This is a simple test is check if all the components are recoverable.
Test data setup:
- Deploy Longhorn on a 3 nodes cluster.
- Create a volume
vol-1
using Longhorn UI. - Create a volume
vol-2
using the Longhorn storage class. - Create a volume
vol-3
with backing image. - Create an RWX volume
vol-4
. - Write some data in all the volumes created and compute the md5sum.
- Have all the volumes in attached state.
Test steps:
- Delete the IM-e from every volume and make sure every volume recovers. Check the data as well.
- Start replica rebuilding for the aforementioned volumes, and delete the IM-e while it is rebuilding. Verify the recovered volumes.
- Delete the Share-manager pod and verify the RWX volume
vol-4
is able recover. Verify the data too. - Delete the backing image manager pod and verify the pod gets recreated.
- Delete one pod of all the Longhorn components like longhorn-manager, ui, csi components etc and verify they are able to recover.