OpenStack Multi-Node Deployment

This repository provides Vagrant-based environments and deployment scripts for testing OpenStack with Kolla-Ansible. The examples cover a single-node deployment, a no-HA cluster, and a distributed topology. They are intended for development and testing, not as production reference architectures.

Why containerized OpenStack?

Kolla provides container images for OpenStack services, while Kolla-Ansible configures and deploys those services. Container-based deployment offers several benefits:

• Repeatable environments: Reusing the same image versions keeps service binaries and dependencies consistent across deployments.
• Faster provisioning: Prebuilt images reduce the need to install service packages individually on each host.
• Scaling: Additional compute nodes can run the same service images, with host preparation and inventory changes managed through Kolla-Ansible.
• Service selection: Services can be enabled or disabled through Kolla-Ansible configuration rather than manually installing or removing their packages.
• Image-based updates: Service containers can be recreated with updated images. Previous images can help recovery, but database migrations and persistent data mean upgrades are not automatically atomic or reversible.
• Service recovery: Container restart policies can restart failed containers; they do not by themselves provide high availability or recovery from host failures.

Containers still depend on a supported host operating system, networking, and persistent storage. Image portability does not remove those requirements.

Deployment examples

• All-in-one: run the OpenStack services on one VM.
• No high availability: run a controller and compute nodes, using Vagrant or physical Raspberry Pi hardware.
• Distributed: provision a multi-VM topology with separate controller, compute, storage, registry, and monitoring roles.

For prerequisites, setup guidance, and networking references, start at the documentation index.

The deployment scripts prepare nodes, configure Kolla-Ansible, and optionally build and publish container images to a local registry:

• node.sh prepares hosts and can partition, format, and mount additional volumes. Only pass disposable disks to its volume options.
• registry.sh sets up the local container image registry.
• undercloud.sh configures and runs Kolla-Ansible. The distributed sample uses a separate undercloud VM to provision service nodes.
• install.sh coordinates an all-in-one deployment, including optional registry setup and post-deployment initialization.

Deploy all-in-one directly

The all-in-one Vagrant guide provides the VM-based path. You can also run the installer directly on a dedicated, supported Linux host with passwordless sudo. This changes host packages and system configuration; do not run it on your workstation or a host with workloads you need to keep.

From the repository root, replace eno1 and eno2 with your host's interfaces:

OS_FOLDER="$PWD" \
  OS_KOLLA_NETWORK_INTERFACE=eno1 \
  OS_KOLLA_NEUTRON_EXTERNAL_INTERFACE=eno2 \
  bash ./install.sh

The management interface must have an IPv4 address. The Neutron external interface must be separate, dedicated to Neutron, and have no host IP address. Review the networking reference and upstream prerequisites before deploying.

To bootstrap without an existing checkout, download and review the installer before running it with the same interface settings:

curl -fsSL https://raw.githubusercontent.com/electrocucaracha/openstack-multinode/master/install.sh -o install-openstack.sh
less install-openstack.sh
OS_KOLLA_NETWORK_INTERFACE=eno1 \
  OS_KOLLA_NEUTRON_EXTERNAL_INTERFACE=eno2 \
  bash ./install-openstack.sh

In this case, the installer clones the repository into /opt/openstack-multinode unless OS_FOLDER points to another location.

Contributing

Contributions are welcome, including code, documentation, testing, and community outreach.