ComfyUI: handle errors on load and various bugfixes
4 min read
comfyui handle errors on load and various bugfixes featured meme
ComfyUI will spit out lots of errors on load, after updates especially. Upgrade a python package? error! Update a custom node? error! Update the core? errors! Here are few of the error I came accross and how I fixed them.
ComfyUI Load Errors
dwpose.py:26: UserWarning: DWPose: Onnxruntime not found or doesn’t come with acceleration providers
E:\GPT\ComfyUI\custom_nodes\comfyui_controlnet_aux\node_wrappers\dwpose.py:26: UserWarning: DWPose: Onnxruntime not found or doesn't come with acceleration providers, switch to OpenCV with CPU device. DWPose might run very slowly
warnings.warn("DWPose: Onnxruntime not found or doesn't come with acceleration providers, switch to OpenCV with CPU device. DWPose might run very slowly")
Error discussed here
That’s a risky take to fix, involving re-installing onnxruntime and onnxruntime-gpu and Gigabytes of download. You will need a valid MSVC / VisualStudio2022 environment.
pip uninstall onnxruntime onnxruntime-gpu pip install onnxruntime-gpu --extra-index-url https://aiinfra.pkgs.visualstudio.com/PublicPackages/_packaging/onnxruntime-cuda-12/pypi/simple/ --no-cache-dir
FutureWarning: Using TRANSFORMERS_CACHE is deprecated
transformers\utils\hub.py:124: FutureWarning: UsingTRANSFORMERS_CACHEis deprecated and will be removed in v5 of Transformers. UseHF_HOMEinstead.
add this in your Comfy load batch: set HF_HOME=%stableROOT%\.cache\huggingface with stableROOT=<path>\sd.webui\webui
Cannot import efficiency-nodes-comfyui module for custom nodes: name ‘latent_versions_updated’ is not defined
Cannot import efficiency-nodes-comfyui module for custom nodes: name 'latent_versions_updated' is not defined AttributeError: 'Block' object has no attribute 'drop_path'
This error is discussed here, but resolving it was not easy.
First I tried updating the requirements as recommended:
vi custom_nodes\comfyui_controlnet_aux\requirements.txt timm=0.6.7 pip install timm==0.6.7 pip upgrade
But then other errors started popping up:
model.onnx: 100%|███████████████████████████████████████████████████████████████████| 326M/326M [00:24<00:00, 13.4MB/s] selected_tags.csv: 100%|████████████████████████████████████████████████████████████| 254k/254k [00:00<00:00, 1.77MB/s] EP Error D:\a\_work\1\s\onnxruntime\core\session\provider_bridge_ort.cc:1131 onnxruntime::ProviderLibrary::Get [ONNXRuntimeError] : 1 : FAIL : LoadLibrary failed with error 126 "" when trying to load "E:\GPT\stable-diffusion-webui\venv\lib\site-packages\onnxruntime\capi\onnxruntime_providers_tensorrt.dll" when using ['TensorrtExecutionProvider', 'CUDAExecutionProvider', 'CPUExecutionProvider'] Falling back to ['CUDAExecutionProvider', 'CPUExecutionProvider'] and retrying. outdoors, sky, day, cloud, water, tree, no_humans, window, ocean, cloudy_sky, plant, building, scenery, city, watercraft, bridge, boat
onnxruntime error discussed here, and also here
so you check for onnxruntime version and nvidia-smi:
# required dlls: pip show onnxruntime # 1.15 nvidia-smi # 12.0
But no idea how I solved it. maybe just reinstlaling Comfy.
