https://gerrit.akraino.org/r/c/ta/cloudtaf/+/1809/2/testcases/basic_func_tests/tc_008_storage_check.py

Naga is waiting for review.

https://gerrit.akraino.org/r/c/ta/cloudtaf/+/1800

Can I assume Ceph is already installed?
Can I assume RBD is also installed as part of Ceph?
Can I assume Retain/Delete StorageClass(s) provisioned with RBD? If TestCase has to create, how to get credentials and pool details of Ceph?

Naga is working on the patch. He will commit when it is completed.

Naga is working on this to-day.

Naga did not have time to work on this; he will work on test automation to complete this item

Naga working to automate these test cases.

1. These are the manual steps to test the ask.
3. Following yaml can be used to create a PVC.
   $ cat pvc.yaml
   kind: PersistentVolumeClaim
   apiVersion: v1
   metadata:
   name: ceph-claim
   spec:
   accessModes:

• ReadWriteOnce
  resources:
  requests:
  storage: 2Gi

1. Get the persitent volument name
   $ kubectl get pvc ceph-claim | awk 'NR==2 {print $3}

   '

2. Describe PV to get Reclaim Policy, RBDImage, RBDPool
   $ kubectl describe pv pvc-e6aad6b5-ffb6-422c-ae9b-0d547a9a4685
   $ sudo rbd list -p <RBDPool> | grep <RBDImage>
3. Incase of Reclaim Policy: Retain rbd list should return the image.

This is tested on a environment where Ceph is intstalled from its packages ( not docker container or k8s pods)

Naga is working to manually test RBD installation and RBD+K8s integration. He will finalize the manual steps to-day.

In this case, we can modify the test case as follows:

1. Delete the PV
2. If the RBD image still exists:
       a. Check if the StorageClass mandates retention, pass TC
       b. Check if the StorageClass does not mandate retention, fail TC
3. Alternatively, if the RBD image does not exist:
       a. Check if the StorageClass mandates retention, fail TC
       b. Check if the StorageClass does not mandate retention, pass TC

cc: Levovar

By default requested PVs retained in REC. It's controlled by the StorageClass object: https://gerrit.akraino.org/r/gitweb?p=ta/caas-kubernetes.git;a=blob;f=ansible/roles/kubernetes_ceph/templates/ceph-storageclass.yaml.j2;h=d6988730ca27b7267cd17b0e4e2a5a4853c5645c;hb=HEAD#l35 So this TC will surely fail.

cc: Levovar

Thanks for your feedback. Please see clarifications below.

Delete the PV (ensuring that the underling RBD image goes away) - The idea of this test is that when you delete a Ceph-backed PV, the deletion should cascade to the Ceph RBD image that implements the PV in Ceph. This is an important test because if RBD images do not go away on PV deletion, we may end up with slowly-accumulating wasted space in Ceph instance.

REC-34 in not related to Ceph, docker registry uses different storage backend (swift) - Thank you very much. We will provide a swift test for docker registry.

Some points about the test case:

• About

  Delete the PV (ensuring that the underling RBD image goes away)

  I am not 100% sure will work, I think deletion of a bound PV won't trigger a new PV provisioning.

• REC-34 in not related to Ceph, docker registry uses different storage backend (swift).