Issue when deploying Wyoming container | Home Assistant | Page 1

minor escarp Nov 20, 2024, 10:52 PM

#

I am trying to run Wyoming whisper in a docker container because I am running HA core and add-ons do not work when I try to deploy the container I get this error:

2024-11-20T22:45:25.473858870Z huggingface_hub.errors.LocalEntryNotFoundError: Cannot find an appropriate cached snapshot folder for the specified revision on the local disk and outgoing traffic has been disabled. To enable repo look-ups and downloads online, pass 'local_files_only=False' as input.

my yaml is:

services:
whisper:
container_name: whisper
image: rhasspy/wyoming-whisper
command: --model medium-int8 --language en
volumes:
- ./whisper-data:/data
environment:
- TZ=US/Eastern
- local_files_only=False
- HF_HUB_CACHE=/tmp
restart: unless-stopped
ports:
- 10300:10300

piper:
container_name: piper
image: rhasspy/wyoming-piper
command: --voice en_US-lessac-medium
volumes:
- ./piper-data:/data
environment:
- TZ=US/Eastern
- local_files_only=False
- HF_HUB_CACHE=/tmp
restart: unless-stopped
ports:
- 10200:10200```

nova plover Nov 20, 2024, 11:20 PM

#

for wyoming/piper I was able to issue the following command on my Debian Linux server:docker run -it -d -p 10200:10200 -v /path/to/local/data:/data rhasspy/wyoming-piper --voice en_US-lessac-medium
and it started right up. The docker-compose.yaml file for wyoming/whisper is as follows. I am using the GPU so I had to install some certain Nvidia libs as show in the yaml file.

#

services:
  whisper:
    container_name: whisper
    image: rhasspy/wyoming-whisper:latest
    command: --model Systran/faster-distil-whisper-large-v3 --language en --beam-size 5 --device
 cuda
    volumes:
      - /home/whisper/whisper-data:/data
      - /usr/lib/x86_64-linux-gnu/libcudnn_ops_infer.so.8:/usr/lib/x86_64-linux-gnu/libcudnn_ops
_infer.so.8
      - /usr/lib/x86_64-linux-gnu/libcudnn_cnn_infer.so.8:/usr/lib/x86_64-linux-gnu/libcudnn_cnn
_infer.so.8
      - /usr/lib/x86_64-linux-gnu/libcublasLt.so.12:/usr/lib/x86_64-linux-gnu/libcublasLt.so.12
      - /usr/lib/x86_64-linux-gnu/libcublas.so.11:/usr/lib/x86_64-linux-gnu/libcublas.so.11
    restart: always
    ports:
      - 10300:10300
    runtime: nvidia
    deploy:
        resources:
          reservations:
            devices:
              - driver: nvidia
                count: 1
                capabilities: [gpu]

minor escarp Nov 20, 2024, 11:30 PM

#

hm let me try

minor escarp Nov 22, 2024, 12:27 AM

#

I am running this on a intel server no GPU so i dont need drivers I still get the error weird

#

Still get this

#

nova plover Nov 22, 2024, 2:42 AM

#

Is the Server running Linux or are you using WSL2? I don't know if it makes any difference but I am running Python 3.11.2.

mighty scroll Nov 22, 2024, 2:53 AM

#

Seems like the pod may be having issues reaching the Internet to pull the asr model

#

Maybe try exec into the pod and run a ping or curl to google or something to see if it can get anywhere?

minor escarp Nov 22, 2024, 11:02 AM

#

im running stock debian

#

I cant exec into it because it has not started from the error

mighty scroll Nov 22, 2024, 2:03 PM

#

hm looking at your yaml, shouldn't all your environment values have two spaces in front?

#

you have this:

environment:
- TZ=US/Eastern
- local_files_only=False
- HF_HUB_CACHE=/tmp

But think it needs to be this:

environment:
  - TZ=US/Eastern
  - local_files_only=False
  - HF_HUB_CACHE=/tmp

#

though oddly I don't think i set any of these on my container and it is working fine..

#

Yeah took a look at my containers, I don't set any of these

minor escarp Nov 22, 2024, 9:23 PM

#

yea these where some of the things i tried to fix it

#

oh sorry read it wrong

#

#

i copied it wrong

mighty scroll Nov 22, 2024, 9:28 PM

#

not sure any of the env values are needed

#

by default that local_giles_only thing should already be false

#

and the hf hub cache just goes to the data directory

#

only difference I have is that I am pulling a different model by using the direct name from huggingface

#

and I don't have any of those env variables. Also using a persistent volume store for my data directory and not a local file system mount, but I am using k8s 😅

#

still wondering if it is some kind of docker networking thing where it can't make it's way to huggingface

#

cuz that error is a bit generic I think, I think even with localfilesoff it would throw if it can't connect to huggingface

minor escarp Nov 22, 2024, 9:59 PM

#

ok i have checked and it is just isolated to this device the rest of my network does not have this problem I get this error when installing ollama to

mighty scroll Nov 22, 2024, 11:27 PM

#

Picture is too tiny to see 😅

minor escarp Nov 23, 2024, 12:49 AM

#

mb

#

copy paste broken from portainer

#

2024-11-23T00:49:14.937506858Z huggingface_hub.errors.LocalEntryNotFoundError: Cannot find an appropriate cached snapshot folder for the specified rev

mighty scroll Nov 23, 2024, 1:43 AM

#

Same in ollama, interesting

#

Can other containers reach out?

#Issue when deploying Wyoming container