一篇文章解决ComfyUI常见的故障报错！

前言

学习和使用ComfyUI最痛苦的是什么？就是这满屏的红色方框和和[报错信息] “报错信息”)，处理完一批又一批，很多人玩了一两个流程就搞不下去了，很多初学者因此就放弃了。

有道是：配置流程大半天，跑通出图三分钟。ComfyUI的快感可能就是这三分钟。体验[comfyUI]给你带来的高效和快感，希望这篇文章能帮助解决常见的技术问题，让你畅快淋漓的使用ComfyUI。

最近因为部分[SD]的流程需要自动化，批量化，所以开始学习和使用ComfyUI，我搞了一个多月了，期间经历过各种问题，由于是技术出身，对troubleshooting本身就执着，所以一步一步的解决问题过程中积累了很多经验，同时也在网上做一些课程，帮助一些非技术出身的小白学员入门了comfyUI.希望通过本文就记录一些安装和使用的常见问题，让大家踩坑后能尽快解决吧。其实github上issue里面也有很多有用的内容，如果英文好的小伙伴可以直接去哪里解决。

以我的使用经验问题大多出在各种[Custom node]安装和模型下载上，另外由于开源组件更新快，很多相关组件都需要及时更新到最新版本，所以有些时候出了问题，把相关的包更新一下，问题就可能解决了。

1. 关于一键安装包

首先要对一键安装包的作者致敬，花了精力方便很多技术小白用户入门学习。

但是我自己使用和教学员的经历来看，一键安装包在后期使用上存在着很多不方便的地方，因为是[开源项目]，主版本和各种custom node更新非常快，很多问题就是版本更新不及时产生的。另外会出现各种客户化节点缺失，最后还是需要学习手动安装[custom node]. 所以对于学习来讲，可能是无法偷懒的。

我对学员说的是：你既然选择了学习ComfyUI,就是选择了挑战自己，自己安装custom node也只是挑战的一小部分。

总结一下：尽量自己直接通过github拉项目或者下载[Portable]安装包，学习安装各种节点。确实不愿意弄，那就去搞一键安装包。

所有的AI设计工具，模型和插件，都已经整理好了，👇获取~

2.安装的常见问题

本文不讨论安装过程，因为安装的指南文章很多，只简要说一下安装需要注意的问题.

ComfyUI和其它sd的工具一样，非常依赖cuda和c语言的开发环境，所以cuda相关的包, windows上的微软开发工具一定要事先安装好。即便你用的是[portable]或者一键安装包，后面使用过程中会给你带来极大的便利。如果使用conda或python的[venv],那就要安装好python环境。

总结：

必须安装好cuda，[cudnn]等相关的工具环境
安装[msvs]，gcc这类开发环境
python环境。（用[embeding python]的记住安装要用 …/

b.与其他[Stable diffusion]软件共享模型文件

在[comfyui]目录下有个extra_model_paths.yaml.example文，你需要重命名为extra_model_paths.yaml,

里面有个a111的配置就是[webui]，你把webui的路径配置上就可以了。

#Rename this 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
a111:base_path: C:/dev/sd160/stable-diffusion-webuicheckpoints: models/Stable-diffusionconfigs: models/Stable-diffusionvae: models/VAE[lora](https://www.zhihu.com/search?q=lora&search_source=Entity&hybrid_search_source=Entity&hybrid_search_extra=%7B%22sourceType%22%3A%22article%22%2C%22sourceId%22%3A%22660102125%22%7D "lora")s: |models/[Lora](https://www.zhihu.com/search?q=Lora&search_source=Entity&hybrid_search_source=Entity&hybrid_search_extra=%7B%22sourceType%22%3A%22article%22%2C%22sourceId%22%3A%22660102125%22%7D "Lora")models/LyCORISupscale_models: |models/ESRGANmodels/RealESRGANmodels/SwinIRembeddings: embeddingshypernetworks: models/hypernetworkscontrolnet: models/ControlNet#other_ui:
#    base_path: path/to/ui
#    checkpoints: models/checkpoints
#    gligen: models/gligen
#    custom_nodes: path/custom_nodes

c. ComfyUI的更新

从评论区可以看到，很多时候遇到的问题都是ComfyUI没有更新导致的，如果你经常使用它，并且经常跑网上最新的一些流程例子，你就需要经常性的更新ComfyUI。更新的方法一般有以下几种。

