Instance Created Using Server Snapshot Fails to Spawn

Problem

• An instance booted from a server snapshot that is created from a boot from volume instance fails to spawn and goes into an ERROR state.
• The following error is observed in the nova-scheduler logs.

Failed to call glance method data with args (u'[IMAGE_UUID]',), {} on endpoints [u'http://10.78.8.144:9292', ...]

Environment

• Platform9 Managed OpenStack - All Versions
• Glance

Cause

• The image created from the server snapshot operation does not have the correct metadata.
• The server snapshot is created for a boot from volume instance that was migrated to Platform9 using a 3rd party tool like Coriolis.

Resolution

1. Create a new image-backed instance using the same Linux distribution as the source instance.

Note: Please select the flavor size to be similar to the size of the volume-backed instance.

2. Shut down the newly created instance.
3. Locate the source volume on the Cinder host, convert the volume to a qcow2 format, and save it in the instance directory of the instance created in Step #1.

Note: Please create a backup of the volume before performing the following command.

$ qemu-img convert -f raw -O qcow2 [src-volume] /mnt/nfs/instances/[instance-UUID]/disk2

4. Navigate to the instance (created in Step #1) directory and rename the original disk file to disk.bak.
5. Rename the disk2 file created in Step #3 to disk.
6. Start the instance created in Step #1.