Sygil-Dev/sygil-webui

EOFError: Ran out of input - Unraid Docker

redthista opened this issue · 6 comments

Discussed in #1672

Originally posted by Mizerka November 19, 2022
Hey, using https://hub.docker.com/r/hlky/sd-webui to build the docker instance, it builds fine and webui comes up with no errors in logs, when attempting to generate text to image it just spits out this error;

EOFError: Ran out of input

Traceback:
File "/opt/conda/lib/python3.8/site-packages/streamlit/runtime/scriptrunner/script_runner.py", line 562, in _run_script
exec(code, module.dict)
File "/sd/scripts/webui_streamlit.py", line 174, in
layout()
File "/sd/scripts/webui_streamlit.py", line 138, in layout
layout()
File "/sd/scripts/txt2img.py", line 320, in layout
load_models(False, st.session_state["use_GFPGAN"], st.session_state["use_RealESRGAN"], st.session_state["RealESRGAN_model"], server_state["CustomModel_available"],
File "/sd/scripts/sd_utils.py", line 303, in load_models
config, device, model, modelCS, modelFS = load_sd_model(custom_model)
File "/sd/scripts/sd_utils.py", line 928, in load_sd_model
model = load_model_from_config(config, ckpt_path)
File "/sd/scripts/sd_utils.py", line 332, in load_model_from_config
pl_sd = torch.load(ckpt, map_location="cpu")
File "/opt/conda/lib/python3.8/site-packages/torch/serialization.py", line 713, in load
return _legacy_load(opened_file, map_location, pickle_module, **pickle_load_args)
File "/opt/conda/lib/python3.8/site-packages/torch/serialization.py", line 920, in _legacy_load
magic_number = pickle_module.load(f, **pickle_load_args)

tried to recrete it and it just did the same thing, not sure where im failing, the documentation is... lacking. fwiw it's unraid OS.

running nvidia-smi in docker command shows its seeing the 1050ti just fine

nvidia-smi

Sat Nov 19 02:40:02 2022
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 515.65.01 Driver Version: 515.65.01 CUDA Version: 11.7 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|===============================+======================+======================|
| 0 NVIDIA GeForce ... Off | 00000000:84:00.0 Off | N/A |
| 25% 60C P8 N/A / 75W | 2MiB / 4096MiB | 0% Default |
| | | N/A |
+-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=============================================================================|
| No running processes found |
+-----------------------------------------------------------------------------+

any ideas?

+1, getting the same error on unraid docker.

Looking at the logs, seems like it's not downloading it properly?

01/07/2023 2:28:58 PM
Downloading: https://www.googleapis.com/storage/v1/b/aai-blog-files/o/sd-v1-4.ckpt?alt=media please wait...
01/07/2023 2:28:58 PM
Username/Password Authentication Failed.

Have the same issue. Other models work.

I got it to work by getting the original weights from huggingface.co directly.
The SHA256 matches the file at https://huggingface.co/CompVis/stable-diffusion-v-1-4-original/blob/main/sd-v1-4.ckpt

From a terminal in unraid

cd /mnt/user/appdata/sygil-webui/sd/user_data/model_cache
wget https://huggingface.co/CompVis/stable-diffusion-v-1-4-original/resolve/main/sd-v1-4.ckpt -O model.ckpt.fe4efff1e174c627256e44ec2991ba279b3816e364b49f9be2abc0b3ff3f8556

I got it to work by getting the original weights from huggingface.co directly. The SHA256 matches the file at https://huggingface.co/CompVis/stable-diffusion-v-1-4-original/blob/main/sd-v1-4.ckpt

From a terminal in unraid

cd /mnt/user/appdata/sygil-webui/sd/user_data/model_cache
wget https://huggingface.co/CompVis/stable-diffusion-v-1-4-original/resolve/main/sd-v1-4.ckpt -O model.ckpt.fe4efff1e174c627256e44ec2991ba279b3816e364b49f9be2abc0b3ff3f8556

This fixes it - the original download had 0 bytes.

I got it to work by getting the original weights from huggingface.co directly. The SHA256 matches the file at https://huggingface.co/CompVis/stable-diffusion-v-1-4-original/blob/main/sd-v1-4.ckpt

From a terminal in unraid

cd /mnt/user/appdata/sygil-webui/sd/user_data/model_cache
wget https://huggingface.co/CompVis/stable-diffusion-v-1-4-original/resolve/main/sd-v1-4.ckpt -O model.ckpt.fe4efff1e174c627256e44ec2991ba279b3816e364b49f9be2abc0b3ff3f8556

thanks,that works fine now

Strange this bug hasn't been fixed in updates yet. In any case, the fix above still works.