Story (Required)

As a QE trying to test product I want not to depend on OpenStack storage. I want to install a fresh OpenShift cluster and store its metadata in a secret. Later, after test finished running, I want to use the metadata from the secret to uninstall the cluster.

Background (Required)

Main pipelines (acceptance-tests, upgrade-tests, flexy-install) depend on a workspace/PVC called install-dir.

First we used standard storage class in OpenStack (Cinder) but then we found out that pipeline runs often blocked each other because it only supports read-write-once mode. Later, we started using Manila-backed volumes which support read-write-many. Both are not 100% reliable and when there is a problem with storage, we need to switch from one to another (and lose data) or wait until outage is resolved.

Saving all critical data in a secret will make the whole system more resilient and reliable.

Out of scope

Approach (Required)

Leverage secret named "cluster-xxx" that is already being created in flexy-install task. Add envvars and metadata.json to this secret so that it's possible to uninstall the cluster when not needed anymore.

Dependencies

Acceptance Criteria (Mandatory)

Install-dir workspace/PVC is completely removed from tasks, pipelines, pipeline runs, README and from cluster.