CODEWITHSUNDEEP
pythondockertensorflow
mihaela7089
mihaela7089

Reputation: 3

Open existing nvidia-docker container

I am new to tensorflow and docker and I am trying to make a docker container where I can use tensorflow with GPU.

After trying different things I managed to set up my nvidia drivers and I can run:

nvidia-docker run -it gcr.io/tensorflow/tensorflow bash

which opens me a docker container with a tensorflow image.

This is good as it creates me the tensorflow docker container where I can install also keras and use it, but if I exit from this container I loose all my created files and installed packages.

How is it possible once I exit this container to go back to it and have all my files and packages still in there?

Below is a screenshot with my docker images (which are too many now because of my many tries to go around this issue and because some tensorflow images did not install properly):

I hope there is an easy way to do this?

Upvotes: 0

Views: 2340

Answers (1)

xxi
xxi

Reputation: 1510

nvidia-docker run -it create NEW container

you want to open existed container use

nvidia-docker start -i CONTAINER_ID or
nvidia-docker start -i NAMES

and if you want more terminals
nvidia-docker exec -it CONTAINER_ID bash or
nvidia-docker exec -it NAMES bash

Update
Run container when the container is close

$ docker ps -a                                                                                                  [ruby-2.3.1p112]
CONTAINER ID        IMAGE               COMMAND             CREATED             STATUS                    PORTS               NAMES
acd83fb407b4        rfcn:rfcn           "/bin/bash"         4 days ago          Exited (0) 28 hours ago                       rfcn
$ nvidia-docker start -i rfcn                                                                                   [ruby-2.3.1p112]
root@acd83fb407b4:/#

Open another terminal when the container is running 

$ nvidia-docker exec -it rfcn bash                                                                              [ruby-2.3.1p112]
root@acd83fb407b4:/#

Upvotes: 1

Related Questions