Unable to Snapshot Instance: "Image <path>/disk is unacceptable: Unable to convert image to qcow2"

Problem

• Unable to snapshot an instance (in any power state).
• The following logs are present in /var/log/pf9/ostackhost.log file on the hypervisor.

2023-09-23 17:59:11.504 TRACE oslo_messaging.rpc.server nova.exception.ImageUnacceptable: Image /opt/pf9/data/instances/84277b18-00d2-44f9-bc9d-744786857d36/disk is unacceptable: Unable to convert image to qcow2: Unexpected error while running command.

2023-09-23 17:59:11.504 TRACE oslo_messaging.rpc.server Command: qemu-img convert -t none -O qcow2 -f qcow2 /opt/pf9/data/instances/84277b18-00d2-44f9-bc9d-744786857d36/disk /opt/pf9/data/instances//snapshots/tmp15vuki7h/c15b44c0dc1749f5ac739322076aafde

2023-09-23 17:59:11.504 TRACE oslo_messaging.rpc.server Stderr: "qemu-img: Could not open '/opt/pf9/data/instances/84277b18-00d2-44f9-bc9d-744786857d36/disk': Could not open backing file: Could not open '/opt/pf9/data/instances/_base/f3eb8b1b8773b5bb41db6349d2170e270eb38fb5': No such file or directory\n"

Environment

• Platform9 Managed OpenStack - All Versions
• Nova
• Glance

Cause

The backing file for the instance is missing, which can occur if an instance becomes un-defined in Libvirt – for example, due to a partially failed migration. The root cause for the backing file missing may be different under certain circumstances.

Resolution

Option A: Copy the Backing File from Another Hypervisor

1. From a working, functional hypervisor – if an instance is scheduled with the same image as the one which is failing to snapshot, it may have a copy of the same data which you can use to copy. Check for any matching folders/files in /opt/pf9/data/instances/_base/ .
2. Copy the directory/files to the same path on the non-functional hypervisor.
3. Attempt to re-build the instance.

Option B: Migrate the Instance

1. Migrate the instance. The image should be re-downloaded from Glance to the hypervisor.
2. Snapshot the instance.