#iGPU setup swap memory to 100% concern

1 messages · Page 1 of 1 (latest)

upbeat sandal
#

Every time i try to process a large amt of images, the swap usage goes to 100%

My setup is an intel n95 with the openvino docker image selected. Since the igpu doesnt have its own ram, it has to use a lot of system ram I am sure. RAM looks under control but the swap just climbs till its full

Is this expected or a possible issue?

image.png
wet vineBOT
#

:wave: Hey @upbeat sandal,

Thanks for reaching out to us. Please carefully read this message and follow the recommended actions. This will help us be more effective in our support effort and leave more time for building Immich immich.

References

#

Checklist

I have...

  1. :ballot_box_with_check: verified I'm on the latest release(note that mobile app releases may take some time).
  2. :ballot_box_with_check: read applicable release notes.
  3. :ballot_box_with_check: reviewed the FAQs for known issues.
  4. :ballot_box_with_check: reviewed Github for known issues.
  5. :ballot_box_with_check: tried accessing Immich via local ip (without a custom reverse proxy).
  6. :ballot_box_with_check: uploaded the relevant information (see below).
  7. :ballot_box_with_check: tried an incognito window, disabled extensions, cleared mobile app cache, logged out and back in, different browsers, etc. as applicable

(an item can be marked as "complete" by reacting with the appropriate number)

Information

In order to be able to effectively help you, we need you to provide clear information to show what the problem is. The exact details needed vary per case, but here is a list of things to consider:

  • Your docker-compose.yml and .env files.
  • Logs from all the containers and their status (see above).
  • All the troubleshooting steps you've tried so far.
  • Any recent changes you've made to Immich or your system.
  • Details about your system (both software/OS and hardware).
  • Details about your storage (filesystems, type of disks, output of commands like fdisk -l and df -h).
  • The version of the Immich server, mobile app, and other relevant pieces.
  • Any other information that you think might be relevant.

Please paste files and logs with proper code formatting, and especially avoid blurry screenshots.
Without the right information we can't work out what the problem is. Help us help you ;)

If this ticket can be closed you can use the /close command, and re-open it later if needed.

wet vineBOT
blissful falcon
#

looks like you're maxing out your system ram of 8gb. I needed 16gb to get openvino to run with my intel iGPU without using swap. Swap gets used when ram is full

upbeat sandal
#

I have 10 for that LXC, its using 8 then leveled off. Anyways - so you think it best to not use the openvino image for the immich machinelearning container?

blissful falcon
#

depends, I use it because it makes my textual searches faster (smart search). It could also be that there's a supposed memory leak with the openvino ML container - a rumor I've heard but haven't confirmed myself. It sounded like an upsteam issue and not related to immich

upbeat sandal
#

I see. Yeah, I would like to keep it on, and that swap accumulation sure looks sus

blissful falcon
#

perhaps it's the ocr job having issues?

#

I'm running it on a intel j4125 with 18gb ram. It runs great for everything except ocr (im using the server model) - I needed to build a bit of a hack to get the ocr to offload to my more powerful remote machine

upbeat sandal
#

This happens during OCR on mine. Its an N95

blissful falcon
#

if you don't notice a difference in search speed, I'd probably turn it off for awhile and check back later to see if its been resolved

upbeat sandal
#

that is a good idea. sold

#

just use the latest tag on the machine learning container instead?

blissful falcon
#

my cpu is noticeably slower than my gpu for search - so that wasn't an option for me

blissful falcon
upbeat sandal
blissful falcon
#

optionally you could just comment out the ml file, which would cause it to fail loading, and then fallback to cpu

#

this would be super easy to uncomment in the future

blissful falcon
upbeat sandal
#

i actually dont have that env attribute

    container_name: immich_machine_learning
    #image: ghcr.io/immich-app/immich-machine-learning:release-openvino
    image: ghcr.io/immich-app/immich-machine-learning:latest
    volumes:
      - model-cache:/cache
    environment:
      # Tells the ML service to use the iGPU via OpenVINO
      - IMMICH_DEVICE=openvino
      - MACHINE_LEARNING_MODEL_TTL=30

    devices:
      - /dev/dri:/dev/dri```
blissful falcon
# 
    container_name: immich_machine_learning
    #image: ghcr.io/immich-app/immich-machine-learning:release-openvino
    image: ghcr.io/immich-app/immich-machine-learning:release
    volumes:
      - model-cache:/cache
    environment:
      # Tells the ML service to use the iGPU via OpenVINO
      - IMMICH_DEVICE=openvino
      - MACHINE_LEARNING_MODEL_TTL=30

    devices:
      - /dev/dri:/dev/dri```
#

any reason the release portion isn't defined in your env file?

#

I'd likely suggest making them v2 so that it'll keep working and you won't need to worry about breaking changes when updating. When v3 gets released you'd need to change v2 to v3 while following breaking changes to ensure it continues to operate

upbeat sandal
#

hmm no I am just trying to figure out what the tag should be now - I see in the example docker compose its

image: ghcr.io/immich-app/immich-machine-learning:${IMMICH_VERSION:-release}
#

then IMMICH_VERSION=v2.5.6 can be an env variable, i suppose

blissful falcon
upbeat sandal
#

alright sweet