Prefer contributions over in the official fork under the rh-ecosystem-edge organization:

https://github.com/rh-ecosystem-edge/bip-orchestrate-vm

See openshift/enhancements#565

Note that this repo is just a proof-of-concept. This repo is for debugging / experimenting with single-node bootstrap-in-place installation. Clusters created by this repo are not officially supported.

bootstrap-in-place is currently unsupported (and doesn't even work) on any cloud providers, it's meant for baremetal / virtual machines that can boot arbitrary ISO files. Even for those purposes, it might be easier for you to just use the Red Hat OpenShift Assisted Installer - it's a much more friendly interface to install Single Node OpenShift on baremetal with proper configurations, validations and bootstrap-in-place support.

If you need a single-node cluster on a cloud provider - the recommended (but still currently not officially supported) way is by just using regular IPI installer and setting the install-config.yaml control plane replicas to 1 and the compute replicas to 0. This will create, during installation, a temporary extra bootstrap node which will get automatically torn down by the installer when installation is done, leaving you with a single-node OpenShift installation.

Since the manual mode does not have strong dependencies on any platform (i.e., platform-agnostic), it is therefore the recommended mode for this repo. However, automatic provisioning for libvirt and vsphere can be found in the How to run - automatic makefile section below.

• Create a workdir for the installer - mkdir sno-workdir
• Create an install-config.yaml in the sno-workdir. An example file can be found in ./install-config.yaml.template. There are a small number of fields in the template that should be set by hand. Reasonable defaults are given below.
  • MACHINE_NETWORK - the machine network CIDR. A good default is 192.168.126.0/24.
  • CLUSTER_SVC_NETWORK - the cluster service network CIDR. A good default is 172.30.0.0/16.
  • CLUSTER_NETWORK - the cluster network CIDR. A good default is 10.128.0.0/14.
  • CLUSTER_NAME - the cluster name, the default is test-cluster.
  • BASE_DOMAIN - the cluster base domain, the default is redhat.com.
• Download the ISO to the workdir ./download_live_iso.sh sno-workdir/base.iso
• Get an installer binary using oc adm release extract --command=openshift-install --to ./bin ${RELEASE_IMAGE}
• (optional) If custom manifests are defined, generate manifests with ./manifests.sh and then copy custom manifests into generated folder. Invocation examples:

INSTALLATION_DISK=/dev/sda \
RELEASE_IMAGE=quay.io/openshift-release-dev/ocp-release:4.13.5-x86_64 \
INSTALLER_BIN=./bin/openshift-install \
INSTALLER_WORKDIR=./sno-workdir \
./manifests.sh

cp ./manifests/*.yaml $INSTALLER_WORKDIR/manifests/

• Generate an ignition file using the installer with ./generate.sh. Invocation example:

INSTALLATION_DISK=/dev/sda \
RELEASE_IMAGE=quay.io/openshift-release-dev/ocp-release:4.13.5-x86_64 \
INSTALLER_BIN=./bin/openshift-install \
INSTALLER_WORKDIR=./sno-workdir \
./generate.sh

• Embed the ignition file inside the ISO using ./embed.sh. Invocation example:

ISO_PATH=./sno-workdir/base.iso \
IGNITION_PATH=./sno-workdir/bootstrap-in-place-for-live-iso.ign \
OUTPUT_PATH=./sno-workdir/embedded.iso \
./embed.sh

You can now use sno-workdir/embedded.iso to install a single node cluster. The kubeconfig file can be found in ./sno-workdir/auth/kubeconfig

Automatic mode using Makefiles, currently supports SNO deployments on two virtualization providers, namely libvirt and vSphere.

• Set PULL_SECRET environment variable to your pull secret
• make start-iso - Spins up a VM with the liveCD. This will automatically perform the following actions:
  • Extract the openshift installer from the release image.
  • Generate the install-config.yaml.
  • Execute the openshift-installer create single-node-ignition-config command to generate the bootstrap-in-place-for-live-iso.ign.
  • Add the complete-installation.service to bootstrap-in-place-for-live-iso.ign.
  • Download the RHCOS live ISO.
  • Embed the bootstrap-in-place Ignition to the ISO.
  • Create a libvirt network & VM.
  • Boot the VM with that ISO.
• You can now monitor the progress using make ssh and journalctl -f -u bootkube.service or kubectl --kubeconfig ./sno-workdir/auth/kubeconfig get clusterversion.

• Update vSphere values and credentials in the Makefile.vsphere.
  • VSPHERE_DATACENTER_NAME
  • VSPHERE_DATASTORE_NAME
  • VSPHERE_NETWORK_NAME
  • VSPHERE_USER
  • VSPHERE_PASSWORD
  • VSPHERE_SERVER
  • VSPHERE_VM_NAME
• Set PULL_SECRET environment variable to your pull secret
• make deploy-vsphere - Spins up a VM with the liveCD in vSphere. This will automatically perform the following actions:
  • Create a workdir for the installer - mkdir sno-workdir.
  • Extract the openshift installer from the release image.
  • Generate the install-config.yaml.
  • Execute the openshift-installer create single-node-ignition-config command to generate the bootstrap-in-place-for-live-iso.ign.
  • Add the complete-installation.service to bootstrap-in-place-for-live-iso.ign.
  • Download the RHCOS live ISO.
  • Embed the bootstrap-in-place Ignition to the ISO.
  • Upload embedded ISO to vSphere datastore.
  • Create a VM in vSphere
  • Boot the VM with that ISO

• Set PULL_SECRET environment variable to your pull secret
• make start-iso-abi - Spins up a VM with the ABI ISO. This will automatically perform the following actions:
  • Extract the openshift installer from the release image.
  • Generate the install-config.yaml.
  • Execute the openshift-installer agent create image command to generate the agent.iso.
  • Create a libvirt network & VM.
  • Boot the VM with that ISO.
• You can now monitor the progress using make ssh and journalctl -f -u assisted-service.service or kubectl --kubeconfig ./sno-workdir/auth/kubeconfig get clusterversion.

• Default release image is quay.io/openshift-release-dev/ocp-release:4.13.5-x86_64 you can override it using RELEASE_IMAGE env var.
• make will execute the generate.sh script with INSTALLATION_DISK=/dev/vda
• if you’re running the installation on a BM environment, it should be updated.