I want to use Jellyfin on Proxmox, if that is a thing. After reading a post here where most people recommended Debian as host OS I want to make a VM running Debian and install Jellyfin Server there.

Now I have a few questions:

  • I see many people install Jellyfin via docker. Does that have any advantages? I would prefer to avoid docker as it adds a level of complexity for me.

  • where do I save my media? I have a loose plan to run a second VM running openMediaVault where all my HDDs are passed through and then use NFS to mount a folder on the Jellyfin VM. Is that a sane path?

  • what do I have to consider on Proxmox, to get the best hardware results on Jellyfin? Do I need some special passthrough magic to get it running smoothly? I don’t have a dedicated GPU, does that make the configuration easier?

  • CouncilOfFriends@slrpnk.netEnglish
    1·
    1 year ago

    I forget the specific issue I had trying the LXC route, but the easiest one for me was Docker in a VM. This LinuxServer.io guide really helped me. I’ve been meaning to move from Docker to Podman so I can learn how that works and get the security benefits, but the existing setup works and there were a few stumbling blocks I don’t feel like dealing with again. For example, and in case any other newbies like me run into these:

    • First time I [passed through my disks](https://pve.proxmox.com/wiki/Passthrough_Physical_Disk_to_Virtual_Machine_(VM) to the VM from the Proxmox host I didn’t document well what commands I had to run. When I later backed up the VM and then deleted to restore from backup I had to go look up device IDs again so made sure to save it. Now after a restore I only have to remember to run one command to give it the drives:

    qm set 101 -scsi1 /dev/disk/by-id/ && qm set 101 -scsi2 /dev/disk/by-id/ && qm set 101 -scsi3 /dev/disk/by-id/

    • Changing docker storage location so your VM doesn’t fill with docker data.

    • I like to specify a docker folder, with a config folder for each service. With my frankenstein server of mismatched drives pooled via MergerFS however, I found writes for my merged path ( /mnt/storage ) were slower than when using a specific drive’s path (such as /mnt/data2). Giving paths on data2 for containers to store configs, and data1 for qBittorrent and NZBGet gave me the best download speeds. For the same reason I specify a data2 path for Jellyfin and Plex to use for transcoding and reads use the merged path. That way completed downloads are moved to whichever data drive has the most free space