Update error: ComfyUI-Allor
Allor errors are discussed here and here
What I did to fix it:
pushd custom_nodes\ComfyUI-Allor\ git fetch origin main git reset --hard origin/main git pull git config --global --add safe.directory E:/GPT/ComfyUI/custom_nodes/ComfyUI-post-processing-nodes
[comfy_mtb] | WARNING -> Found multiple match
FETCH DATA from: E:\GPT\ComfyUI\custom_nodes\ComfyUI-Manager\extension-node-map.json [comfy_mtb] | WARNING -> Found multiple match, we will pick the first E:\GPT\ComfyUI\models\upscale_models ['E:\\GPT\\ComfyUI\\models\\upscale_models', '../stable-diffusion-webui/models/ESRGAN', '../stable-diffusion-webui/models/GFPGAN', '../stable-diffusion-webui/models/RealESRGAN', '../stable-diffusion-webui/models/SwinIR']
Simply edit extra_model_paths.yaml and remove the upscale duplicates:
#Rename extra_model_paths.yaml.example to extra_model_paths.yaml and ComfyUI will load it
# config for a1111 ui
# all you have to do is change the base_path to where yours is installed
#your base path should be either an existing comfy install or a central folder where you store all of your models, loras, etc.
comfyui:
base_path: ./
checkpoints: |
models/checkpoints
configs: |
models/checkpoints
vae: |
models/vae
models/vae_approx
loras: |
models/loras
# [comfy_mtb] | WARNING -> Found multiple match, we will pick the first E:\GPT\ComfyUI\models\upscale_models
# upscale_models: |
# models/upscale_models/ESRGAN
# models/upscale_models/GFPGAN
# models/upscale_models/RealESRGAN
# models/upscale_models/SwinIR
clip: |
models/clip
clip_vision: |
models/clip_vision
hypernetworks: |
models/hypernetworks
controlnet: |
models/ControlNet
embeddings: |
models/embeddings
a111:
base_path: ../sd.webui/webui/
checkpoints: |
models/Stable-diffusion
configs: |
models/Stable-diffusion
vae: |
models/vae
models/vae_approx
loras: |
models/Lora
models/LyCORIS
upscale_models: |
models/ESRGAN
models/GFPGAN
models/RealESRGAN
models/SwinIR
clip: |
models/clip
clip_vision: |
models/clip_vision
hypernetworks: |
models/hypernetworks
controlnet: |
models/ControlNet
gligen: |
models/gligen
embeddings: |
embeddings
WAS Node Suite Warning: ffmpeg_bin_path is not set
WAS Node Suite Warning:ffmpeg_bin_pathis not set inE:\GPT\ComfyUI\custom_nodes\was-node-suite-comfyui\was_suite_config.jsonconfig file. Will attempt to use system ffmpeg binaries if available.
As long as ffmpeg is in your PATH, you can ignore, or edit custom_nodes\was-node-suite-comfyui\was_suite_config.json as it says
WARNING: Ignoring invalid distribution -rotobuf
WARNING: Ignoring invalid distribution -rotobuf (e:\gpt\stable-diffusion-webui\venv\lib\site-packages)
Try deleting the venv folder completely and let it install once again, they possibly updated torch so you need to reinstall all.
If doesn’t work, try doing this before the git pull:
pushd stable-diffusion-webui\venv\Lib\site-packages\ git stash git pull
When loading the graph, the following node types were not found
When loading the graph, the following node types were not found: InsightFaceLoader PrepImageForInsightFace IPAdapterApply IPAdapterApplyFaceID
But they were installed… This happened after I left Comfy asleep for 3 months, then ran an update.
Solution: reinstall sd.webui from scratch, previous A1111 installer with venv had become incompatible.
! No module named ‘gray2color’
classic pip module missing because the custom node maker doesn’t know about requirements.txt
Solution: pip install gray2color simpleeval
WARNING ⚠️ user config directory ‘sd.webui\webui\.cache\AppData\Roaming\Ultralytics’ is not writeable
### Loading: ComfyUI-Impact-Pack (V5.3.1) WARNING ⚠️ user config directory 'E:\GPT\sd.webui\webui\.cache\AppData\Roaming\Ultralytics' is not writeable, defaulting to '/tmp' or CWD.Alternatively you can define a YOLO_CONFIG_DIR environment variable for this path.
