Of course you can. Traefik can proxy almost anything. Check out my traefik guide. I talk about adding external apps there. Or hop on to our discord and there are many who can help you out.

My dev environment is lxc. So it definitely works. What errors? Join the discord for help

Request fulfilled. This series is ongoing. First one out last Sunday. Check it out.

@@AnandsLab thanks! I did see it. I figured it out with other resources but I'll def be watching the series.

Thank you!

Yup I was right, I had to assign a user that was in group of the card0 and renderD128. That allowed transcoding in docker. Now if you could just help me figure out handbrake! 😁

nevermind, that didn't work a second time

Absolutely.

In that case, you have two options. But first try this: Install acl on host and lxc. Then on the host run sudo setfacl -Rm u:101000:rw /dev/dri. Now if you check inside the LXC the user 1000 should have rw access. Check your user id inside the lxc with id command. If it is other than 1000 then you will have to adjust the setfacl command. The second option is to change ownership of render128 on proxmox host. But i prefer the first.

With an Unpriviledged LXC, the way I get around that is to do some user and group mappings so things look right and you can assign groups to the (jellyfin) user to use the hardware correctly. First you will need to Find Device Numbers ls -l /dev/dri In My case with an Intel APU the device numbers are 226 and 128. Process is the same for AMD iGPUs Now Find the group numbers for the group(s) for these devices (renderer and video) cat /etc/group Mine are 44 and 104. Now you need to add these groups to the SubGroup file in the host. nano /etc/subgid Add at the bottom of the file as below. root:100000:65536 root:44:1 root:104:1 In the above I also added the 100000 group so root maps correctly too. We are going to have to do some user and group remapping so the devices come through to the LXC correctly. This is done in the .conf file for your container. I will just past as an example what I do. These lines are just for the devices and group mappings. lxc.cgroup2.devices.allow: c 226:0 rwm lxc.cgroup2.devices.allow: c 226:128 rwm lxc.mount.entry: /dev/dri/renderD128 dev/dri/renderD128 none bind,optional,create=file lxc.idmap: u 0 100000 65536 lxc.idmap: g 0 100000 44 lxc.idmap: g 44 44 1 lxc.idmap: g 45 100045 62 lxc.idmap: g 107 104 1 lxc.idmap: g 108 100108 65428 The above mappings will map the devices and groups into your LXC Container. After starting your container check /dev/dri sith an ls -l to ensure the groups mapped correctly. If so, add these groups to the jellyfin user and restart jellyfin. Now your hardware render will work You can also pass in an Nvidea GPU but it is a bit more involved. Hope this is helpful.

I had to add these lines to my docker compose file devices: - /dev/dri:/dev/dri

@@nickmichniewicz3519 of course I thought that was obvious but you have a good point.

@@nickmichniewicz3519 That is one way to do it, passing dri as a device. In my opinion the clearner way is to is to mount the actual device(s) you need and not entire structure. This works with both Inel and AMD APUs. Nvidea cards are another story. lxc.mount.entry: /dev/dri/renderD128 dev/dri/renderD128 none bind,optional,create=file lxc.mount.entry: /dev/dri/card0 dev/dri/card0 none bind,optional,create=file But like I always say. There is usually more than one way to do what we do and if it works, it is not wrong.

don't worry I reviewed your video again saw you mention enabling the api.insecure=true and it's associated ports. all working now!

iGPU passthrough to a VM is a bit more tricky. I am not doing this yet as I still need the iGPU available on host, which won't be the case if i pass it through. But out of interest I will try to cover this in a separate video.

@@AnandsLabi know that for older versions of Intel cpus you could have both host video and also pass the gpu to a vm.

@@Sherbibv I am not saying it can't happen but I have never seen it be successful due to passing the iGPU to a VM as a hardware device. This gives the VM full control over the hardware. That is providing you tell the Proxmox host to not use said iGPU, otherwise it does not pass into the VM. But, you can pass an iGPU to many LXC containers without issue. I am sure there is a limit but I have had 6 seperate LXC containers running at the same time without issue. All using the iGPU for transcoding.

Oops! Adding it to the description.

Please elaborate to help others.

@@AnandsLab It's a hot mess of confusing overlays and hides how applications really work in a native environment. I can copy a LXC and VM system to bare metal and vice versa, but a docker environment totally messes that up unless it's docker to docker.

It was, kind of. I had planned it but moved it up in the schedule :-)

It definitely is. Just need to ensure good backups although I never had to restore one in 2 years

Already recorded. And I am passing through my igpu. Not a dedicated card.