1）如果是[windows portable]版本，有个update文件夹，直接执行里面的update_comfyui.bat就可以。另外一个命令会更新一些python依赖包，

2）重新在comfy网址下载最新版本，这个方法不方便的地方是需要你把所有custom_nodes文件夹拷贝过去，而且有些节点需要手工安装python包和依赖。

3）如果是用[git clone]方式安装的，你重新git pull一下就可以了。

3.ComfyUI节点的安装

很多人说用Manager,[节点管理器]节点管理器")来管理custom node,但是现实情况是国内由于墙的原因，经常下载失败。

所以学会手动安装是ComfyUI必备技能。如果你引入的流程上，出现大量的红色节点和红色的提示，那么说明你缺少对应节点，

你可以首先点击"[Install]Missing [Custom Node]s

这个时候注意，要开着科学上网，因为它要访问这个两个github的网址

FETCH DATA from: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/custom-node-list.json
FETCH DATA from: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/extension-node-map.json

你点击Install后，虽然显示安装成功，并且让你重新启动，但是不要被表面现象迷惑，很可能安装是失败的。

只有如下图显示了紫色disable和红色[uninstall]的色块才是安装成功的标志.

要在文件浏览器里检查，对应的文件夹是不是出现了, ComfyUI/custom_nodes下按照时间排序，看看有没有当前时间段创建的文件夹，如果没有说明需要手工安装。

通过两个方式拿到节点的git库地址,一是点击install之后看后台的信息, 看到有git clone https://….

b.手工安装[节点node]

可以打开一个cmd窗口，进到comfyUI/custom_node路径下，把命令拷贝过去 git clone ‘(https://link.zhihu.com/?target=https%3A//… “https://…”).’,这会儿不要执行，因为访问git需要加一个代理。你可以把引号删除，同时加上 . 如图所示：

ghproxy是非常好用的git 代理。但是最近被墙了，可以改用，非常推荐大家使用。

注：我个人使用是comfyUI开两个窗口，不要直接点run_nvidia_gpu.bat 这个文件启动，开一个cmd窗口或者[poweshell]，进入命令行启动它，因为后面安装节点要反复执行这个命令，你这个窗口只需要按ctrl+c 停止服务器，再按上箭头和回车就可以重新启动新的comfyUI,效率高很多。另外一个窗口只负责更新节点和python库，也省去了每次都要新开窗口进入[custom nodes路径]的时间。

c.安装python依赖包

有些节点如果重启后还是import failed,可能是需要你手工安装一些python包，你需要仔细查看对应github库的[readme.md]文件。在启动时，会报一些 No module named xxxx 这类错误信息，那就是需要手动安装这些包。

像下面这个示例中charactor face swap包，就写明了手动安装时，需要执行的命令

这时可以进入另一个专门安装包和节点的窗口。执行…\…\…\python_embeded\[python]

注意linux环境和windows环境斜杠方向不同。

这里面要注意一下，其实install.py是comfyUI启动或安装时会自动调用的脚本.只不过由于我们有些时候安装不成功才需要执行它。另外一些节点常见的安装过程是需要 …\…\…\python_embeded\python -m [pip] install -r requirements.txt 来手动安装各种python包。我的这种安装方式适用于安装portable包的小伙伴，要是你通过venv或者[conda环境]安装的comfyUI,那你直接用pip就好了。

还有些节点需要额外的安装步骤，请自行仔细阅读开源项目的readme.md文件，答案都在那里。

d.一些特殊的节点

有些特殊的节点，其实只是一个python文件，你下载了github文件夹后还需要把文件直接放到custom_nodes的目录下。比如这个[clipseg]节点组件。

