TFTP, or DNS services
  • Configure specific machines, subnets, or architectures to chainload from Matchbox
  • Place Matchbox behind a menu entry (timeout and default to Matchbox)

!!! note "" TFTP chainloading to modern boot firmware, like iPXE, avoids issues with old NICs and allows faster transfer protocols like HTTP to be used.

!!! warning Compile iPXE from source with support for HTTPS downloads. iPXE's pre-built firmware binaries do not enable this. If you cannot enable HTTPS downloads, set download_protocol = "http" (discouraged).

Terraform Setup

Install Terraform v0.13.0+ on your system.

$ terraform version
Terraform v1.0.0

Read concepts to learn about Terraform, modules, and organizing resources. Change to your infrastructure repository (e.g. infra).

cd infra/clusters

Provider

Configure the Matchbox provider to use your Matchbox API endpoint and client certificate in a providers.tf file.

provider "matchbox" {
  endpoint    = "matchbox.example.com:8081"
  client_cert = file("~/.config/matchbox/client.crt")
  client_key  = file("~/.config/matchbox/client.key")
  ca          = file("~/.config/matchbox/ca.crt")
}

provider "ct" {}

terraform {
  required_providers {
    ct = {
      source  = "poseidon/ct"
      version = "0.11.0"
    }
    matchbox = {
      source = "poseidon/matchbox"
      version = "0.5.2"
    }
  }
}

Cluster

Define a Kubernetes cluster using the module bare-metal/flatcar-linux/kubernetes.

module "mercury" {
  source = "git::https://github.com/poseidon/typhoon//bare-metal/flatcar-linux/kubernetes?ref=v1.27.4"

  # bare-metal
  cluster_name            = "mercury"
  matchbox_http_endpoint  = "http://matchbox.example.com"
  os_channel              = "flatcar-stable"
  os_version              = "2345.3.1"

  # configuration
  k8s_domain_name    = "node1.example.com"
  ssh_authorized_key = "ssh-rsa AAAAB3Nz..."

  # machines
  controllers = [{
    name   = "node1"
    mac    = "52:54:00:a1:9c:ae"
    domain = "node1.example.com"
  }]
  workers = [
    {
      name   = "node2",
      mac    = "52:54:00:b2:2f:86"
      domain = "node2.example.com"
    },
    {
      name   = "node3",
      mac    = "52:54:00:c3:61:77"
      domain = "node3.example.com"
    }
  ]

  # set to http only if you cannot chainload to iPXE firmware with https support
  # download_protocol = "http"
}

Workers with similar features can be defined inline using the workers field as shown above. It's also possible to define discrete workers that attach to the cluster. Discrete workers are more advanced, but more verbose.

module "mercury-node1" {
  source = "git::https://github.com/poseidon/typhoon//bare-metal/flatcar-linux/kubernetes/worker?ref=v1.27.4"

  # bare-metal
  cluster_name = "mercury"
  matchbox_http_endpoint  = "http://matchbox.example.com"
  os_channel              = "flatcar-stable"
  os_version              = "2345.3.1"

  # configuration
  name               = "node2"
  mac                = "52:54:00:b2:2f:86"
  domain             = "node2.example.com"
  kubeconfig         = module.mercury.kubeconfig-admin
  ssh_authorized_key = "ssh-rsa AAAAB3Nz..."

  # optional
  snippets       = []
  node_labels    = []
  node_taints    = []
  install_disk   = "/dev/vda"
  cached_install = false
}

...

Reference the variables docs or the variables.tf source.

ssh-agent

Initial bootstrapping requires bootstrap.service be started on one controller node. Terraform uses ssh-agent to automate this step. Add your SSH private key to ssh-agent.

ssh-add ~/.ssh/id_rsa
ssh-add -L

Apply

Initialize the config directory if this is the first use with Terraform.

terraform init

Plan the resources to be created.

$ terraform plan
Plan: 55 to add, 0 to change, 0 to destroy.

Apply the changes. Terraform will generate bootstrap assets and create Matchbox profiles (e.g. controller, worker) and matching rules via the Matchbox API.

$ terraform apply
module.mercury.null_resource.copy-controller-secrets.0: Still creating... (10s elapsed)
module.mercury.null_resource.copy-worker-secrets.0: Still creating... (10s elapsed)
...

Apply will then loop until it can successfully copy credentials to each machine and start the one-time Kubernetes bootstrap service. Proceed to the next step while this loops.

Power

Power on each machine with the boot device set to pxe for the next boot only.

ipmitool -H node1.example.com -U USER -P PASS chassis bootdev pxe
ipmitool -H node1.example.com -U USER -P PASS power on

Machines will network boot, install Container Linux to disk, reboot into the disk install, and provision themselves as controllers or workers.

!!! tip "" If this is the first test of your PXE-enabled network boot environment, watch the SOL console of a machine to spot any misconfigurations.

