Goal

In the broader context of  of "AI at the edge",  we want the  whole stack clean and fully supported. This means we also need the accelerator vendor agreement/support for MicroShift.

For NVIDIA,  Red Hat actually runs/operates  the CI / Testing pipline (check URL in comment).

The goal of this initiative is to integrate MicroShift testing into this pipeline. Therefore, engineering teams from team MicroShift and team Ecosystem need to collaborate.

Responsibilities

Team MicroShift:

• Creating the pipelines in MicroShift CI using cloud resources to test with NVIDIA GPU using existing MicroShift Prow infra.
• The CI tests must validate the documented procedure: https://docs.nvidia.com/datacenter/cloud-native/edge/latest/nvidia-gpu-with-device-edge.html 
• Currently, this is without OLM - in the future, we might want to add NVIDIA GPU operator deployment via OLM on Microshift. This is optional. 
• Maintain the pipleines, e.g. adding new versions of MicroShift, removing out of support versions.
• Create and Fix bugs when the pipeline should fail

Team Ecosys:

• Integrate the new MicroShift prow jobs into their existing dashboard.
•  

Success Criteria

The pipline dashboard (see comment for URL) shows test results:

• Weekly test runs for all released and supported MicroShift versions (incl. EUS), e.g. V4.14, V4.15, V4.16 V4.17 as time of this writing, see product lifecyle page for the official list. 
• Nightly test runs for all yet un-released MicroShift versions (e.g. V4.18 and V4.19 at the time of this writing), see product page for the list of versions in development/testing