还有这几个组件。

e.网络不通怎么办？

如果网络不通，一般会报"[WinError 121]信号灯超时时间已到”或者

Error handling request
Traceback (most recent call last):File "C:\DEV\ComfyUI_windows_portable\python_embeded\lib\site-packages\[aiohttp](https://www.zhihu.com/search?q=aiohttp&search_source=Entity&hybrid_search_source=Entity&hybrid_search_extra=%7B%22sourceType%22%3A%22article%22%2C%22sourceId%22%3A%22660102125%22%7D "aiohttp")\web_protocol.py", line 433, in _handle_requestresp = await request_handler(request)File "C:\DEV\ComfyUI_windows_portable\python_embeded\lib\site-packages\aiohttp\web_app.py", line 504, in _handleresp = await handler(request)File "C:\DEV\ComfyUI_windows_portable\python_embeded\lib\site-packages\aiohttp\web_[middlewares.py](https://www.zhihu.com/search?q=middlewares.py&search_source=Entity&hybrid_search_source=Entity&hybrid_search_extra=%7B%22sourceType%22%3A%22article%22%2C%22sourceId%22%3A%22660102125%22%7D "middlewares.py")", line 117, in implreturn await handler(request)File "C:\DEV\ComfyUI_windows_portable\ComfyUI\[server.py](https://www.zhihu.com/search?q=server.py&search_source=Entity&hybrid_search_source=Entity&hybrid_search_extra=%7B%22sourceType%22%3A%22article%22%2C%22sourceId%22%3A%22660102125%22%7D "server.py")", line 46, in cache_controlresponse: web.Response = await handler(request)File "C:\DEV\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Manager\__init__.py", line 522, in fetch_customnode_mappingsjson_obj = await get_data(uri)File "C:\DEV\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Manager\__init__.py", line 367, in get_dataasync with session.get(uri) as resp:File "C:\DEV\ComfyUI_windows_portable\python_embeded\lib\site-packages\[aiohttp\client.py](https://www.zhihu.com/search?q=aiohttp%5Cclient.py&search_source=Entity&hybrid_search_source=Entity&hybrid_search_extra=%7B%22sourceType%22%3A%22article%22%2C%22sourceId%22%3A%22660102125%22%7D "aiohttp\client.py")", line 1141, in __aenter__self._resp = await self._coroFile "C:\DEV\ComfyUI_windows_portable\python_embeded\lib\site-packages\aiohttp\client.py", line 560, in _requestawait resp.start(conn)File "C:\DEV\ComfyUI_windows_portable\python_embeded\lib\site-packages\aiohttp\client_[reqrep.py](https://www.zhihu.com/search?q=reqrep.py&search_source=Entity&hybrid_search_source=Entity&hybrid_search_extra=%7B%22sourceType%22%3A%22article%22%2C%22sourceId%22%3A%22660102125%22%7D "reqrep.py")", line 899, in startmessage, payload = await protocol.read()  # type: ignore[union-attr]File "C:\DEV\ComfyUI_windows_portable\python_embeded\lib\site-packages\aiohttp\streams.py", line 616, in readawait self._waiter
[aiohttp.client_exceptions](https://www.zhihu.com/search?q=aiohttp.client_exceptions&search_source=Entity&hybrid_search_source=Entity&hybrid_search_extra=%7B%22sourceType%22%3A%22article%22%2C%22sourceId%22%3A%22660102125%22%7D "aiohttp.client_exceptions").ClientOSError: [WinError 121] 信号灯超时时间已到

无法连接github的错误信息"Cannot connect to host [raw.githubusercontent.com:443]ssl:default [None]"

