This part contains a collection of basic and useful Docker commands in daily using. A full list of Docker commands can be found at here. For the Docker commands for NVIDIA DGX Cloud Service usage can be found at 1 and 2.
Some common use commands are listed below:
1. List images on the system
docker images
It will return the Docker images on the system (below is an example):
REPOSITORY TAG IMAGE ID CREATED SIZE
nvcr.io/nvidia/theano 18.01 dc2fda091dd4 5 weeks ago 4GB
nvcr.io/nvidia/mxnet 18.01-py3 1a70c143e041 5 weeks ago 2.67GB
nvcr.io/nvidia/mxnet 18.01-py2 ffe7de2f34c3 5 weeks ago 2.64GB
nvcr.io/nvidia/pytorch 17.12 5ac6ff8f9a81 2 months ago 4.52GB
nvcr.io/nvidia/tensorflow 17.12 19afd620fc8e 2 months ago 2.88GB
nvcr.io/nvidia/caffe2 17.12 ac17621b6d70 2 months ago 2.8GB
nvcr.io/nvidia/caffe 17.12 cd94fcaaec3f 2 months ago 3.25GB
2. List running containers
docker ps
It will return the Docker containers running by all users now on the system (below is an example):
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
9172d62a0f3f nvcr.io/nvidia/pytorch:19.07-py3 "/usr/local/bin/nvid…" 4 days ago Up 4 days test_attn
2f48d36a73c0 dls/tensorflow:v2.1-cuda10.1-new2 "/usr/local/bin/nvid…" 12 days ago Up 12 days 6006/tcp, 6064/tcp, 8888/tcp clrn_run
d09e539854a9 dls/tensorflow:v2.1-cuda10.1 "/usr/local/bin/nvid…" 13 days ago Up 13 days 6006/tcp, 6064/tcp, 8888/tcp multi-1d-GRU
c7d73d8dbc0e dls/tensorflow:v2.1-cuda10.1 "/usr/local/bin/nvid…" 13 days ago Up 13 days 6006/tcp, 6064/tcp, 8888/tcp music-lstm
2506b4262924 nvcr.io/hunter/pyg1.4.2:latest "/usr/local/bin/nvid…" 3 weeks ago Up 3 weeks pyg1
2b62b12570e2 hlim_tensorflow3 "/usr/local/bin/nvid…" 2 months ago Up 2 months 6006/tcp, 6064/tcp, 8888/tcp interesting_visvesvaraya
6fd51c5709ba quay.io/coreos/etcd:v3.2.26 "/usr/local/bin/etcd" 2 months ago Up 2 months etcd1
76f5d2b3329b nvcr.io/nvidia/pytorch:19.07-py3 "/usr/local/bin/nvid…" 3 months ago Up 2 days drug_attn
288cf9489f6b nvcr.io/nvidia/tensorflow:19.08-py3 "/usr/local/bin/nvid…" 7 months ago Up 2 months 6006/tcp, 6064/tcp, 8888/tcp hlim_tensorflow2
6615744e4934 nvcr.io/nvidia/tensorflow:19.08-py3 "/usr/local/bin/nvid…" 7 months ago Up 2 months 6006/tcp, 6064/tcp, 8888/tcp hlim_tensorflow1
The CONTAINER ID can be used to identify your container in other commands.
3. Run a container
nvidia-docker run --name containerName -it --network=host --rm -v local_dir:container_dir repository:tag
4. Exit a container
Exit a container and close it, then container cannot be found via docker ps:
exit
Exit a container without closing it, container can be found via docker ps:
ctrl+P+Q
5. Resume a container
To resume a container after exiting with “ctrl+P+Q”:
docker exec -it containerID /bin/bash
or
docker attach containerID
Using the second one will sometimes get stuck and have no response for a long time. The first one is preferred.
6. Kill a container
docker kill containerID
7. Save a customized container
Sometimes if you have done some changes (e.g. install new packages or drivers) in a container, it is better to save the modified container as a new container image before exiting it. Then next time you can directly run your saved image to resume the container environment.
docker commit containerID repository:tag
The repository and tag are your customized ones for the next time using.
8. Upload a customized container to NGC
Sometimes a customized container image is built and saved on one HPC, and you want to use the image on another machine. Then it is better to upload the customized container image to our group's cloud space in NGC. Then the image can be pulled from our cloud space on the other machine.
