Skip to content

Ability to resume a failed deployment

Ability to resume a failed deployment

Proposer

  • Gerardo Garcia (Telefonica)
  • Alfonso Tierno (Telefonica)
  • Francisco Javier Ramon (Telefonica)

Type

Feature

Target MDG/TF

IM-NBI, LCM

Description

Currently, whenever the instantiation of a NS/Slice fails in the middle of the process (e.g. due to lack of resources, consumption of the VIM quota, etc.), OSM behaves as specified in ETSI NFV:

  • Report that the NS/Slice is in failed state.
  • The resources that were successfully created during the instantiation operation are kept and associated to the failed NS/Slice instance.
  • In order to release such resources associated to a failed instance, a deletion operation should be explicitly requested.

In this case, it would be highly beneficial enabling a mechanism that allowed resuming the instantiation operation at the point where it was interrupted, so that once the underlying issue in the environment is solved (e.g. the quota has been increased by the VIM administrator), the operation can be completed efficiently, avoiding the need of deleting the whole instance and re-creating all the VDUs and VLs again. This mechanism would be particularly beneficial in large NS/NSI.

This mechanism is expected to be based on SOL005's retry mechanism for NS instantiation.

Demo or definition of done

  1. Configure a VIM target which uses a tenant with a limited quota.
  2. Define a NS Package that requires more resources than the ones allowed by the quote.
  3. Launch the instantiation of the NS over the VIM target and wait until the operation fails due to lack of resources (as expected).
  4. Extend the quota so that the whole NS fits.
  5. Resume the instantiation with the retry mechanism and check that the operation does not start from scratch but that it reuses the resources already assigned to the failed NS instance.