Traceback (most recent call last):File "C:\DEV\ComfyUI_windows_portable\python_embeded\lib\site-packages\aiohttp\web_[protocol.py](https://www.zhihu.com/search?q=protocol.py&search_source=Entity&hybrid_search_source=Entity&hybrid_search_extra=%7B%22sourceType%22%3A%22article%22%2C%22sourceId%22%3A%22660102125%22%7D "protocol.py")", line 433, in _handle_requestresp = await [request_handler](https://www.zhihu.com/search?q=request_handler&search_source=Entity&hybrid_search_source=Entity&hybrid_search_extra=%7B%22sourceType%22%3A%22article%22%2C%22sourceId%22%3A%22660102125%22%7D "request_handler")(request)File "C:\DEV\ComfyUI_windows_portable\python_embeded\lib\site-packages\aiohttp\web_app.py", line 504, in _handleresp = await handler(request)File "C:\DEV\ComfyUI_windows_portable\python_embeded\lib\site-packages\aiohttp\web_middlewares.py", line 117, in implreturn await handler(request)File "C:\DEV\ComfyUI_windows_portable\ComfyUI\server.py", line 46, in cache_controlresponse: web.Response = await handler(request)File "C:\DEV\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Manager\__init__.py", line 582, in fetch_customnode_listjson_obj = await get_data(uri)File "C:\DEV\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Manager\__init__.py", line 367, in get_dataasync with session.get(uri) as resp:File "C:\DEV\ComfyUI_windows_portable\python_embeded\lib\site-packages\aiohttp\client.py", line 1141, in __aenter__self._resp = await self._coroFile "C:\DEV\ComfyUI_windows_portable\python_embeded\lib\site-packages\aiohttp\client.py", line 536, in _requestconn = await self._connector.connect(File "C:\DEV\ComfyUI_windows_portable\python_embeded\lib\site-packages\aiohttp\connector.py", line 540, in connectproto = await self._create_connection(req, traces, timeout)File "C:\DEV\ComfyUI_windows_portable\python_embeded\lib\site-packages\aiohttp\connector.py", line 899, in _create_connection_, proto = await self._create_proxy_connection(req, traces, timeout)File "C:\DEV\ComfyUI_windows_portable\python_embeded\lib\site-packages\aiohttp\[connector.py](https://www.zhihu.com/search?q=connector.py&search_source=Entity&hybrid_search_source=Entity&hybrid_search_extra=%7B%22sourceType%22%3A%22article%22%2C%22sourceId%22%3A%22660102125%22%7D "connector.py")", line 1328, in _create_proxy_connectionreturn await self._start_tls_connection(File "C:\DEV\ComfyUI_windows_portable\python_embeded\lib\site-packages\aiohttp\connector.py", line 1111, in _start_tls_connectionraise client_error(req.connection_key, exc) from exc
aiohttp.client_exceptions.ClientConnectorError: Cannot connect to host raw.githubusercontent.com:443 ssl:default [None]

这个时候不要着急，赶紧把科学上网打开。同时它不会影响comfyUI的运行.

f. Failed to connect to [http://raw.githubusercontent.com:443]

这个错误也是由于网络原因引起的。但是原因可能不尽相同。除了科学上网，有些时候国内的dns会解析错误的[http://raw.githubusercontent.com]域名，还有些ISP([互联网服务提供商]在[DNS]层面出于某些未知原因屏蔽了

翻译关键点:可以通过在windows手工配置hosts文件解决。

将185.199.108.133 [raw.githubusercontent.com]直接加到C:\Windows\System32\drivers\etc\hosts文件的最后一行。也有将dns改为国外服务器的。总之保证这个域名可以正常访问就可以。