Solution: md sd.webui\webui\.cache\AppData\Roaming\Ultralytics
open-clip-torch 2.20.0 requires protobuf<4, but you have protobuf 5.26.1 which is incompatible
Downloading onnx-1.16.0-cp310-cp310-win_amd6[!] ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts. Successfully installed onnx-1.16.0 protobuf-5.26.1 [!] open-clip-torch 2.20.0 requires protobuf<4, but you have protobuf 5.26.1 which is incompatible.
Solution: upgrade everything
Cmd(‘git’) failed due to: exit code(1)
This happens all the time after my latest Comfy upgrade. EVERY node I instlal or upgrade via the Node Manager error out. Installs are just fine, it’s like Git command is timing out or smth.
No idea how to fix that.
dwpose.py:26: UserWarning: DWPose: Onnxruntime not found
[comfyui_controlnet_aux] | INFO -> Using ckpts path: E:\GPT\ComfyUI\custom_nodes\comfyui_controlnet_aux\ckpts
[comfyui_controlnet_aux] | INFO -> Using symlinks: False
[comfyui_controlnet_aux] | INFO -> Using ort providers: ['CUDAExecutionProvider', 'DirectMLExecutionProvider', 'OpenVINOExecutionProvider', 'ROCMExecutionProvider', 'CPUExecutionProvider', 'CoreMLExecutionProvider']
E:\GPT\ComfyUI\custom_nodes\comfyui_controlnet_aux\node_wrappers\dwpose.py:26: UserWarning: DWPose: Onnxruntime not found or doesn't come with acceleration providers, switch to OpenCV with CPU device. DWPose might run very slowly
warnings.warn("DWPose: Onnxruntime not found or doesn't come with acceleration providers, switch to OpenCV with CPU device. DWPose might run very slowly")
I believe a node update took care of it
Error. No styles.csv found
Error. No styles.csv found. Put your styles.csv in E:\GPT\ComfyUI\custom_nodes\ComfyUI_hus_utils\styles.csv. Then press "Refresh".
Solution:
cd custom_nodes\ComfyUI_hus_utils copy example styles.csv styles.csv
raise AssertionError(“Torch not compiled with CUDA enabled”)
Prestartup times for custom nodes:
0.0 seconds: E:\GPT\ComfyUI\custom_nodes\rgthree-comfy
0.0 seconds: E:\GPT\ComfyUI\custom_nodes\comfyui-deploy
0.2 seconds: E:\GPT\ComfyUI\custom_nodes\ComfyUI-Manager
Traceback (most recent call last):
File "E:\GPT\ComfyUI\main.py", line 76, in <module>
import execution
File "E:\GPT\ComfyUI\execution.py", line 11, in <module>
import nodes
File "E:\GPT\ComfyUI\nodes.py", line 21, in <module>
import comfy.diffusers_load
File "E:\GPT\ComfyUI\comfy\diffusers_load.py", line 3, in <module>
import comfy.sd
File "E:\GPT\ComfyUI\comfy\sd.py", line 5, in <module>
from comfy import model_management
File "E:\GPT\ComfyUI\comfy\model_management.py", line 119, in <module>
total_vram = get_total_memory(get_torch_device()) / (1024 * 1024)
File "E:\GPT\ComfyUI\comfy\model_management.py", line 88, in get_torch_device
return torch.device(torch.cuda.current_device())
File "E:\GPT\sd.webui\system\python\lib\site-packages\torch\cuda\__init__.py", line 778, in current_device
_lazy_init()
File "E:\GPT\sd.webui\system\python\lib\site-packages\torch\cuda\__init__.py", line 284, in _lazy_init
raise AssertionError("Torch not compiled with CUDA enabled")
AssertionError: Torch not compiled with CUDA enabled
Certainly a version mismatch during the last update as discussed here
Solution: pip install torch torchvision torchaudio --index-url <a href="https://download.pytorch.org/whl/cu121`">https://download.pytorch.org/whl/cu121
NameError: name ‘CLIPTemperaturePatch’ is not defined
