视频自动翻译剪辑
生成基础镜像，将所有依赖都放到基础镜像中
docker rmi harbor.5jstore.com:8020/ai/wm_video_cut:v0 docker build -f Dockerfile_v0 -t harbor.5jstore.com:8020/ai/wm_video_cut:v0 .
docker rmi harbor.5jstore.com:8020/ai/wm_video_cut:v0.1 docker run --gpus all --runtime=nvidia -it harbor.5jstore.com:8020/ai/wm_video_cut:v0 /bin/bash wget https://developer.download.nvidia.com/compute/cuda/11.3.0/local_installers/cuda_11.3.0_465.19.01_linux.run sh cuda_11.3.0_465.19.01_linux.run 1、不选驱动 2、报警说发现一个已存在的cuda，是否update，选择否。 3、修改PATH等环境变量： vi ~/.bashrc export PATH=/usr/local/cuda-11.3/bin:$PATH export LD_LIBRARY_PATH=/usr/local/cuda-11.3/lib64:$LD_LIBRARY_PATH export CUDA_HOME=/usr/local/cuda-11.3 删除下载文件 rm cuda_11.3.0_465.19.01_linux.run 提交新镜像 docker commit c9c2a347491d harbor.5jstore.com:8020/ai/wm_video_cut:v0.1
docker rmi harbor.5jstore.com:8020/ai/wm_video_cut:v0.2 docker build -f Dockerfile_v0.2 -t harbor.5jstore.com:8020/ai/wm_video_cut:v0.2 .
生成主镜像
docker rmi harbor.5jstore.com:8020/ai/wm_video_cut:v2 docker build -f Dockerfile -t harbor.5jstore.com:8020/ai/wm_video_cut:v2 .
docker-compose up -d
上述命令会开启一个http服务，公布如下接口:
网关可根据接口做路由和负载
docker run -it harbor.5jstore.com:8020/ai/wm_video_cut:v0 /bin/bash docker run -it harbor.5jstore.com:8020/ai/wm_video_cut:v1 /bin/bash docker run -it harbor.5jstore.com:8020/ai/wm_video_cut:v2 /bin/bash
docker run --gpus all --runtime=nvidia -v ./inputs/:/app/inputs/ -v ./outputs/:/app/outputs/ -it harbor.5jstore.com:8020/common/ai_generate_video:proc_v4 /bin/bash
ffmpeg -hwaccels
python start.py gunicorn start:app -c ./gunicorn.conf.py
flask --app start run --debug