Bootstrap

Wait for the bootstrap step to finish bootstrapping the Kubernetes control plane. This may take 5-15 minutes depending on your network.

module.mercury.null_resource.bootstrap: Still creating... (6m10s elapsed)
module.mercury.null_resource.bootstrap: Still creating... (6m20s elapsed)
module.mercury.null_resource.bootstrap: Still creating... (6m30s elapsed)
module.mercury.null_resource.bootstrap: Still creating... (6m40s elapsed)
module.mercury.null_resource.bootstrap: Creation complete (ID: 5441741360626669024)

Apply complete! Resources: 55 added, 0 changed, 0 destroyed.

To watch the install to disk (until machines reboot from disk), SSH to port 2222.

# before v1.10.1
$ ssh debug@node1.example.com
# after v1.10.1
$ ssh -p 2222 core@node1.example.com

To watch the bootstrap process in detail, SSH to the first controller and journal the logs.

$ ssh core@node1.example.com
$ journalctl -f -u bootstrap
The connection to the server cluster.example.com:6443 was refused - did you specify the right host or port?
Waiting for static pod control plane
...
serviceaccount/calico-node unchanged
systemd[1]: Started Kubernetes control plane.

Verify

Install kubectl on your system. Obtain the generated cluster kubeconfig from module outputs (e.g. write to a local file).

resource "local_file" "kubeconfig-mercury" {
  content  = module.mercury.kubeconfig-admin
  filename = "/home/user/.kube/configs/mercury-config"
}

List nodes in the cluster.

$ export KUBECONFIG=/home/user/.kube/configs/mercury-config
$ kubectl get nodes
NAME                STATUS  ROLES   AGE  VERSION
node1.example.com   Ready   <none>  10m  v1.27.4
node2.example.com   Ready   <none>  10m  v1.27.4
node3.example.com   Ready   <none>  10m  v1.27.4

List the pods.

$ kubectl get pods --all-namespaces
NAMESPACE     NAME                                       READY     STATUS    RESTARTS   AGE
kube-system   calico-node-6qp7f                          2/2       Running   1          11m
kube-system   calico-node-gnjrm                          2/2       Running   0          11m
kube-system   calico-node-llbgt                          2/2       Running   0          11m
kube-system   coredns-1187388186-dj3pd                   1/1       Running   0          11m
kube-system   coredns-1187388186-mx9rt                   1/1       Running   0          11m
kube-system   kube-apiserver-node1.example.com           1/1       Running   0          11m
kube-system   kube-controller-node1.example.com          1/1       Running   1          11m
kube-system   kube-proxy-50sd4                           1/1       Running   0          11m
kube-system   kube-proxy-bczhp                           1/1       Running   0          11m
kube-system   kube-proxy-mp2fw                           1/1       Running   0          11m
kube-system   kube-scheduler-node1.example.com           1/1       Running   0          11m

Going Further

Learn about maintenance and addons.

Variables

Check the variables.tf source.

Required

Name Description Example
cluster_name Unique cluster name "mercury"
matchbox_http_endpoint Matchbox HTTP read-only endpoint "http://matchbox.example.com:port"
os_channel Channel for a Container Linux derivative flatcar-stable, flatcar-beta, flatcar-alpha
os_version Version for a Container Linux derivative to PXE and install "2345.3.1"
k8s_domain_name FQDN resolving to the controller(s) nodes. Workers and kubectl will communicate with this endpoint "myk8s.example.com"
ssh_authorized_key SSH public key for user 'core' "ssh-rsa AAAAB3Nz..."
controllers List of controller machine detail objects (unique name, identifying MAC address, FQDN) [{name="node1", mac="52:54:00:a1:9c:ae", domain="node1.example.com"}]

Optional

Name Description Default Example
workers List of worker machine detail objects (unique name, identifying MAC address, FQDN) [] [{name="node2", mac="52:54:00:b2:2f:86", domain="node2.example.com"}, {name="node3", mac="52:54:00:c3:61:77", domain="node3.example.com"}]
download_protocol Protocol iPXE uses to download the kernel and initrd. iPXE must be compiled with crypto support for https. Unused if cached_install is true "https" "http"
cached_install PXE boot and install from the Matchbox /assets cache. Admin MUST have downloaded Container Linux or Flatcar images into the cache false true
install_disk Disk device where Container Linux should be installed "/dev/sda" "/dev/sdb"
networking Choice of networking provider "cilium" "calico" or "cilium" or "flannel"
network_mtu CNI interface MTU (calico-only) 1480 -
snippets Map from machine names to lists of Container Linux Config snippets {} examples
network_ip_autodetection_method Method to detect host IPv4 address (calico-only) "first-found" "can-reach=10.0.0.1"
pod_cidr CIDR IPv4 range to assign to Kubernetes pods "10.2.0.0/16" "10.22.0.0/16"
service_cidr CIDR IPv4 range to assign to Kubernetes services "10.3.0.0/16" "10.3.0.0/24"
kernel_args Additional kernel args to provide at PXE boot [] ["kvm-intel.nested=1"]
worker_node_labels Map from worker name to list of initial node labels {} {"node2" = ["role=special"]}
worker_node_taints Map from worker name to list of initial node taints {} {"node2" = ["role=special:NoSchedule"]}
oem_type An OEM type to install with flatcar-install. "" "vmware_raw"