I’m using docker on AWS with a Nvidia Volta AMI and P3.2Xlarge instance and 32 Gb volume attached and have signed in ok with API key and have run the pytorch example successfully (nvidia-docker run --rm -it nvcr.io/nvidia/pytorch:17.10):
docker pull nvcr.io/nvidia/pytorch:17.10
nvidia-docker run --rm -it nvcr.io/nvidia/pytorch:17.10 ....................... Works ok.
Run the MNIST example:
cd /opt/pytorch/examples/mnist ............... in root@fflklsk:/workspace#
python main.py ....................... Works ok.
I try to run: nvidia-docker run --name digits -d -p 8888:5000nvcr.io/nvidia/digits
ubuntu@ip-172-31-27-120:~$ nvidia-docker run --name digits -d -p 8888:5000nvcr.io/nvidia/digits
"docker run" requires at least 1 argument.
See 'docker run --help'.
Usage: docker run [OPTIONS] IMAGE [COMMAND] [ARG...]
Run a command in a new container
ubuntu@ip-172-31-27-120:~$]
I then tried all kinds of permutations as below, with no success.
Please can anybody see what the mistake I am making is? ………. Thanks!
Logging into the NGC Registry at nvcr.io....Login Succeeded
ubuntu@ip-172-31-27-120:~$ docker pull nvcr.io/nvidia/digits:18.11-tensorflow
18.11-tensorflow: Pulling from nvidia/digits
Digest: sha256:c42b32e5c0ca3d428a72df0683f0c7cfd7ef022d544bc0257370d8ad468893c0
Status: Image is up to date for nvcr.io/nvidia/digits:18.11-tensorflow
ubuntu@ip-172-31-27-120:~$ nvidia-docker run --name digits -d -p 8888:5000 -v /h ome/ubuntu/data:/data:ro -v /home/ubuntu/digits- jobs:/workspace/jobs nvcr.io/nv idia/digits
docker: invalid reference format.
See 'docker run --help'.
ubuntu@ip-172-31-27-120:~$ nvidia-docker run --name digits -d -p 8888:5000 -v /h ome/ubuntu/data:/data:ro -v /home/ubuntu/digits-jobs:/jobs nvcr.io/nvidia/digit s
Unable to find image 'nvcr.io/nvidia/digits:latest' locally
docker: Error response from daemon: manifest for nvcr.io/nvidia/digits:latest no t found.
See 'docker run --help'.
ubuntu@ip-172-31-27-120:~$ nvidia-docker run --name digits -d -p 8888:5000nvcr.io/nvidia/digits
"docker run" requires at least 1 argument.
See 'docker run --help'.
Usage: docker run [OPTIONS] IMAGE [COMMAND] [ARG...]
Run a command in a new container
ubuntu@ip-172-31-27-120:~$ nvidia-docker run --name digits -d -p 8888:5000nvcr.io/nvidia/digits:18.11
"docker run" requires at least 1 argument.
See 'docker run --help'.
Usage: docker run [OPTIONS] IMAGE [COMMAND] [ARG...]
Run a command in a new container
ubuntu@ip-172-31-27-120:~$ nvidia-docker run --name digits -d -p 8888:5000nvcr.io/nvidia/digits:18.11-tensorflow
"docker run" requires at least 1 argument.
See 'docker run --help'.
Usage: docker run [OPTIONS] IMAGE [COMMAND] [ARG...]
Run a command in a new container
ubuntu@ip-172-31-27-120:~$ nvidia-docker run --rm -it -d -p 8888:5000nvcr.io/nvidia/digits:18.11-tensorflow
"docker run" requires at least 1 argument.
See 'docker run --help'.
Usage: docker run [OPTIONS] IMAGE [COMMAND] [ARG...]
Run a command in a new container
ubuntu@ip-172-31-27-120:~$ nvidia-docker run --rm -it -d -p 8888:5000nvcr.io/nvidia/digits:18.11
"docker run" requires at least 1 argument.
See 'docker run --help'.
Usage: docker run [OPTIONS] IMAGE [COMMAND] [ARG...]
Run a command in a new container
ubuntu@ip-172-31-27-120:~$ nvidia-docker run --rm -it -d -p 8888:5000nvcr.io/nvidia/digits
"docker run" requires at least 1 argument.
See 'docker run --help'.
Usage: docker run [OPTIONS] IMAGE [COMMAND] [ARG...]
Run a command in a new container
ubuntu@ip-172-31-27-120:~$
ubuntu@ip-172-31-27-120:~$ nvidia-docker run -d --name digits-18:11 -p 8888:5000 --shm-size=1g --ulimit memlock=-1 --ulimit stack=67108864 nvcr.io/nvidia/digits:18:11
docker: invalid reference format.
See 'docker run --help'.
ubuntu@ip-172-31-27-120:~$ nvidia-docker run -d --name digits-18:11-tensorflow -p 8888:5000 --shm-size=1g --ulimit memlock=-1 --ulimit stack=67108864 nvcr.io/nvidia/digits:18:11-tensorflow
docker: invalid reference format.
See 'docker run --help'.
ubuntu@ip-172-31-27-120:~$
ubuntu@ip-172-31-27-120:~$ nvidia-docker run -d --name digits-18:11 -p 8888:5000 --shm-size=1g --ulimit memlock=-1 --ulimit stack=67108864 nvcr.io/nvidia/digits s
Unable to find image 'nvcr.io/nvidia/digits:latest' locally
docker: Error response from daemon: manifest for nvcr.io/nvidia/digits:latest not found.
See 'docker run --help'.
ubuntu@ip-172-31-27-120:~$
…… So I uploaded all my data and tried to create a new object detection dataset, but acter specifying the 4 folders I get this error in digits:
New Object Detection Dataset
train_image_folder
Folder does not exist or is not reachable
val_image_folder
Folder does not exist or is not reachable
val_label_folder
Folder does not exist or is not reachable
train_label_folder
Folder does not exist or is not reachable
Is this something to do with mounting the directories?
I don’t really understand the docker file system or the below:
nvidia-docker run -it -v local_dir:container_dir nvcr.io/nvidia/digits:<xx.xx>
I can create a volume:
$ docker volume create myData
What’s the difference between a volume and a directory?
I can list my volumes:
ubuntu@ip-172-31-24-74:~$ docker volume ls
DRIVER VOLUME NAME
local 1b3c2f8357f2d5ab2ac3b31e3983bb7b1c2b7dae1f8ff4c5568dde9a5a316e43
local 507447a5246619a4f4e5b227f0e9486cadb54c619d95799f6c2f373855814b7c
local myData
ubuntu@ip-172-31-24-74:~$
I can run digits and access it on port 8888, but what to type into the image path?
… seems to create a directory structure that I can see in my SHH console:
/home/ubuntu/data/wasp …… but not accepted by digits browser as a valid directory. Maybe I should try /data/wasp ?
… but I cant SHH any files into ‘wasp’ so I figure it must be describing what’s in the container?
if so, how do I SHH my files into this directory in the container?
Where and how should I put my image and label files such that digits browser can access them?