4.comfyUI的使用

a.正常启动

下面是正常启动的界面，所有节点都正常的加载。

如果安装失败，会显示IMPORT FAILED 如图：

b. 启动参数

运行一般windows下都是运行run_nvidia_gpu.bat这个脚本，linux环境下，就是直接python main.py。

启动的所有参数可以通过python main.py -h 参数来查看。愿意学习代码的，也可以看cli_args.py这个文件来学习。

E:\ComfyUI_windows_portable\ComfyUI>..\python_embeded\python.exe main.py -h
** ComfyUI startup time: 2024-01-06 15:42:26.697935
** Platform: Windows
** Python version: 3.11.6 (tags/v3.11.6:8b6ee5b, Oct  2 2023, 14:57:12) [MSC v.1935 64 bit (AMD64)]
** Python executable: E:\ComfyUI_windows_portable\python_embeded\python.exe
** Log path: E:\ComfyUI_windows_portable\ComfyUI\[comfyui.log](https://www.zhihu.com/search?q=comfyui.log&search_source=Entity&hybrid_search_source=Entity&hybrid_search_extra=%7B%22sourceType%22%3A%22article%22%2C%22sourceId%22%3A%22660102125%22%7D "comfyui.log")Prestartup times for custom nodes:0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\rgthree-comfy0.1 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Managerusage: main.py [-h] [--listen [IP]] [--port PORT] [--[enable-cors-header](https://www.zhihu.com/search?q=enable-cors-header&search_source=Entity&hybrid_search_source=Entity&hybrid_search_extra=%7B%22sourceType%22%3A%22article%22%2C%22sourceId%22%3A%22660102125%22%7D "enable-cors-header") [ORIGIN]] [--max-upload-size MAX_UPLOAD_SIZE][--extra-model-paths-config PATH [PATH ...]] [--output-directory OUTPUT_DIRECTORY][--temp-directory TEMP_DIRECTORY] [--[input-directory](https://www.zhihu.com/search?q=input-directory&search_source=Entity&hybrid_search_source=Entity&hybrid_search_extra=%7B%22sourceType%22%3A%22article%22%2C%22sourceId%22%3A%22660102125%22%7D "input-directory") INPUT_DIRECTORY] [--auto-launch][--disable-auto-launch] [--cuda-device DEVICE_ID] [--[cuda-malloc](https://www.zhihu.com/search?q=cuda-malloc&search_source=Entity&hybrid_search_source=Entity&hybrid_search_extra=%7B%22sourceType%22%3A%22article%22%2C%22sourceId%22%3A%22660102125%22%7D "cuda-malloc") | --disable-cuda-malloc][--dont-upcast-attention] [--force-fp32 | --force-fp16][--bf16-unet | --fp16-unet | --fp8_e4m3fn-unet | --fp8_e5m2-unet][--fp16-vae | --fp32-vae | --bf16-vae][--fp8_e4m3fn-text-enc | --fp8_e5m2-text-enc | --fp16-text-enc | --fp32-text-enc][--directml [DIRECTML_DEVICE]] [--disable-ipex-optimize][--preview-method [none,auto,[latent2rgb](https://www.zhihu.com/search?q=latent2rgb&search_source=Entity&hybrid_search_source=Entity&hybrid_search_extra=%7B%22sourceType%22%3A%22article%22%2C%22sourceId%22%3A%22660102125%22%7D "latent2rgb"),taesd]][--use-split-cross-attention | --use-quad-cross-attention | --use-[pytorch-cross-attention](https://www.zhihu.com/search?q=pytorch-cross-attention&search_source=Entity&hybrid_search_source=Entity&hybrid_search_extra=%7B%22sourceType%22%3A%22article%22%2C%22sourceId%22%3A%22660102125%22%7D "pytorch-cross-attention")][--disable-xformers] [--gpu-only | --highvram | --normalvram | --lowvram | --novram | --cpu][--disable-smart-memory] [--deterministic] [--dont-print-server] [--quick-test-for-ci][--[windows-standalone](https://www.zhihu.com/search?q=windows-standalone&search_source=Entity&hybrid_search_source=Entity&hybrid_search_extra=%7B%22sourceType%22%3A%22article%22%2C%22sourceId%22%3A%22660102125%22%7D "windows-standalone")-build] [--disable-metadata]options:-h, --help            show this help message and exit--listen [IP]         Specify the IP address to listen on (default: 127.0.0.1). If --listen is provided without an[argument](https://www.zhihu.com/search?q=argument&search_source=Entity&hybrid_search_source=Entity&hybrid_search_extra=%7B%22sourceType%22%3A%22article%22%2C%22sourceId%22%3A%22660102125%22%7D "argument"), it defaults to 0.0.0.0. (listens on all)--port PORT           Set the listen port.--enable-cors-header [ORIGIN]Enable CORS (Cross-Origin Resource Sharing) with optional origin or allow all with default'*'.--[max-upload-size](https://www.zhihu.com/search?q=max-upload-size&search_source=Entity&hybrid_search_source=Entity&hybrid_search_extra=%7B%22sourceType%22%3A%22article%22%2C%22sourceId%22%3A%22660102125%22%7D "max-upload-size") MAX_UPLOAD_SIZESet the maximum upload size in MB.--extra-model-paths-config PATH [PATH ...]Load one or more extra_model_paths.yaml files.--output-directory OUTPUT_DIRECTORYSet the ComfyUI output directory.--temp-directory TEMP_DIRECTORYSet the ComfyUI temp directory (default is in the ComfyUI directory).--input-directory INPUT_DIRECTORYSet the ComfyUI input directory.--auto-launch         Automatically launch ComfyUI in the default browser.--disable-auto-launchDisable auto launching the browser.--cuda-device DEVICE_IDSet the id of the cuda device this instance will use.--cuda-malloc         Enable cudaMallocAsync (enabled by default for torch 2.0 and up).--disable-cuda-mallocDisable cudaMallocAsync.--dont-upcast-attentionDisable upcasting of attention. Can boost speed but increase the chances of black images.--force-fp32          Force fp32 (If this makes your GPU work better please report it).--force-fp16          Force fp16.--bf16-unet           Run the UNET in bf16. This should only be used for testing stuff.--fp16-unet           Store unet weights in fp16.--fp8_e4m3fn-unet     Store unet weights in fp8_e4m3fn.--fp8_e5m2-unet       Store unet weights in fp8_e5m2.--fp16-vae            Run the VAE in fp16, might cause black images.--fp32-vae            Run the VAE in full precision fp32.--bf16-vae            Run the VAE in bf16.--fp8_e4m3fn-text-encStore text encoder weights in fp8 (e4m3fn variant).--fp8_e5m2-text-enc   Store text encoder weights in fp8 (e5m2 variant).--fp16-text-enc       Store text encoder weights in fp16.--fp32-text-enc       Store text encoder weights in fp32.--directml [DIRECTML_DEVICE]Use torch-directml.--disable-ipex-optimizeDisables ipex.optimize when loading models with Intel GPUs.--preview-method [none,auto,latent2rgb,taesd]Default preview method for sampler nodes.--use-split-cross-attentionUse the split cross attention optimization. Ignored when xformers is used.--use-quad-cross-attentionUse the sub-quadratic cross attention optimization . Ignored when xformers is used.--use-pytorch-cross-attentionUse the new pytorch 2.0 cross attention function.--disable-xformers    Disable xformers.--gpu-only            Store and run everything (text encoders/CLIP models, etc... on the GPU).--highvram            By default models will be unloaded to CPU memory after being used. This option keeps them inGPU memory.--normalvram          Used to force normal vram use if lowvram gets automatically enabled.--lowvram             Split the unet in parts to use less vram.--novram              When lowvram isn't enough.--cpu                 To use the CPU for everything (slow).--disable-smart-memoryForce ComfyUI to agressively offload to regular ram instead of keeping models in vram when itcan.--deterministic       Make pytorch use slower deterministic algorithms when it can. Note that this might not makeimages deterministic in all cases.--dont-print-server   Don't print server output.--quick-test-for-ci   Quick test for CI.--windows-standalone-buildWindows standalone build: Enable convenient things that most people using the standalonewindows build will probably enjoy (like auto opening the page on startup).--disable-metadata    Disable saving prompt metadata in files.

