Migrating Image Library from Local Drive to Persistent Volume

This guide provides step-by-step instructions for migrating the Private Cloud Director Image Library backend from a local drive (file backend) to a volume (block storage service backend), along with important considerations for High Availability (HA) mode and multi-host Image Library configurations.

Overview

By default, the Image Service in the Private Cloud Director uses a local file-based backend to store VM images. For enhanced scalability, persistence, and centralized management, it is recommended that the image library be migrated to use a persistent storage volume.

Migrating to a persistent storage backend is particularly beneficial in High Availability (HA) environments and supports future scalability through multi-host image library configuration.

Prerequisites

• The Private Cloud Director environment is already deployed with a functional Image Library using the file backend.
• A block storage service compatible storage backend is configured and available.
• At least one available volume for storing the image library.
• Image Library role is assigned on one or more hosts.
• Administrator access to all relevant nodes and services.
• Downtime window for migration (this process involves stopping image services).

Step-by-Step Instructions

Step 1: Prepare the Volume

Create a new volume with sufficient capacity to hold all existing images (plus future growth).

Attach the volume to the host where the Image Library role is currently assigned.

Bash

pcdctl volume create --size <size-in-GB> pcd-image-repo-vol
pcdctl server add volume <image-node-hostname> pcd-image-repo-vol

On the host, identify the volume device (e.g., /dev/vdb) and mount it:

Bash

sudo mkfs.ext4 /dev/vdb
sudo mkdir -p /mnt/glance-cinder
sudo mount /dev/vdb /mnt/glance-cinder

Copy existing image data from local image library (default: /opt/pf9/pf9-glance/images) to the mounted volume:

YAML

sudo rsync -avh /opt/pf9/pf9-glance/images/ /mnt/glance-cinder/

Update the system to mount the volume persistently:

Bash

echo '/dev/vdb /mnt/glance-cinder ext4 defaults 0 0' | sudo tee -a /etc/fstab

Step 2: Update Image Library Configuration

Edit the image library service configuration file (e.g., /etc/glance/glance-api.conf) and set the following backend settings:

Bash

[DEFAULT]
enabled_backends = cinder:cinder

[glance_store]
default_backend = cinder

[cinder]
volume_type = <your-volume-type-if-applicable>

Restart the image library service:

Bash

sudo systemctl restart glance-api
sudo systemctl restart pf9-ostackhost

Verify the backend is now using block storage volume:

YAML

pcdctl image show <image-id> -f json | grep location

Info

Ensure the Image Service on each node is configured correctly with the same backend settings and the volume is mounted identically across all nodes.

High Availability and Multi-Host Image Library Considerations

• HA Mode Support: Using a blocks storage volume backend for the image library is fully supported in HA mode, as the block storage service provides persistent and shared storage accessible across multiple controller nodes.
• Multi-Host Image Library: If multiple hosts have the Image Library role:

  • The persistent storage backend ensures centralized image access without needing data synchronization between nodes.
  • All image library nodes must have access to the persistent storage service and appropriate authentication configured.