Hi everyone, I’ve been working on my homelab for a year and a half now, and I’ve tested several approaches to managing NAS and selfhosted applications. My current setup is an old desktop computer that boots into Proxmox, which has two VMs:
- TrueNAS Scale: manages storage, shares and replication.
- Debian 12 w/ docker: for all of my selfhosted applications.
The applications connect to the TrueNAS’ storage via NFS. I have two identical HDDs as a mirror, another one that has no failsafe (but it’s fine, because the data it contains is non-critical), and an external HDD that I want to use for replication, or some other use I still haven’t decided.
Now, the issue is the following. I’ve noticed that TrueNAS complains that the HDDs are Unhealthy and has complained about checksum errors. It also turns out that it can’t run S.M.A.R.T. checks, because instead of using an HBA, I’m directly passing the entire HDDs by ID to the VM. I’ve read recently that it’s discouraged to pass virtualized disks to TrueNAS, as data corruption can occur. And lately I was having trouble with a selfhosted instance of gitea, where data (apparently) got corrupted, and git was throwing errors when you tried to fetch or pull. I don’t know if this is related or not.
Now the thing is, I have a very limited budget, so I’m not keen on buying a dedicated HBA just out of a hunch. Is it really needed?
I mean, I know I could run TrueNAS directly, instead of using Proxmox, but I’ve found TrueNAS to be a pretty crappy Hypervisor (IMHO) in the past.
My main goal is to be able to manage the data that is used in selfhosted applications separately. For example, I want to be able to access Nextcloud’s files, even if the docker instance is broken. But maybe this is just an irrational fear, and I should instead backup the entire docker instances and hope for the best, or maybe I’m just misunderstanding how this works.
In any case, I have some data that I want to store and want to reliably archive, and I don’t want the docker apps to have too much control over it. That’s why I went with the current approach. It has also allowed for very granular control. But it’s also a bit more cumbersome, as everytime I want to selfhost a new app, I need to configure datasets, permissions and mounting of NFS shares.
Is there a simpler approach to all this? Or should I just buy an HBA and continue with things as they are? If so, which one should I buy (considering a very limited budget)?
I’m thankful for any advice you can give and for your time. Have a nice day!
You should have all your data separately stored, it shouldn’t be locked inside containers, and using a VM hosted on a device to serve the data is a little convoluted
I personally don’t like TrueNAS - I’m not a hater, it just doesn’t float my boat (but I suspect someone will rage-downvote me 😉)
So, as an alternative approach, have a look at OpenMediaVault
It’s basically a Debian based NAS designed for DIY systems, which serves the local drives but it also has docker on, so feels like it might be a better fit for you.
I tried OMV in the past, but I found TrueNAS to be more intuitive… but that’s just personal preference I guess, and I’m not opposed to using OMV. Are you suggesting, then, that I run OMV on bare metal, and use it for everything? Or should it be inside a VM? If it’s the former, how easy is it to setup docker, because I’m not that familiar with OMV (it’s been a long time since I last checked it out). Is it like installing it in Debian directly? How does it handle the storage?
I always prefer bare metal for the core NAS functionality. There’s no benefit in adding a hypervisor layer just to create an NFS / SMB / iSCSI share
OMV comes with it’s own bare metal installer, based on Debian, so it’s as stable as a rock.
If you’ve used it before, you’re probably aware that it needs it’s own drive to install on, then everything else is the bulk storage pool… I’ve used various USB / mSATA / M.2 drives over the years and found it’s a really good way to segregate things.
I stopped using OMV when - IMO - “core” functions I was using (ie syncthing) became containers, because I have no use for that level of abstraction (but it’s less work for the OMV dev to maintain addons, so fair enough)
So, you don’t have to install docker, OMV automatically handles it for you.
How much OMV’s moved on, I don’t know, but I thought it would simplify your setup.
Okay, thank you, that’s good to know. However, I don’t have two separate devices that I can use to separate the NAS functionality from the Docker functionality, that’s why I was using Proxmox in the first place. And, I’m not sure how well Docker can run in OMV. But I’ll still keep it in mind as an option, thank you!
I think you’ve misunderstood
Ok, OMV needs a separate (small) boot drive to install on (ie consider a M.2 / SSD on a USB adapter)
But, then all your (large) storage is used for the NAS.
OMV will run Docker containers, but their data would also be pointed to the large NAS storage.
| Small | Large | |--------+-----------| | OMV | Your Files| | Docker | Data, etc |You’re right, I misunderstood. I understand now, thank you for replying with detail. I’m currently still not over my “I like Hypervisors now” phase. If I go back to bare metal, I will most probably use the setup you described. Still, thank you very much, and I’m keeping this thread for future reference.
you can install dockge in truenas and then all your docker data is not “locked” inside their application data
Yeah, actually I had that setup for a brief amount of time, but I had issues because my system has too little RAM, and TrueNAS and Docker were constantly fighting over RAM, and the system hang frequently. For some mindboggling reason, if I ran TrueNAS and a Docker VM separately on Proxmox, were I could manually specify allowed RAM, everything worked perfectly fine, even though the available RAM was still very low. If I ever go down this road again, I’ll need to buy more RAM sticks.
