ProxMox[TrueNAS(VM) + JellyFin(VM/LXC)] or [TrueNAS+JF plugin]

[trentmu]

I have a JellyFin LXC which accesses shared storage (from the proxmox node) which works for my needs.

I am planning to add a "new" (refurb) machine with much more storage - 1 HBA [with 12 drives (for ZFS vdevs) + 2 SSDs (for boot/OS)].
I will configure the SSDs as a mirrored boot/OS pair, and then either

• 3 vdevs of raidz1(3+1)
  • to get better performance
• or
• 2 vdevs of raidz2(4+2)
  • to get better redundancy

Is it a better option to use

• TrueNAS Scale
  • directly controlling the HBA etc, with a JF plugin
• OMV
  • directly controlling the HBA etc, with a JF plugin
• proxmox
  • controlling the ZFS/drives with a JF LXC
• proxmox
  • TrueNAS or OMV as a VM with shared CIFS
  • JellyFin as a VM accessing the CIFS share from TrueNAS VM

Specifically the last option doesn't "feel" right - I can't passthough the full HBA to the VM since I need it to boot the proxmox OS - unless there's a better way?

 

[Dunuin]

I would PCI passthrough the HBA into the TrueNAS VM, create the ZFS pools there and then shrem them though NFS/SMB.
Then I would create a Jellyfin guest on the PVE host as an unprivileged LXC or VM and mount your NFS/SMB shares there to access your media.

Specifically the last option doesn't "feel" right - I can't passthough the full HBA to the VM since I need it to boot the proxmox OS - unless there's a better way?

Then option 1 and 2 wont work as you can only passthrough the whole HBA with all of its ports. Or do you mean installing TrueNAs Scale/OMV bare metal instead of PVE?

 

[trentmu]

I would PCI passthrough the HBA into the TrueNAS VM, create the ZFS pools there and then shrem them though NFS/SMB.
Then I would create a Jellyfin guest on the PVE host as an unprivileged LXC or VM and mount your NFS/SMB shares there to access your media.

Then option 1 and 2 wont work as you can only passthrough the whole HBA with all of its ports. Or do you mean installing TrueNAs Scale/OMV bare metal instead of PVE?

For 1 & 2 I was implying a baremetal TN/OMV instead of PVE - but I am really enjoying proxmox so would prefer to stay with it if possible

 

[wbravin]

Hello all

I'm using this thread because i would have the same question.

I currently have a proxmox 8.2.2 server with 3 vms and 2 LXC and all works well (until i screw it up).

Currently I have on this server jellyfin installed as an app within truenas scale . again it works fine

I have a second server that has truenas installed bare metal. I use this server as a backup to my main server

This setup is working fine and fills mostly all my need

Because i have time on my hands and i always like to learn new things, i would like to get a 3rd server to create a high availability environment

Therefore, if i understand things as they are written hear.

I could install proxmox on the new server create the same vms. I would then add in a new LXC jellyfin and when installing jellyfin i just add the libraries from my truenas datasets with the same credentials.

If so this would be very simple

But is it better (more efficient ,secure and/or less resource intensive)than my original setup on server 1?
I'm I understanding this thread correctly?

Thank you for your patience and help

 

[Dunuin]

For HA you need at least 2 PVE nodes + a third machine (could be a VM on the bare metal TrueNAS) as a qdevice. Then a shared filesystem like NFS (problem with single point of failure) or ceph (needs 3+ nodes) or ZFS replication (ZFS used on at least two PVE nodes). And you wouldn`t create new copies of your guests. There is only one VM/LXC and that is shared or gets replicated.

See here for the cluster requirements and basics: https://pve.proxmox.com/wiki/Cluster_Manager

 

[wbravin]

For HA you need at least 2 PVE nodes + a third machine (could be a VM on the bare metal TrueNAS) as a qdevice

Thank you for your reply
Yes . Currently i have a dell r720 as my first pve. I plan to purchase a dell r730 for building my HA. I will then have my first nas (a A supermicro built nas is currently running truenas on bare metal)

Then a shared filesystem like NFS (problem with single point of failure) or ceph (needs 3+ nodes) or ZFS replication (ZFS used on at least two PVE nodes). And you wouldn`t create new copies of your guests. There is only one VM/LXC and that is shared or gets replicated.

Yes I understand

Currently, my dell is backed up every Saturday to my supermicro

See here for the cluster requirements and basics: https://pve.proxmox.com/wiki/Cluster_Manager

Once again thank you for the teaching