#Utilize GPU on another machine for machine learning

1 messages · Page 1 of 1 (latest)

silver saffron
#

I was following this video https://www.youtube.com/watch?v=QHWNu_in0Zc

I thought I had set everything up as instructed but when I enter tailscale ip address i'm unable to connect.

i'm using docker desktop on my main windows machine which is currently housing immich_machine_learning and tailscale

i'm not familiar at all with how this is supposed to work, I simply tried to follow the video above

In today's video we explore sharing an NVIDIA GPU from Windows 11, running a containerized workload with docker for windows, over Tailscale to connect Immich - a self-hosted photo library - to hardware accelerate machine learning tasks.

Personal accounts are always free on Tailscale and can include up to 3 users and 100 devices. Get started tod...

▶ Play video
#

i'm sure this has something to do with how i set up tailscale but i'm so lost, i have no idea what it could be lol

warped coral
#

It looks like you pulled a tailscale docker image?
What the video is suggesting you install tailscale natively to your PC, then use the IP address assigned to your device to connect to the immich ML port.

i.e., if your tailscale ip for your computer with the GPU is 100.10.123.12, in immich you should set the address for ML as 100.10.123.12:3003

warped coral
silver saffron
#

ohhh so need to install tailscale directly on my pc, not in docker

silver saffron
#

ok, i've installed tailscale on my winpc but i'm still not able to connect

#

in the video he sends the sharelink to the computer that runs immich, in my case, it's running in container station, how in the world would i be able to share my tailscale if all that's there is immich?

warped coral
#

https://tailscale.com/qnap

You should be able to connect your qnap to your tail scale network

Tailscale for QNAP NAS. Securely access your QNAP NAS from anywhere in the world, on any device.

silver saffron
#

yeah, i still can't figure it out

#

i think i just need to figure out how to expose port 3003 in my windows docker

#

i tried to forward that port on my router but no go lol

warped coral
silver saffron
#

ahhh kk, let me just delete that

warped coral
silver saffron
#

my apologies but i honestly have no idea how to do that - please excuse my ignorance, still have a lot to learn

warped coral
#

All good.

Is your QNAP and desktop on the same lan? Do you have access to the terminal on your qnap?

silver saffron
#

yes, it's on the same lan

#

i can definitely putty into the qnap

warped coral
#

Alright, you don't need tailscale at all then. Since you asked a question about tailscale, I assumed your QNAP was in a different lan altogether.

#

So on your qnap terminal, try if you can ping your desktop.
ping 192.168.xx.xx, replace the IP with your desktop's IP address

silver saffron
#

sure thing

warped coral
#

Can you post the output?

silver saffron
#

looks like i cannot

#

i get a "permission denied (are you root?)"

#

if i sudo ping, i get pings

#

ummm.. how do i stop it from pinging? lol

devout hedge
silver saffron
silver saffron
warped coral
# warped coral e.g.,

In your immich, set the ML address to http://192.168.88.44:3003, like you see in this screenshot.

silver saffron
#

i guess at least it's not getting a connecting error? it's something else entirely

#

GPU passthrough for Machine learning

silver saffron
#

it seems i can see the container on a different machine within LAN as well

warped coral
#

Also, GPU passthrough means something else entirely...

silver saffron
#

oh sorry, i just wanted to name it something more appropiate

#

no wonder i was having such a hard time finding info lol

#

looks like it's working?!