Skip to content

Latest commit

 

History

History
66 lines (49 loc) · 1.91 KB

DEVELOPMENT.md

File metadata and controls

66 lines (49 loc) · 1.91 KB
Raw

Development Setup

Disclaimer: A lot of this makes assumptions about my own setup, so it will most likely not work out of the box if you are not using my configurations and environment.

Assumptions

  • Hypervisor setup with QEMU and KVM
  • qemu0 bridge exists
  • Host is runing a recent enough version of Linux with all required tools intalled

Development Cluster

NOTE: The dev cluster does not add ZFS support or rebase to ghcr.io/alxbl/lab/fcos for the time being as that is not necessary for development.

The purpose of this is to be able to iterate and test the bootstrapping process quickly without having to deal with hardware.

It takes about 5-6 minutes to wipe the devnet VMs and reprovision the cluster from scratch:

pwd # $HOME/Code/github.com/alxbl/lab
source .env
cd scripts/

# Stop qemu0 bridge DNSmasq if it's running
sudo systemctl stop dnsmasq@qemu0

# Download FCOS images if necessary
# Start PXE server and matchbox
docker-compose -f ../build/bootstrap/devnet.yml up -d

# Cycle the disks and VMs
sudo ./vm.sh destroy && ./vm.sh create

# Bootstrap the cluster using terraform
cd ../infra/clusters/dev
terraform apply

# This might be necessary if nothing else is running on qemu0 since 
# the iface will be down and dnsmasq won't be bound to it, causing
# iPXE to fail to boot.
sudo virsh reset node1
sudo virsh reset node2

To go a step further, it's possible to actually launch the full cluster provisioning, but this might not scale.

REPO_ROOT=$HOME/Code/github.com/alxbl/lab
cd $REPO_ROOT
export KUBECONFIG=$REPO_ROOT/infra/clusters/dev/kube.conf
kubectl apply -k $REPO_ROOT/infra/k8s/argocd

This will deploy ArgoCD and the root application, which follows the app-of-apps pattern to provision and configure everything else in the cluster.