I’ve been running TrueNAS SCALE on two servers for a little over a year now.[1] Since almost a month after I started, I’ve been planning to virtualize these installations inside Proxmox because SCALE’s Kubernetes implementation is pretty awful, not to mention gone in newer releases. (I’m overly fond of Kubernetes and want much more control.)
Over the past couple of months, I’ve been considering eliminating TrueNAS entirely and letting Proxmox manage the same pools. Don’t get me wrong, the SCALE GUI has been a big help for someone as new to ZFS as I am and it’s great to be able to do things quickly with a few clicks, but I can build my own scripts and aliases for the CLI. It seems worth it in return for freeing my storage from the confines of TrueNAS (thereby allowing me to mount some parts of it directly rather than use NFS and SMB everywhere).
My main question is aimed at anyone who’s done a similar migration: is there anything significant you miss after having left TrueNAS behind? And, as an aside, are there any dashboards out there like the one TrueNAS provides? (I don’t mean full-blown observability stacks like PLG, which I see as complementing the simple realtime metrics I currently have access to via the TrueNAS GUI.) Proxmox doesn’t quite cover everything.
My first foray into real servers, a homelab, rack mounting, etc. ↩︎
This sounds like a great idea to me. If you’re find if having a ui for samba, I’d consider bind mounting directories into an LXC container with cockpit and the file sharing plugin.
Also webmin.
zfs plugins do exist (examples here and here) but I’ve no idea how usable these are. One appears to be abandoned and the other a bit stale?
I have a lot of love for the “it just works” aspect of TrueNAS along with their stewardship during updates (i.e. they leverage ZFSBootMenu to preserve one’s ability to instantly boot into prior releases with zero fuss). At the same time I’m tired of fighting with the middleware when I change something under the hood. Each new release steamrolls a bunch of my changes and I have to redo them (which is scriptable… but still).
I’m also a little put out by their cherry-picking patches from OpenZFS. I totally get why they’re doing it – they subject updates to additional testing before rolling to users. But personally I’d prefer to run official OpenZFS releases.
I left behind TrueNAS (as a primary storage) and went all hyperconverged on proxmox. Only hiccups were NFS sharing (ironed that out to my k8s nodes, etc.) and then using sanoid/syncoid to take snapshots. That and setting up TrueNAS SCALE as a backup server to pull snapshots from proxmox (just had to get the naming scheme right.)
Again, these weren’t insurmountable and I have no regrets about going all in on just proxmox. I’ve replaced a couple drives in it without issue as well. TrueNAS is great for just storage but it is very, very lacking in doing just about anything else IMO but hey- that is its primary goal so can’t really complain.
I use cockpit on Debian, which is able to manage most aspects of the system. I do use the 45Drives ZFS plugin, and I can manage shares, firewalls, podman containers, VMs, etc., through other parts of Cockpit.
Works well for me, and I appreciate the added layer of control over the system I have but, of course, that’s a double edged sword. I build up my system via Ansible and have a very stable platform but there was some trial and error to land where I did.
Thanks for all the replies. It sounds like I can dispense with TrueNAS, and Cockpit with the ZFS plugin should provide a small GUI veneer. I agree with @adaptive_chance that it’s nice to have an appliance OS that Just Works™ but the lack of control is frustrating.
Would it be possible to elaborate on these? What sort of issues did you have with NFS and syncing snapshots?
Oh this was more just on me. I had shared via the TrueNAS gui and even though I had copied off the /etc/exports from there the way ZFS shares vs. a native Linux install is different.
As for syncing snapshots, instead of using syncoid to push (to the truenas box, I use it to sync to other linux boxes) I decided to use TrueNAS as a backup server and pull the snapshots. I just had to modify either 1) sanoid to name the snapshots differently like TrueNAS expects or 2) modify TrueNAS to pick them up with a naming schema.
I decided on 2 so when I set up the replication task I just picked Include snapshots with the name “Matching naming schema” and then like
autosnap_%Y-%m-%d_%H:%M:%S_hourly and repeat for daily, monthly (could’ve done regex here too- would’ve been easier most likely.)