VNF development - Resource Orchestration and Management


I’m new to NFV and I’m starting to work on a VNF for a PPDR scenario. I have already followed some tutorials online but I’m still struggling with this. Many times, when I try to deploy my VNF, I get several errors and the deployment is not completed. Because of this, I have to delete the VNF deployment in OSM but it seems to take ages to complete the deletion so I just click on the “Force Delete” button. This indeed deletes my VNF deployment but leaves a lot of garbage behind. The VM in the VIM is not deleted sometimes and neither are the Juju models that were instantiated to perform operations in my VNF. Is this normal? How can I prevent this?

Thank you in advance!

Hello @pedro.k.

This is a common issue with OSM. In fact, if you force the deletion of a resource, most of the time the dependencies (in regards to other resources) are not deleted. I advise you not to force the deletion of any resource and just wait a bit longer to check if the resource is removed or not. Most of the time, indeed, it takes quite some time. But in the research facility where I work we ended up upgrading the specs of the OSM base VM and it helped a lot.

Still, sometimes, we have to force the deletion of VNF instances. In that case, we usually have to access our VIM (Openstack, in our case) and manually delete the VM that was previously instantiated. It’s not a complex problem but is rather dull. When using Juju Charms in the VNFs, we also sometimes have to manually delete the model that was left hanging. To do so, we access OSM’s VM, list all the instantiated models (juju models), and then just force the removal of the model (juju destroy-model --force --no-wait <model_id>). It is quite important that you delete these juju models, or else they’ll start consuming HW resources, and your OSM will become slower.

Hope it helps :slight_smile: