Centos7 部署 Stable Diffusion

参考：https://www.jianshu.com/p/ff81bb76158a

一、安装最新版 git

yum install libcurl-devel curl-devel -y

# 安装依赖库
yum -y install zlib-devel perl-ExtUtils-MakeMaker asciidoc xmlto openssl-devel curl-devel libcurl-devel
 
# 下载安装包 
wget https://www.kernel.org/pub/software/scm/git/git-2.42.0.tar.xz --no-check-certificate
 
# 解压安装包，进入目录
tar -vxf git-2.42.0.tar.xz
cd git-2.40.0
 
# 编译、安装
./configure
 
make -j4 && make -j4 install
 
# 查看是否安装成功
git --version

clone 代码到本地：

stable diffusion webui：https://github.com/AUTOMATIC1111/stable-diffusion-webui.git

stable diffusion webui forge：https://github.com/lllyasviel/stable-diffusion-webui-forge.git

建议使用 sd webui forge ，优化更好

二、其余步骤：详看参考链接

遇到的问题：

1、git clone 比较慢

解决办法：设置代理 https://blog.csdn.net/dszgf5717/article/details/130735389

2、pip install 比较慢

解决办法：更换源或设置代理 https://blog.csdn.net/dszgf5717/article/details/53138298

3、无法使用GPU

Torch is not able to use GPU; add --skip-torch-cuda-test to COMMANDLINE_ARGS variable to disable this check

解决办法：一般是 pytorch 装的版本不对，重装

i、先卸载原来的

pip uninstall torch torchvision torchaudio

ii、安装与显卡相匹配的版本

pytorch 官网 https://pytorch.org/get-started/locally/

参考文章：https://blog.csdn.net/weixin_41803482/article/details/109231249

4、运行报错

OSError: Can't load tokenizer for 'openai/clip-vit-large-patch14'. If you were trying to load it from 'https://huggingface.co/models', make sure you don't have a local directory with the same name. Otherwise, make sure 'openai/clip-vit-large-patch14' is the correct path to a directory containing all relevant files for a CLIPTokenizer tokenizer.

解决办法：

手动下载 https://huggingface.co/openai/clip-vit-large-patch14/tree/main
下载文件为：
1、config.json
2、flax_model.msgpack
3、merges.txt
4、preprocessor_config.json
5、pytorch_model.bin
6、special_tokens_map.json
7、tokenizer.json
8、tokenizer_config.json
9、vocab.json
 
在项目目录下创建（stable-diffusion-webui）
 
mkdir -p openai/clip-vit-large-patch14
 
把以上九个文件放到新文件夹下

5、运行报错

# 报错1
 
ImportError: libX11.so.6: cannot open shared object file: No such file or directory
 
# 报错2
 
ImportError: libXext.so.6: cannot open shared object file: No such file or directory

解决办法：

yum install libX11 libXext -y

6、运行报错

Cannot locate TCMalloc (improves CPU memory usage)

解决办法：

yum install gperftools gperftools-devel -y

参考文章：https://blog.csdn.net/weixin_57451673/article/details/133919472

7、运行报错

出现 No module named ‘_lzma’

解决方法：

yum install xz-devel python-backports-lzma backports.lzma -y
pip install backports.lzma
pip install xformers

修改文件第 27 行左右的地方

vim /usr/local/lib/python3.10/lzma.py

try:
    from _lzma import *
    from _lzma import _encode_filter_properties, _decode_filter_properties
except ImportError:
    from backports.lzma import *
    from backports.lzma import _encode_filter_properties, _decode_filter_properties

8、运行报错

  File "/data/stable-diffusion/stable-diffusion-webui/modules/processing.py", line 2, in 
    import cv2
  File "/data/stable-diffusion/stable-diffusion-webui/venv/lib/python3.10/site-packages/cv2/__init__.py", line 181, in 
    bootstrap()
  File "/data/stable-diffusion/stable-diffusion-webui/venv/lib/python3.10/site-packages/cv2/__init__.py", line 153, in bootstrap
    native_module = importlib.import_module("cv2")
  File "/usr/local/python3/lib/python3.10/importlib/__init__.py", line 126, in import_module
    return _bootstrap._gcd_import(name[level:], package, level)
ImportError: dlopen: cannot load any more object with static TLS

解决办法：

vim /data/stable-diffusion/stable-diffusion-webui/modules/initialize.py

import importlib
import logging
import os
import cv2 
import sys 
 
...
 
def imports():
    logging.getLogger("torch.distributed.nn").setLevel(logging.ERROR)  # sshh...
    logging.getLogger("xformers").addFilter(lambda record: 'A matching Triton is not available' not in record.getMessage())
 
    import cv2    # 在 import torch 之前 import cv2
    import torch  # noqa: F401
    startup_timer.record("import torch")

三、启动参数设置

编辑文件 webui-user.sh

# 设置启动参数
export COMMANDLINE_ARGS="--precision autocast --listen --api --no-half --xformers --enable-insecure-extension-access"
 
# 设置模型共享或单独指定模型所在目录
export A1111_HOME="/data/stable-diffusion/stable-diffusion-webui"
export COMMANDLINE_ARGS="${COMMANDLINE_ARGS} \
--ckpt-dir ${A1111_HOME}/models/Stable-diffusion \
--hypernetwork-dir ${A1111_HOME}/models/hypernetworks \
--embeddings-dir ${A1111_HOME}/embeddings \
--lora-dir ${A1111_HOME}/models/Lora \
--vae-dir ${A1111_HOME}/models/VAE}"
 
# 设置虚拟环境所在目录
venv_dir="/data/webui-forge/venv"

四、启动 webui

sh webui.sh