Heal operation for VDU

Proposers

• Alexis Romero (Indra)
• Francisco Rodríguez (Indra)

Description

In the event of failure of one or more virtual machines in a VNF, for instance due to the loss of one physical host, there should be a way to re-create them without needing a full re-deployment of the network service.

The SOL005 interface defines a heal operation for the Network Service, which is currently not implemented. Also, the required granularity should be at the VDU level, instead of the VNF level.

The operation should remove the failing VM instance in the VIM and then create a new one with the same networking, IP addressing and persistent volumes attached. New VIM Ids and IP addresses (if applicable) need to be propagated to OSM Databases and execution environments. Then, day-0 primitives will be executed.

Regarding day-1 operations, the ssh keys should be injected if needed, and the vdu-level primitives triggered.

Auto-healing is not yet considered in this feature. Notice that there is another feature aimed at auto-healing #9617

Demo or definition of done

Invoke a VIM operation for deleting one Virtual Machine. Then, execute the Heal primitive over that specific VDU. The VM should be instantiated and be accessible using ssh again. Check that the day-0 and vdu level day-1 primitives have also been executed.