其中常用的几个参数，简单介绍一下。

–listen 如果你想让别人也可以访问你的comfyui，或者提供API访问，就需要设置这个参数，默认是监听机器上所有的ip。如果是只想让它监听特定的IP，那你加上就行比如”192.168.0.18“

–port 指定监听端口

–lowvram 降低显存的占用。

c manager访问不到

我们在使用manager时，会遇到下面这个报错。

FETCH DATA from: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/custom-node-list.json
[ComfyUI-Manager] Due to a network error, switching to local mode.
=> custom-node-list.json
=> [WinError 121] 信号灯超时时间已到那如何解决呢？

首先，这个问题原因是国内对这个域名[raw.githubusercontent.com]存在DNS不准确的问题。大家可以通过[https://sites.ipaddress.com/raw.githubusercontent.com/] 方式直接查询域名的ip，结果如下：

然后手工修改 ”c:\windows\System32\drivers\etc\hosts“文件

增加一行：

然后执行：

ipconfig /flushdns

再尝试访问，就可以了。

5. 模型文件找不到

模型文件分为很多类别，一类是和sd相关的checkpoint, Lora, embedding这，常见的是你拉过来别的workflow,作者命名这些文件有自己的习惯，拿过来需要确认和自己本地保存的文件名字一致。一类是custom node自己需要的模型文件，第一次运行时会在相应的网站，一般就是github,huggingface这类，国内也有modelscope这类。

