democratic-csi VS rook

The interface software regarding is not a worry of mine, as democratic-csi does the storage management for me, thus the compatibility it is not limited to the application using the storage per se, as this is handled by Kubernete's CSI drivers, being application-agnostic when utilizing the storage provided.My main worry is not latency, but rather RAM

Check out https://github.com/democratic-csi/democratic-csi

in k3s, i'm using https://github.com/democratic-csi/democratic-csi (was using iSCSI before, now everything is NFS)

Have you checked this out? GitHub democratic-csi I have yet to test this in my @home K8s cluster. It supports iSCSI volume management for FreeNAS, Synology and other CSI backends.

If you ever migrate from docker up to kubernetes, then take a look into democratic-csi. For a modest homelab, it is a valid option (said from somebody who manages a small homelab and plays around with kubernetes).

For persistent storage have you looked into using TrueNAS with a CSI provider with your container orchestrator? I'm assuming your orchestrator is Nomad or Kubernetes.

Yes, Rancher does support FreeNAS, TrueNAS. and Scale using the storage class provider https://github.com/democratic-csi/democratic-csi It's important to remember Rancher is the server. And in this case, you need to ask the question does the k8s cluster that Rancher is managing support this storage class provider? If you are using RKE the answer is Yes.

I use https://github.com/democratic-csi/democratic-csi to mount nfs/iscsi shares (and manage the shares) from my SAN (truenas box).

iSCSI with democratic-csi (https://github.com/democratic-csi/democratic-csi) works great for me on truenas. I use iSCSI for any PVs that don't need to be shared and NFS for anything I'd like to share between different pods (like movies, music).

Storage - democratic-csi looked the most promising, it has worked well so far. I am using zfs-generic-iscsi against an Ubuntu 20.04 storage server. I also tried zfs-generic-nfs and it worked successfully with the caveat of having to deal with NFS file permissions.

I have some experience with Ceph, both for work, and with homelab-y stuff.

First, bear in mind that Ceph is a distributed storage system - so the idea is that you will have multiple nodes.

For learning, you can definitely virtualise it all on a single box - but you'll have a better time with discrete physical machines.

Also, Ceph does prefer physical access to disks (similar to ZFS).

And you do need decent networking connectivity - I think that's the main thing people think of, when they think of high hardware requirements for Ceph. Ideally 10Gbe at the minimum - although more if you want higher performance - there can be a lot of network traffic, particularly with things like backfill. (25Gbps if you can find that gear cheap for homelab - 50Gbps is a technological dead-end. 100Gbps works well).

But honestly, for a homelab, a cheap mini PC or NUC with 10Gbe will work fine, and you should get acceptable performance, and it'll be good for learning.

You can install Ceph directly on bare-metal, or if you want to do the homelab k8s route, you can use Rook (https://rook.io/).

Hope this helps, and good luck! Let me know if you have any other questions.

Another option is to leverage a Kubernetes-native distributed storage solution such as Rook Ceph as the storage backend for stateful components running on Kubernetes. This has the benefit of simplifying application configuration while addressing business requirements for data backup and recovery such as the ability to take volume snapshots at a regular interval and perform application-level data recovery in case of a disaster.

This is beyond your question but might help someone else: I switch from docker-compose to kubernetes for my home lab a while ago. The storage solution I've settled on is Rook. It was a bit of up-front work learning how to get it up but now that it's done my storage is automatically managed by Ceph. I can swap out drives and Ceph basically takes care of everything itself.

The stumbling point I am at is I want to use rook.io(Ceph) as my storage solution for the cluster. The Ceph prerequisites are one of the following:

Have you looked at Rook?

I've briefly run ceph in an external mode, you can actually use a rook deployment to manage it (sort of). Here is the documentation for doing that. For me it didn't pass my testing phase because I need better networking equipment before I can try that.

This article explains the data corruption issue happened in Rook in 2021. The root cause lies in an unexpected place and can also occurs in all Ceph environment. It's interesting that Rook had started to encounter this problem recently even though this problem has existed for a long time. It's due to a series of coincidences. I wrote this article because the word "Atari" used in a non-historical context in 2021.

Rook (this is a nice article for Rook NFS)

Storage: Favor any distributed storage you know to start with for Persistent Volumes: Ceph maybe via rook.io, Longhorn if you go rancher etc

I've dealt with a lot of issues that are very close to just unplugging a node. Unfortunately on node lost, my stateful workloads using rook-ceph block storage won't migrate over to another node automatically due to an issue with rook. Stateless apps (ingress nginx, etc..) not using rook-ceph block failover to another node just fine. I've kind of accepted this for now and I know Longhorn has a feature that makes this work but I find rook-ceph to be more stable for my workloads.