Traceback (most recent call last):
File "E:\GPT\ComfyUI\nodes.py", line 1864, in load_custom_node
module_spec.loader.exec_module(module)
File "<frozen importlib._bootstrap_external>", line 883, in exec_module
File "<frozen importlib._bootstrap>", line 241, in _call_with_frames_removed
File "E:\GPT\ComfyUI\custom_nodes\ComfyUI-AutomaticCFG\__init__.py", line 19, in <module>
"Temperature settings CLIP": CLIPTemperaturePatch,
NameError: name 'CLIPTemperaturePatch' is not defined
Cannot import E:\GPT\ComfyUI\custom_nodes\ComfyUI-AutomaticCFG module for custom nodes: name 'CLIPTemperaturePatch' is not defined
(IMPORT FAILED): E:\GPT\ComfyUI\custom_nodes\ComfyUI-AutomaticCFG
Identified and solved here: edit the __init__.py of the node as explained here
Or just wait for a patch
clip missing: [‘clip_l.logit_scale’, ‘clip_l.transformer.text_projection.weight’]
Error discussed here, seemed to be caused by an update of Comfy
clip missing: ['clip_l.logit_scale', 'clip_l.transformer.text_projection.weight'] Loading 1 new model C:\Users\heruv\ComfyUI\comfy\ldm\modules\attention.py:345: UserWarning: 1Torch was not compiled with flash attention. (Triggered internally at ..\aten\src\ATen\native\transformers\cuda\sdp_utils.cpp:263.) out = torch.nn.functional.scaled_dot_product_attention(q, k, v, attn_mask=mask, dropout_p=0.0, is_causal=False)
solution: edit comfy\supported_models.py and make the pop_keys list empty: (or wait for patch)
def process_clip_state_dict_for_saving(self, state_dict):
# pop_keys = ["clip_l.transformer.text_projection.weight", "clip_l.logit_scale"]
pop_keys = []
for p in pop_keys:
if p in state_dict:
state_dict.pop(p)
Custom Node Related Errors
comfyui-mixlab-nodes: ConnectionResetError: [WinError 10054] An existing connection was forcibly closed by the remote host
FETCH DATA from: E:\GPT\ComfyUI\custom_nodes\ComfyUI-Manager\extension-node-map.json [DONE] Exception in callback _ProactorBasePipeTransport._call_connection_lost(None) handle: <Handle _ProactorBasePipeTransport._call_connection_lost(None)> Traceback (most recent call last): File "asyncio\events.py", line 80, in _run File "asyncio\proactor_events.py", line 162, in _call_connection_lost ConnectionResetError: [WinError 10054] An existing connection was forcibly closed by the remote host #read_workflow_json_files_all E:\GPT\ComfyUI\custom_nodes\comfyui-mixlab-nodes\app\
This is a known issue linked to Comfy itself, discussed here
Solution: ignore it. This happens when you refresh the browser, just restart Comfy instead.
WARNING: shape mismatch when trying to apply embedding
happens when using embeddings with sdxl:
embedding:bad_prompt_version2.pt, embedding:ng_deepnegative_v1_75t.pt, embedding:bad-hands-5.pt, sunbeam, embedding:easynegative, watermark, signature, WARNING: shape mismatch when trying to apply embedding, embedding will be ignored 768 1280
Solution: must use XL negatives:
A:Standard
B:Realistic
C:Anime like
embedding:peopleneg, embedding:unaestheticXL_AYv1, embedding:bad-XL.pt, embedding:negativeXL_A, watermark, signature,
the following node types were not found: IPAdapterApply
Happened when trying to load the ComfyUICharacterCreator from r/ComfyUI. Nodes evolve so fast… IPAdapter_plus errors discussed here.
Solution 1: use old IPAdapter
Solution 2: just work and update the workflow to use the new updated nodes, dammit