a. checkpoint,lora名字不一致

是一种常见的错误，把别人的workflow加载进来时，记得要看model名字和自己本地的是否一样

File “C:\DEV\ComfyUI_windows_portable\ComfyUI\ execution.py”, line 153, in recursive_execute
output_data, output_ui = get_output_data(obj, input_data_all)
File “C:\DEV\ComfyUI_windows_portable\ComfyUI\ execution.py”, line 83, in get_output_data
return_values = map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True)
File “C:\DEV\ComfyUI_windows_portable\ComfyUI\ execution.py”, line 76, in map_node_over_list
results.append(getattr(obj, func)(**slice_dict(input_data_all, i)))
File “C:\DEV\ComfyUI_windows_portable\ComfyUI\ nodes.py”, line 476, in load_checkpoint
out = comfy.sd.load_checkpoint_guess_config(ckpt_path, output_vae=True, output_clip=True, embedding_directory=folder_paths.get_folder_paths(“embeddings”))
File “C:\DEV\ComfyUI_windows_portable\ComfyUI\comfy\ sd.py”, line 400, in load_checkpoint_guess_config
sd = comfy.utils.load_torch_file(ckpt_path)
File “C:\DEV\ComfyUI_windows_portable\ComfyUI\comfy\ utils.py”, line 12, in load_torch_file
if ckpt.lower().endswith(“.safetensors”):
AttributeError: ‘NoneType’ object has no attribute ‘lower’

b.无法连接hugging face

关键字 huggingface.co "Connection to http://huggingface.co timed out类。

‘(MaxRetryError("HTTPSConnectionPool(host=’ Hugging Face – The AI community building the future.‘, port=443): Max retries exceeded with url: /lllyasviel/Annotators/resolve/main/ZoeD_M12_N.pt (Caused by ConnectTimeoutError(<urllib3.connection.HTTPSConnection object at 0x000001D934781DB0>, ‘Connection to Hugging Face – The AI community building the future. timed out. (connect timeout=10)’))"), ‘(Request ID: dc2a2763-47a3-4862-b320-9aacf3a28b19)’)’ thrown while requesting HEAD https://huggingface.co/lllyasviel/Annotators/resolve/main/ZoeD_M12_N.pt
WARNING:huggingface_hub.utils._http:‘(MaxRetryError("HTTPSConnectionPool(host=’ Hugging Face – The AI community building the future.‘, port=443): Max retries exceeded with url: /lllyasviel/Annotators/resolve/main/ZoeD_M12_N.pt (Caused by ConnectTimeoutError(<urllib3.connection.HTTPSConnection object at 0x000001D934781DB0>, ‘Connection to Hugging Face – The AI community building the future. timed out. (connect timeout=10)’))"), ‘(Request ID: dc2a2763-47a3-4862-b320-9aacf3a28b19)’)’ thrown while requesting HEAD https://huggingface.co/lllyasviel/Annotators/resolve/main/ZoeD_M12_N.pt

如何从huggingface下载模型

补充一个新的方法，可以在通过环境变量设置huggingface代理来解决这个问题。

set HF_ENDPOINT=https://hf-mirror.com

6.执行的节点没有配置好

这是前几天评论区小伙伴留下，这个我们可以看到执行错误的节点有一个红色边框，并报错。报错是错模型没有找到，[undefine]是用户没有选择本机上的模型会出现的。同类的错误都可以这个方式来进行排查，比如sampler节点如果忘记输入positive condition，就会报如下错误。

再比如这个apply faceID节点忘记选择model，model就会出现红圈。

执行节点时，我们经常会遇到错误，很多时候需要检查是不是我们节点的版本太旧了，和其它节点匹配出了问题。

比如我有一次使用

output = model.apply_model(input_x, timestep_, **c).chunk(batch_chunks)
File “E:\DEV\ComfyUI_windows_portable\ComfyUI\comfy\model_base.py”, line 77, in apply_model
context = context.to(dtype)
AttributeError: ‘NoneType’ object has no attribute ‘to’

上面错误就是代码对一个空对象进行操作，我检查了流程，没发现设置的问题。就怀疑是节点版本的问题。我更新了节点之后，重启服务器，故障解决。

为了帮助大家更好地掌握 ComfyUI，我在去年花了几个月的时间，撰写并录制了一套ComfyUI的基础教程，共六篇。这套教程详细介绍了选择ComfyUI的理由、其优缺点、下载安装方法、模型与插件的安装、工作流节点和底层逻辑详解、遮罩修改重绘/Inpenting模块以及SDXL工作流手把手搭建。