Pulling model in Models section spawns a never ending stream of Pulling Manfiest notifications that animated down screen about ollama-webui

[tjbck]

Seems like I've accidentally closed this issue, Just updated Ollama to the latest version and seeing this issue as well. It will be fixed in a bit. Thanks for bringing this up.

[kevinduffey]

Right on. Glad to help.

[tjbck]

Hi all, The issue should be fixed now with this merge, Please let me if the problem persists. Thanks!

[sanasol]

Nvm, I see there is a lot of issues https://github.com/jmorganca/ollama/issues?q=stuck

Fix is restart and retry until it work :)

Or better to add ollama noprune flag, to make resume possible after stuck/restart

compose.yaml

version: '3.6'

services:
  ollama:
    # Uncomment below for GPU support
    # deploy:
    #   resources:
    #     reservations:
    #       devices:
    #         - driver: nvidia
    #           count: 1
    #           capabilities:
    #             - gpu
    volumes:
      - ollama:/root/.ollama
    # Uncomment below to expose Ollama API outside the container stack
    # ports:
    #   - 11434:11434
    container_name: ollama
    pull_policy: always
    tty: true
    restart: unless-stopped
    image: ollama/ollama:latest
+    environment:
+     OLLAMA_NOPRUNE: true

And then

docker compose restart && docker compose exec -it ollama ollama pull llama2:70b

until win

[tjbck]

Hi, This looks very strange, I'm not sure what would be the cause of this issue as I cannot reproduce your issue in any of my local machines (mac, linux, windows). Could you please try running Ollama and the Web UI in other machines if you have any and see if you can reproduce the issue yourself? Thanks.

[sanasol]

Same here

Tested on macbook and linux server.

Both times opened using Brave on MacOs.

[kevinduffey]

Not sure why it was closed so fast.. but I dont have another machine to run it on. Glad to see I am not alone with this. I suspect the team should look to do some more testing. At least the image actually pulls down once all that animated waterfall notifications stop. So the download does work. It's not a big big deal, but just an oddity I guess. At least it's somewhat cool looking.. what would be better is if they bounced when they got to the bottom of the screen before disappearing. Adding some physics to it would be slick!
If it helps I was using Chrome on Windows hitting my linux ip that was running ollama and ollama-webui. Both of those are running as docker containers in portainer.

[sanasol]

https://github.com/ollama-webui/ollama-webui/blob/main/src/lib/components/chat/SettingsModal.svelte#L134

There is literally while(true) and !data.status.includes('downloading')

But message is always "pulling {hash}" probably

[sanasol]

perfect

[sanasol]

But there is another problem for me, maybe some quickfix possible. Dont want to post issue for it.

Cannot download 70b llama(other models too sometimes), tried multiple times, got stuck at end.

Sometimes server process killed.

No CPU/RAM usage after that, like just stuck.