create docker
nvidia-docker run -it --name bcaffe -v /:/home compute.nvidia.com/nvidia/caffe bash
start and open docker
## start and connect back to previously created container
$ nvidia-docker start bcaffe
$ nvidia-docker attach bcaffe
## delete the container
$ nvidia-docker rm bcaffe
make your own image
docker commit 01cdd31d9f73 bcaffe
See: https://segmentfault.com/a/1190000002567459
https://philipskokoh.github.io/blog/nvidia-docker-for-your-GPU-application-development
View currently running containers
docker ps
View existing mirrors
docker images
View all containers
docker ps -as
delete container
docker rm huwang-pytorch
See: http://www.simapple.com/340.html
Include exited containers, and list the size of the container.
rename image
docker tag IMAGEID(镜像id) REPOSITORY:TAG(仓库:标签)
#例子
docker tag 1febf44a3f77 huwang-bcaffe:v1.0
delete mirror
You need to stop the container derived from this image first
docker rmi [image_id]/[image_name]
push docker
step1——找到本地镜像的ID:docker images
step2——登陆Hub:docker login --username=username --password=password --email=email
step3——tag:docker tag <imageID> <namespace>/<image name>:<version tag eg latest>
简化写法为:docker tag firstimage YOUR_DOCKERHUB_NAME/firstimage
step4——push镜像:docker push <namespace>/<image name>
Reference
[1] Play with Docker in 5 minutes
[2] https://www.oschina.net/question/877522_239204
[3] https://stackoverflow.com/questions/41984399/denied-requested-access-to-the- resource-is-denied-docker
Existing container
huwang-hard
huwang-bcaffe
huwang-ncaffe
huwang-pytorch