n00b q's about shared storage, and app stores

[neek]

Hi, I'm in the process of evaluating migrating my homelab from FreeNAS/ZFS/NFS-share to Proxmox 5.x. I just got Proxmox 5.2.10 running on a new system and have some initial questions. I'm new to proxmox but i'm very experienced in linux/systems/etc.

1. My first proxmox machine (lets call it pve1) has a lot of storage local to it. I'm trying to decide what the best way would be to eventually share the VM portion of that storage with a small cluster (maybe 2-3 other proxmox systems). Is there any recommendation? I did see the docs and wiki regarding storage but there's not really enough guidance there to make recommendations. I'd like to be able to snapshot the VMs, and in my current solution I do that by running ZFS shared via NFS, and then pausing the VMs when I run a ZFS snapshot. What's the recommended way to snapshot VMs in proxmox?

2. I would also like to host some file shares from that big pool of storage on this first system. Is the recommended way to do that to create a VM / container running samba + nfs and attaching it to the underlying storage?

3. If I do that, and later add other pve systems to the cluster, is there a way to enforce / encourage that particular vm/container to remain close to the storage?

4. Finally, since this is a homelab, I'd like to play around with e.g. Turnkey Linux images for a bunch of systems they've setup. The last news I see of any kind of integration is from ~2016, and talks about Proxmox v2.0. Is there any support in Proxmox for adding an appliance app store or anything similar?

thanks in advance!

 

[LnxBil]

What's the recommended way to snapshot VMs in proxmox?

With backend storage support which you lack in a NFS setup from another machine. Sharing ZFS over iSCSI is work in progress for Linux and FreeNAS, but supported with some Solaris-based backends:

https://pve.proxmox.com/wiki/Storage:_ZFS_over_iSCSI

Is the recommended way to do that to create a VM / container running samba + nfs and attaching it to the underlying storage?

Yes, separation of concerns for everything. PVE does the raw storage consistency and redundancy and you will provide file services in a VM from that pool.

3. If I do that, and later add other pve systems to the cluster, is there a way to enforce / encourage that particular vm/container to remain close to the storage?

First: Do not run shared local storage in a clustered environment, you just create a single point of failure, but back to the question:
You have to have it mounted via NFS on all nodes (also on the one with your storage itself) so that you can migrate if to other nodes. This adds additional and redundant layers. There is currently (and most probably never) such a "feature" like prefer local storage in such a setup, because the design itself is kinda broken. If you want to share stuff, make it redundant via CEPH or an external storage pool.

4. Finally, since this is a homelab, I'd like to play around with e.g. Turnkey Linux images for a bunch of systems they've setup. The last news I see of any kind of integration is from ~2016, and talks about Proxmox v2.0. Is there any support in Proxmox for adding an appliance app store or anything similar?

There are Turnkey images available directly in PVE, but I do not know if that is a complete set or not. Please run:

pveam update
pveam available

 

[neek]

Do not run shared local storage in a clustered environment, you just create a single point of failure

Thank you, yes, I agree this is a risk but in a homelab, it's a compromise I will have to make (I do replicate to an offsite server and to the cloud). Having multiple physical compute servers also lets me get more net CPU horsepower in my cluster without needing replicated storage.

I did find the images in 'pveam available', thank you. I guess I'm just used to FreeNAS after several years having that be my main system.. It housed the storage, and the shares, and some of the compute tasks (mostly through FreeBSD jails which are similar to linux LXC) when I needed them close to the storage. Anyway, thank you, this is a big help in getting my arms around this new system.