Use a PVC for Nextcloud AIO Kubernetes

ℹ️ Support
, ,
1

Hello!

I’ve currently got my Nextcloud instance hosted via snap, and I want to migrate it to Kubernetes.

It looks like the best way is via Nextcloud AIO’s helm chart: all-in-one/nextcloud-aio-helm-chart at main · nextcloud/all-in-one · GitHub

It’s almost perfect; the only problem is the Nextcloud PVC: (nextcloud-aio-nextcloud-persistentvolumeclaim.yaml)

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  labels:
    io.kompose.service: nextcloud-aio-nextcloud
  name: nextcloud-aio-nextcloud
  namespace: "{{ .Values.NAMESPACE }}"
spec:
  {{- if .Values.STORAGE_CLASS }}
  storageClassName: {{ .Values.STORAGE_CLASS }}
  {{- end }}
  accessModes:
    - ReadWriteMany
  resources:
    requests:
      storage: {{ .Values.NEXTCLOUD_STORAGE_SIZE }}

All the other volumes are ReadWriteOnce, which I need for how I have Democratic CSI set up. I need this to be ReadWriteOnce also, but I don’t know if I can change this without forking the helm chart.

Is there any way to do this for the AIO? Should I just use the normal helm chart, which is more customizable?

2

Hi, readwritemany is usually required since multiple independent containers need to have access to the same pvc…

3

If I’m going to only host one replication of the Nextcloud pod on one device, I think I should be fine. (I’m still learning kubernetes, so I’m sorry if that’s the wrong terminology).

The reason I’m moving to Kubernetes is so that I can use use TrueNAS as my storage provider and migrate if need be (hence why I’m not using the built-in Truenas nextcloud). Plus, I can use Kubernetes backup tools to back up all of my services together. Also, running raidz on two hard drives is much safer than running my Nextcloud server off of a single USB backup drive :sweat_smile:

I hadn’t realized that democratic-csi using freenas-api-iscsi needs to be ReadWriteOnce, but I don’t want to undo all the work I put into setting that up. I also don’t want to have to fork a kubernetes chart and work to manually keep it up to date with whatever changes are made to the AIO chart.

Basically, the options I see are:

  • Fork the AIO chart to make that volume ReadWriteOnce;
  • Convince the devs of the AIO helm chart to make this configurable via values (which is a terrible idea and probably shouldn’t be done);
  • Use the base Nextcloud helm chart, and deal with that monolith of a values file;
  • Deploy a docker Nextcloud image on my cluster;
  • Run Nextcloud on TrueNAS and have another place that I need to backup and deal with;
  • Let my usb drive eventually burn out and lose all my data.

Are there any other, better options available? I feel like the best option is to use the base helm chart, but I’d like to use the aio, since it seems like it’d be much easier for me to manage.