I got 4 20TB drives from Amazon around Black Friday that I want to get setup for network storage. I’ve got 3 descent Ryzen 5000 series desktops that I was thinking about setting up so that I could build my own mini-Kubernetes cluster, but I don’t know if I have enough motivation. I’m pretty OCD so small projects often turn into big projects.
I don’t have an ECC motherboard though, so I want to get some input if BTRFS, ZFS, TrueNAS, or some other solution should be relatively safe without it? I guess it is a risk-factor but I haven’t had any issues yet (fingers crossed). I’ve been out of the CNCF space for a while but Rook used to be the way to go for Ceph on Kubernetes. Has there been any new projects worth checking out or should I just do RAID and get it over with? Does Ceph offer the same level of redundancy or performance? The boards have a single M.2 slot so I could add in some SSD caching.
If I go with RAID, should I do RAID 5 or 6? I’m also a bit worried because the drives are all the same so if there is an issue it could hit multiple drives at once, but I plan to try to have an online backup somewhere and if I order more drives I’ll balance it out with a different manufacturer.
If you’ve got >=3 machines and >=3 devices, I’d suggest at least strongly considering Rook. It should allow for future growth and will let you tolerate the loss of one node at the storage level too, assuming you have replication configured. Which (replication params) you can set per StorageClass in case you want to squeeze every last byte out for cases where you don’t need storage-level replication.
I’ve run my own k8s cluster for years now, and solid storage from rook really made it take off with respect to how many applications I can build and/or run on it.
As for backup, there’s velero. Though I haven’t gotten it to work on bare metal. My ideal would be to just use it to store backups in Backblaze B2 given the ridiculously low cost. Presumably I could get there with restic, since that’s my outside-k8s backup solution, but I still haven’t gotten that set up since it’s much more cloud-provider friendly.
Thanks. That is what I’m leaning towards. Do you have any suggestions for a particular distro for your K8S nodes? I’m running Arch on my desktop.
The idea of being able to setup different storage classes is very appealing, as well as learning how to build my on K8S cloud.
I’m also running arch. Unfortunately I’ve been running mine long enough that it’s just my own bespoke Ansible playbooks for configs that have morphed only as required by breaking changes or features/security I want to add. I think the best way to start from scratch these days is kubeadm, and I think it should be fairly straightforward on arch or whatever distro you like.
Fundamentally my setup is just kubelet and kubeproxy on every node, the oci runtime (CRIO for me), etcd (set up manually but certs are automated now) and then some k8s manifests templated and dropped into the k8s manifest folder for the control plane on 3 nodes for HA. The more I think about it, the more I remember how complicated it is unless you want a private CA. Which I have and love the convenience and privacy it affords me (no CTL exposing domain names unless I need public certs and they’re public anyway).
I have expanded to 6 nodes (5 of which remain, RIP laptop SSD) and just run arch on all of them because it kinda just works and I like the consistency. I also got quite good at the arch install in the process.
That’s rad… I have a set of Ansible playbooks/roles/collections already for most system-wide settings. I have a love-hate relationship with Ansible though, but it gets the job done. I may try for cloud-init first until I reach its limitations. I’ve gotten pretty good at the Arch install too, although setting up the disks with LUKS was the most challenging part. Fortunately, the few times I’ve broke things I’ve been able to boot the installer ISO and mount my LUKS volumes from memory, but I couldn’t tell you how I set them up in the first place. 🤣 However I do it, I really just want to automate the process so that I can add new nodes and expand should I decide to rent out colocation space someday.
if the data is mission critical have multiple backups for it, and test the restore process of said backups.
As for your raid vs software raid, just install truenas and throw it on there, use a LSI in IT mode and create a ZFS cluster. Don’t over think it.
Thanks. I’ve got Gigabit Fiber so I guess I’ll try Hetzner as a remote backup, or see how much it will cost to upgrade my Google Workspace account since they started enforcing their storage quotas.
It’s a lot cheaper in the long run to do an off site backup rig and just put it at a friend’s house if you can. Google will be expensive.
I don’t have any friends really 😥 and the unlimited storage with Google Workspace was $25/mo. I think it will cost me about $125/mo. now to get enough pooled storage with Google, but it is doable at least in the short-term. I guess I need to make some friends with fiber connections.
If you’re planning to go BSD, or buy all the drives you’re ever gonna have in the cluster up-front, then ZFS is great. Otherwise, be mindful of the hidden cost of ZFS. Personally, for my home server, because I’m gradually adding more drives still, I’m using mdraid on RAID6 with 8 x 8TB WD Reds/HGST Ultrastars, and I’m loving the room for activities.
Having said that, regardless of the solution you go with, since you’ve got only 4 drives, higher RAID level (and equivalent of thereof such as RAIDZ2) might be out of reach as you’d be “wasting” a lot of space for the extra piece of mind. If I were in your situation, I’d probably use RAID5 (despite RAID 5 is dead in 2009, or have they continued chugging on after 2013) for less important data (so sustain 1 drive failure) or RAID 10 if I need more performance (and depending on luck of draw, potentially sustain 2 drive failures depending on which drive fails).
Thanks. I believe TrueNAS does ZFS as well… maybe by default. If I want to keep it simple this will probably be the route I go.