I recently installed Immich, following several of the tutorials that have been posted on the Internet.  It has been working fine, and have about 5100 photos and 200 small video files.

 

FCP warned that my docker.img file was over 80% (24GB out of 30GB), so I looked at my Container Size:

 

A quick question, and an ask  - Does 4 GB+ seem correct?  If you are running Immich, please post what Container Size is reporting for you.

 

I know Immich is still very much in Beta, and understand if they want to keep things in the Docker image vs mapping it out to an Appdata file (which potentially can change in the future).  For me, it really doesn't matter where the data is, as both my Docker image and my Appdata reside on the same 2 TB pool drive.  So increasing the image size is a non-issue.  I just want to know if what I am seeing is normal, as I don't want to be caught unaware if the image keeps growing.

 

Edited February 18, 20242 yr by ConnerVT

2.2Gb for me. 

Interesting.  The 2 GB difference in Writable - May I ask how many photos/videos you have in Immich, and are you using hardware transcoding or machine learning?  I currently have my Nvidia Quardro P400 (Cuda) working with it.

Edited February 18, 20242 yr by ConnerVT

I have about 100k photos/videos. I haven't figured out how to get the iGPU/quicksync to work yet so it's using software transcoding. ML is turned on for facial recognition but that just uses CPU I believe (I don't have any HW accelerators).  

Have you checked your paths? All generated data should be outside of the docker image right?

 

I don't think it is a path issue.  Immich doesn't map to many in the template.  Two for photos which are on my array in it's own share /mnt/user/immich/ (/photos and /import) and for Appdata (which only contains one folder - machine-learning) in my appdata folder with all of my other dockers.

 

Maybe it is hardware transcoding.  I'll need to poke around inside the container at some point, and see where all of this data is stored.

Hi Guys,

Sorry to jump on but I see yous both have immich working. Can I ask about the screenshots you have attached? I have immich running from the community apps and it has its own folder with the machine learning in the app data folder. When you ask about container size is this the same setup as mine? I kind of understand docker and containers but unraid implements them differently? I don’t think my ML is working even though I set it to cuda when configuring it and I just noticed that uploading from the webpage ain’t working either.

 

Im two minds of just deleting everything and starting again but I wouldn’t be too sure on how to setup a proper docker container in unraid, Most of the guides are working with docker.

 

Sorry for the rant and questions but you two seems to have it working as expected.

The size used in the Docker image and what is in the appdata folder are two different things.  My appdata folder (which is data external to the Docker image) only contains the machine learning folder.

 

I am set up with these three containers from CA: immich, PostgreSQL_Immich and redis (jj9987's container).  I assume you have a Nvidia card you are wishing to use, as you mentioned 'cuda'.  I have a Nvidia P400 which has been working in immich.  I didn't do very much special for the setup (pretty much followed the video I'll link below).

 

For the Nvidia card, make sure you have the Nvidia Driver plugin installed.  Then add:

• --runtime=nvidia in Extra Parameters
• cuda in MACHINE_LEARNING_GPU_ACCELERATION
• Your GPU ID in NVIDIA_VISIBLE_DEVICES (copy from the Nvida Plugin page, don't have any leading/trailing space characters
• all in NVIDIA_DRIVER_CAPABILITIES

 

 

Thanks for your quick reply, really appreciate it. So that’s the video I followed but I don’t have some options in the Immich install.

 

Ive just added the --runtime=nvidia in extra parameters. I must of missed that and I’ve set cuda in the MACHINE_LEARNING_GPU_ACCELERATION

 

however I don’t have the 

NVIDIA_VISIBLE_DEVICES Or NVIDIA_DRIVER_CAPABILITIES

 

are these something that have to be out in manually? 
 

thanks again 

 

7 minutes ago, Smokie said:

are these something that have to be out in manually? 

 

Yes.  At the bottom of the immich template, click on "+Add another Path, Port, Variable, Label or Device".  Follow the instructions for filling these out from the second post from the NVidia Driver support thread.

 

 

Thanks again. Seem to have it running but I’m getting some errors when I search or try to kick off the face recognition.

Im waiting on SpaceInvaders part 2 video and I can compare or blow it up and start again following him lol

 

I have face recognition working here.  You may want to check your Settings, try a few things, then go to Jobs and rerun ALL for Face Detection / Face Recognition (note: This will wipe out any previous set manual naming of people you have done).

 

I'm waiting to see Ed's 2nd Immich video as well.  Hoping he has something on setting up NPM to work with this.  I've tried a bunch of stuff, but no joy so far.  😒