Kubernetes-worker charm should support binding to communicate with Ceph
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Kubernetes Worker Charm |
New
|
Undecided
|
Unassigned |
Bug Description
Duplicating https:/
Kubernetes-master currently has an interface "ceph-storage" that allows to define a binding and communicate to Ceph cluster via separate network.
Kubernetes-worker requires the same in order to be able to create separate interface and provision the volumes, as well as to attach them to the workloads.
This doesn't only affect ceph-based solutions but also nfs-based.
Kubernetes-worker accepts the binding of NFS charm and sets up CSI plugin to use this subnet, however NFS server only receives the default binding of the worker and populates the IP default addresses of the workers as allowed into the exports. Which leads to "access denied" error when trying to mount nfs shares into the containers.