If you’re only doing a VM or two, I’d get rid of proxmox and run truenas directly. It’s gotten better for VMs.
Also make sure you read up on the ecc requirements for truenas if you’re not using ecc ram
And is it easy for the docker instances inside the VM to access the host’s datasets? About ECC, thank you for bringing it up, because I actually have no idea on the subject, and I’m sure that my current ram isn’t ECC. I’ll look into it. It could explain the issue I had with gitea, right?
Send your question to the podcasters at 2.5admins, this seems right up their alley.
Thank you, I might raise the issue there, if I struggle to fix the issue. I didn’t know the podcasters, but now I’ll try and listen to a few episodes, and maybe I can continue learning. Thank you for the suggestion
They always do Free consulting on air, so if your issue gets to them and they have a spot, you get a very very competant answer
They’re gonna say the same thing you’ve read here, which is that if you’re going to virtualize TrueNAS, pass through the controller, not just the disks.
I run a docker host in Proxmox using ZFS datasets for the VM storage for things like my mailserver and NexcloudAIO. When I backup the docker VM, it snapshots the VM at a point in time, and backs up the snapshot to PBS. I’ve restored from that backup and it’s like the machine had just shut down as far as the data is concerned. It journals itself back to a consistent state and no data loss.
I wouldn’t run TrueNAS at all because I have no idea how that’s managing it’s storage and wouldn’t trust the result.
Wait, so if I understood correctly, you’re managing the ZFS pools directly in Proxmox, and then you have a VM that’s running docker, and using the storage that is managed by Proxmox, right? Hmm, sounds like a good solution. Is there any documentation or article that you could recommend, so that I can take a closer look? Also, how could I handle SMB shares?
Yes. So my debian docker host has some datasets attached:
mounted via fstab:
and I specify that path as the datadir for NCAIO:
Then when PBS calls a backup of that VM, all the datasets that Proxmox is managing for that backup take a snapshot, and that’s what’s backed up to PBS. Since it’s a snapshot, I can backup hourly if I want, and PBS dedups so the backups aren’t using a lot of space.
Other docker containers might have a mount that’s used as a bind mount inside the compose.yml to supply data storage.
Also, I have more than one backup job running on PBS so I have multiple backups, including on removable USB drives that I swap out (I restart the PBS server to change drives so it automounts the ZFS volumes on those removable drives and is ready for the next backup).
You could mount ZFS datasets you create in Proxmox as SMB shares in a sharing VM, and it would be handled the same.
As for documentation, I’ve never really seen any done this way but it seems to work. I’ve done restores of entire container stacks this way, as well as walked the backups to individually restore files from PBS.
If you try it and have any questions, ping me.
Wow, that’s awesome. I think that’s actually the approach I’m going to go for. This way I don’t need to buy hardware, and I don’t need to work with TrueNAS anymore.
Where you talk about “walking the backups”, do you mean that you can actually see the entire file structure of the container? I mean, I don’t know how virtual disks are stored on the dataset. Like, as far as I know, a VM virtualized disk is just a file, right? So you’d have a ZFS dataset with a single file, for example? Could you then try and navigate the files inside this VM disk file, without the VM? Or did I misunderstand, and you’re mounting the dataset, somehow, directly inside the VM? Is that like a passthrough for datasets?
In any case, thank you for sharing so much information and for offering help. I may take you up on that, as it seems that this is the approach that I feel most comfortable with.
So if I want a new container stack, I make a new Proxmox “disk” in the ZFS filesystem under the Hardware tab of the VM. This adds a “disk” to the VM when I reboot the VM (there are ways of refreshing the block devices online, but this is easier). I find the new block device and mount it in the VM at a subfolder of /stacks, which will be the new container stack location. I also add this mount point to fstab.
So now I have a mounted volume at /stacks/container-name. I put a docker-compose.yml in there and all data that the stack will use will be subfolders of that folder with bind mounts in the compose file. When I back up, that ZFS dataset that contains everything in that compose stack is snapshotted and backed up as a point-in-time. If that stack has a postgres database, it and all the data it references is internally consistent because it was snapshotted before backup. If I restore the entire folder from backup, it just thinks it had a power outage, replays it’s journals in the database, and all’s well.
So when you have a backup in PBS, from your Proxmox node you can access the backups via the filesystem browser on the left.
When you go to that backup, you can choose to do a File Restore instead of restoring the entire VM. Here I am walking the storage for my nextcloud data within the backups, and I can walk this storage for all discrete backups.
If I want to just restore a container, I will download that “partition” and transfer it to the docker VM. Down the container stack in question, blow out everything in that folder and then restore the contents of the download to the container folder. Start up the docker stack for that folder and it’s back to where it was. Alternatively, I could just restore individual files if I wanted.
I have two identical HDDs as a mirror, another one that has no failsafe (but it’s fine, because the data it contains is non-critical)
On separate pools, I hope? My understanding of ZFS is that the loss of any vdev will mean the loss of the pool, so your striped vdev should be in its own pool that you don’t mind losing.
Correct, they’re two separate pools, and the stripe one doesn’t contain any valuable data.
