杭州阿里 mkmk 仓库

registry.cn-hangzhou.aliyuncs.com/mkmk/all

启动 本地 register server

默认部署的 是 http server， 但是 client 端 是 https 的 ，所以 在 client 端 需要 做一些 配置

docker run -d -p 5000:5000 --restart always --name registry registry

查看 所有仓库

http://192.168.99.100:5000/v2/_catalog

http://registry_ip:5000/v2/nginx/tags/list


http://192.168.111.200:5000/v2/tensorflow/tensorflow/tags/list


http://192.168.99.100:5000/v2/python/tags/list

修改 client 端 配置

## 添加 “ insecure-registries ”

cat >  /etc/docker/daemon.json <<"EOF"

{
  "registry-mirrors": ["https://wm12hkla.mirror.aliyuncs.com"],
  "insecure-registries" : ["192.168.111.200:5000"]
}  +

EOF

systemctl restart docker

测试 本地 register

# 测试 本地镜像 仓库
docker pull nginx
docker tag nginx 192.168.19.10:5000/nginx:new
docker push 192.168.19.10:5000/nginx:new



docker images

docker rmi 192.168.19.10:5000/nginx:new

docker images

docker pull 192.168.19.10:5000/nginx:new

私有 镜像仓库

# sever
docker run -d -p 5000:5000 --restart always --name registry1 registry

for i in 'registry:latest' 'owncloud:10' 
do
j=${i//:/-}
remote_path="registry.cn-hangzhou.aliyuncs.com/mkmk/all:${j}"
docker tag ${i} ${remote_path}
docker push  ${remote_path}
done

# client

cat >  /etc/docker/daemon.json <<"EOF"

{
  "registry-mirrors": ["https://wm12hkla.mirror.aliyuncs.com"],
  "insecure-registries" : ["192.168.111.200:5000"]
}

EOF

systemctl restart docker


register_url='192.168.111.200:5000'
image_url='tensorflow/tensorflow:2.4.1-gpu'

docker pull $image_url
docker tag  $image_url  ${register_url}/$image_url
docker push ${register_url}/$image_url

批量推送镜像

# client

echo 'nameserver 114.114.114.114' > /etc/resolv.conf

register_url='192.168.111.200:5000'
for image_url in 'tensorflow/tensorflow:2.4.1-gpu'  'tensorflow/tensorflow:2.4.1-gpu-jupyter'
do
    docker pull $image_url
    docker tag  $image_url  ${register_url}/$image_url
    docker push ${register_url}/$image_url
done


echo 'nameserver 114.114.114.114' > /etc/resolv.conf

register_url='192.168.111.200:5000'
for image_url in 'nvidia/cuda:11.1.1-devel'  'nvidia/cuda:11.1.1-runtime'
do
    docker pull $image_url
    docker tag  $image_url  ${register_url}/$image_url
    docker push ${register_url}/$image_url
done

gpu

apt install nvidia-driver-450-server -y


docker run -it --gpus=all gpu-burn:cuda11.1


register_url='192.168.111.200:5000'
image_url='gpu-burn:cuda11.1'
docker tag  $image_url  ${register_url}/$image_url
docker push ${register_url}/$image_url

带数据 启动 register

docker run -d -p 5000:5000 --restart always --name registry1 -v  /root/registry:/var/lib/registry   registry

欢迎大家一起交流呀
qq群：3638803451
vx：wxid_sgdelhiwombj12

原文地址：http://www.cnblogs.com/ltgybyb/p/16900559.html