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!
Proxmox supports ZFS natively with management in the WebUI. So you could get rid of TrueNAS entirely and not need to deal with HBA pass-through or anything.
You also wouldn’t need NFS or have to deal with shares, as the data is available directly to Proxmox Containers via bind mounts.
Okay, if Proxmox can handle all that, I’ll be glad to ditch TrueNAS. However, I’m afraid that I won’t know how to migrate. I’ve found this reddit thread about someone who tried to do the same thing (I think) and accidentally corrupted their pools. About skipping NFS shares, that would be a big improvement for me, but I’m very unfamiliar with bind mounts. If I understand correctly, you can specify directories that live on the Proxmox Host, and they appear inside the VM, right? How does this compare to using virtual storage? Also, how can I replicate the ZFS pools to an external machine? In any case, thank you for that info!
Migration should be as simple as importing the existing ZFS stuff into the Proxmox OS. Having backups of important data is critical though.
If I understand correctly, you can specify directories that live on the Proxmox Host, and they appear inside the VM, right?
Inside a Container, yep. VMs can’t do bind mounts, and would need to use NFS to share existing data from the host to inside the VM.
How does this compare to using virtual storage?
Like a VM virtual disk? Those are exclusive to each VM and can’t be shared, so if you want multiple VMs to access the same data then NFS would be needed.
But containers with bind mounts don’t have that limitation and multiple containers can access the same data (such as media).
Also, how can I replicate the ZFS pools to an external machine?
ZFS replication would do that.
Like a VM virtual disk? Those are exclusive to each VM and can’t be shared, so if you want multiple VMs to access the same data then NFS would be needed.
But containers with bind mounts don’t have that limitation and multiple containers can access the same data (such as media).
Just to be clear, are you saying that when you’re using bind-mounted ZFS pools, it’s okay to write from two containers (or both the proxmox host and a container) at the same time?
Also, I think I managed to accomplish that for a VM by creating a Proxmox Directory pointing to a path in a zpool, adding it to the VM using virtiofs, and mounting it within the VM. I’m not sure if writes from both the VM and the host are safe in that case either, though.
You wouldn’t want to write to the same file at the same time, but otherwise it works fine.
Okay, I have one of the pools that is pretty empty and has non-critical data, and I think I’ll try migrating that first, and see if it’s imported correctly by Proxmox.
About Containers, I think I’ll have to do some more research because I think I haven’t fully understood yet how they compare to VMs. Or like, when I should use the one over the other. I guess I could have a Container with a bind mount to a dataset that I want to be able to share over NFS or SMB, and handle that from whatever OS I put in the Container, right? But, I could also have a VM do that, and though it wouldn’t be able to share the data with other VMs, it can do it over NFS, can’t it? What are the advantages of doing one thing over the other?
Well, in any case, thank you for your patience, for going over each detail and taking the time to correct me where I’m wrong. I’m learning a lot, so thank you!
Edit: fixing grammar
Essentially a container shares the kernel of the host, so uses less resources to run.
VMs are useful when you need more isolation or a different kernel (or need to add kernel modules).
For most purposes containers are the easy option.
I guess I could have a Container with a bind mount to a dataset that I want to be able to share over NFS or SMB, and handle that from whatever OS I put in the Container, right?
Yep!
But, I could also have a VM do that, and though it wouldn’t be able to share the data with other VMs, it can do it over NFS, can’t it?
Also yes, just a more complex setup with more performance penalty due to using NFS to share data into the VM.
I see, okay, I’ll try out containers then. So far, I’ve been able to migrate a ZFS Pool without issues, so I’ll start migrating them all, create a container that manages NFS and see if the existing Docker VM picks up the NFS shares successfully. Thank you for going in-depth and explaining everything to me. I’ve learnt a lot!
deleted by creator
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.
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.
