Ok, running into hw transcoding issues with Plex now. Any ideas?
not sure about transcoding, but i got the nvidia 1050 gtx to work as machine learning acceleration for immich. so it does work at least for that i can confirm. all i did was in jlmkr config for the docker jail, add the nvidia bind entry. and install the nvidia container toolkit. then for immich docker compose modify to include the parts for enabling the graphics card for machine learning.
didn’t try setting for jellyfin live transcoding because too many steps and it wasn’t that important for me.
You had to install the nvidia container toolkit inside the container?
yep thats what i did. didnt work until i did that.
The NVIDIA Container Toolkit enables users to build and run GPU-accelerated containers. The toolkit includes a container runtime [library] and utilities to automatically configure containers to leverage NVIDIA GPUs.
*named my jail docker
jlmkr shell docker
then proceeded with the steps
https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/latest/install-guide.html
installed via apt method. had to first install some stuff like curl, gpg first before i could run the command. for apt obviously i updated apt first before install.
so the first command was to add the nvidia repo. after that came installing the toolkit.
then the immich portion of the setup in docker compose and the files for it
For jellyfin which i did not do, these are the steps for that
2 Likes
Yeah, I installed the toolkit, restarted the container and Plex still doesn’t recognize the GPU. I checked nvidia-smi and it’s showing no processes using the GPU when it’s trancoding live TV. 
did you make sure to add this in your docker jail config?
`--bind-ro=/usr/lib/x86_64-linux-gnu/nvidia/current`
that’s the only bind i added for nvidia into the config.
afaik that is the only part jailmaker related (modifying the config to add the nvidia bind), other than installing any required app (in my case had to install the toolkit for the nvidia machine learning to work).
Rest being the docker container specific instructions to modify to get that working for the rest (the stuff i previously mentioned).
So check you did the jailmaker portion side of things, then check on the docker side of things. Since you are using plex, refer to their transcoding nvidia graphics card section for that.
For graphic card drivers, if not mistaken that was already built into the truenas release correct me if i am wrong 
2 Likes
Yes, see here for how it’s configured. I just installed the toolkit, then ran their configure command. After that, I restarted the jail.
Issue resolved. I was missing the NVIDIA_* env vars and adding the gpu capabilities to the container. See here for further details.
2 Likes
Found this for you
https://www.reddit.com/r/PleX/comments/virmxi/guide_plex_transcoding_with_docker_nvidia_gpu/
awesome 
yes for troubleshoot i basically went through the checklist of things you have to set. A whole back and forth making sure you didn’t miss a step or did something incorrectly.
To make sure the jailmaker stuff was correct, i enter jlmkr shell docker
then tested if i could detect the nvidia graphics card. These were some of the commands i used to confirm i could interact with the nvidia graphics card from within the docker jail
if that worked, then i moved onto the other steps.
so glad you figured that out 
1 Like
I’ll do a PR and create a GPU document on the repo and add docker config preconfigured for GPU with Nvidia.
1 Like
@Jip-Hop ^ is that ok? Also, is it ok if I create a compose or stacks folder with directories for applications that will work with jailmaker to make it easier for others coming in to get going? At least it can be hosted there for now, then moved wherever in the future.
Thank you for pointing me in the right direction 
1 Like
I don’t see why compose files need to be in the jailmaker repo. Once docker is installed in a jailmaker jail all the compose stuff is generic docker compose. The whole idea is that your compose stack is portable, so you could spin up the stack in a jail, vm or even on a Synology NAS with docker.
But a docker-nvidia config template could be useful. Then you could include a docker run or simple compose file as an example of how to utilize the GPU with docker in the jail.
2 Likes
Ok, I’ll get something going and submit a PR.
I’ll create a separate repo with configs with compose files known to work with jailmaker and docker compose.
Yes, they are generic, but at that l the same time are proven to work with jailmaker and nested docker and will also give newcomers to the project quick and easy ways to get started with some of the more popular apps.
What do you think? Could that be linked in a document in jailmaker then as examples to get started? Thank you for the consideration.
Sure!
1 Like
Ok, here is what I came up with in regards to templates that work with jailmaker. Let me know what you think. I can link to that from the PR I’ll create to add the GPU docs.
2 Likes
Nice new video @Stux! I have added it to the networking docs on the develop branch.
By the way I still have this TODO in the readme.
TODO: write comparison between systemd-nspawn (without
jailmaker), LXC, VMs, Docker (on the host).
Perhaps this could be fixed with a comparison video? ^^ Comparing with Apps would be interesting too.
Anyhow looking forward to the next one!
1 Like
Next one will be about Snapshots I think, will add the comparison to the ideas list 
3 Likes
It’s threads like these that make you face-palm with disappointment at projects like XigmaNAS, whose forums are private and inaccessible to the public.
Who does that benefit? 
1 Like