Unsupported and "Ignored" Value Not Ignored

[DHI-TommyC]

The setting persistence.accessModes is marked as "only to be used by Octopus Developers", and is supposed to be "ignored" but it is not, and its default value does not match the default value in persistence.backingVolume.accessModes. In fact, the persistence.backingVolume.accessModes is not used in the template at all.

from the templates/tentacle-deployment.yml template:

persistence:
<...>
  nfs:
<...>
    backingVolume:
      # -- The storage class name to use for the NFS Server's backing storage - if left as an empty string, an emptyDir will be used
      # @section -- Persistence
      storageClassName: ""

      # -- The access modes to use for the NFS Server's backing storage
      # @section -- Persistence
      accessModes: ["ReadWriteOnce"]
<...>
  # DO NOT USE THIS, it is not a supported configuration for customers and is only to be used by Octopus Developers
  # @ignored
  accessModes: ["ReadWriteMany"]

and templates/nfs-pvc.yml

{{- if not (include "kubernetes-agent.useCustomPvc" .) }}
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: {{ include "nfs.pvcName" . }} 
spec:
  accessModes:
    {{- .Values.persistence.accessModes | toYaml | nindent 4 }}
  resources:
    requests:
      storage: {{ .Values.persistence.size}}
  volumeName: {{ include "nfs.pvName" . }}
  storageClassName: {{ include "nfs.storageClassName" . }} 
{{- end }}

As seen here, with a default install of the agent using the wizard in the Octopus WebUI, the volume shows as RWX. The only way to use a volume in RWO mode is by overriding persistence.accessModes despite its labels.

In my case, my team and I are using an on-prem kubernetes cluster that does not support RWX mode, so the helm install times out and eventually gets a mounting error for the octopus-agent-nfs-<name>-pvc volume because the volume itself cannot be bound to the pod.

[APErebus]

Hi @DHI-TommyC,

It's not currently a supported mode for customers as it was only added for internal development needs, however we are in the process of releasing v3 of the Kubernetes Agent, which will change the default storage from RWX + NFS to RWO.

You can see #549 the in-progress work. We are hopeful of a release in the next week or so

[DHI-TommyC]

Thanks @APErebus for the response and I look forward to the new version. 👍

[APErebus]

@DHI-TommyC We have released V3 of the Kubernetes Agent that defaults to RWO storage. Can you give that a try and report back?

[DHI-TommyC]

So far so good. 👍