A recent case investigation identified that due to an inconsistent API response from Azure, in some scenarios, previously synced virtual images would be deleted and then re-synced (usually at the next sync).
The effect of this was to unlink virtual images from their node types, making it impossible to provision the custom instance which used this node type/virtual image. Our development team will decide if any action is required our side, but here is a Python script which can be run via a job to link the newly synced virtual images back to the node_type that used them…
It works by maintaining last good state of the node types in a cache and consulting this when node types are found to have no virtual image attached. It then looks up the new image ID and links it to the node so that the custom instance remains provisionable.
All operations are performed over the Morpheus REST API.
The script can be found here
It is provided as-is, and users should modify and satisfy themselves that it is working for them in a non-production environment and then run in test mode in production first.
Users with HA appliances will want to create the cache in the morpheus-ui NFS.