When I was new to Docker and was tired of repeatedly searching the internet for docker commands. Therefore I created this guide of my commonly used commands.
Link to the Docker CLI: https://docs.docker.com/engine/reference/commandline/docker/
Listing all pulled and/or build images on your system using docker image ls or for short docker images.
$ docker image ls
REPOSITORY TAG IMAGE ID CREATED SIZE
ai2ys/deep-learning-workstation-setup/examples/tensorflow 22.10-tf2-py3 ac06a1248d15 14 hours ago 14.6GB
ai2ys/deep-learning-workstation-setup/examples/pytorch 22.10-py3 d038ce706291 14 hours ago 17GB
nvidia/cuda 12.0.1-base-ubuntu22.04 cdd8fce1e7b1 4 weeks ago 243MB
nvcr.io/nvidia/tensorflow 22.10-tf2-py3 6165f4ee35c6 4 months ago 14.4GB
nvcr.io/nvidia/pytorch 22.10-py3 b1309fc25c28 4 months ago 16.9GB
When building images you should call docker image prune from time to time to remove dangling images.
This requires having installed the NVIDIA Container Toolkit. Be sure you have followed all instructions mentioned in NVIDIA Container Toolkit instruction previously.
GPUs can get made accessible using the option --gpus all for making all installed GPU accessible. Alternatively dedicated GPU can get made accessible using --gpus <comma separated list of indices or GPU-UUIDS>. Below an example passing the GPU with index 0 and running the command nvidia-smi.
$ docker run --rm --gpus 0 nvidia/cuda:12.0.1-base-ubuntu22.04 nvidia-smi
Sun Mar 5 10:18:51 2023
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 525.85.12 Driver Version: 525.85.12 CUDA Version: 12.0 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|===============================+======================+======================|
| 0 NVIDIA TITAN RTX On | 00000000:65:00.0 On | N/A |
| 40% 40C P0 56W / 280W | 1200MiB / 24576MiB | 2% Default |
| | | N/A |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=============================================================================|
+-----------------------------------------------------------------------------+
Use the command docker ps to list running containers. When additionally using the option --all also stopped container container will be displayed.
$ docker ps
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
e40ed8584543 nvcr.io/nvidia/tensorflow:22.10-tf2-py3 "/opt/nvidia/nvidia_…" About a minute ago Up About a minute 6006/tcp, 6064/tcp, 8888/tcp relaxed_hodgkin
An alternate way is to use the command docker container ls.
docker attach <container name or ID>
docker exec -it <container name or ID> bash
Assuming having started a Jupyter notebook in a container, but the terminal window being detached from the container (option --detach)
$ docker run --rm --detach -it --gpus all -p 8888:8888 nvcr.io/nvidia/tensorflow:22.10-tf2-py3 jupyter notebook --ip=0.0.0.0 --port=8888 --no-browser
0da83916789f149011098ad00cd61a7a7f297cf58e041eee457110cb6ec23c1
$ docker ps
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
0da83916789f nvcr.io/nvidia/tensorflow:22.10-tf2-py3 "/opt/nvidia/nvidia_…" 2 minutes ago Up 2 minutes 6006/tcp, 6064/tcp, 0.0.0.0:8888->8888/tcp, :::8888->8888/tcp dazzling_benz
$ docker exec dazzling_benz jupyter notebook list
Currently running servers:
http://0.0.0.0:8888/?token=f20fd1fd97882e8bb9f211996cbae2b5227099a06ce58154 :: /workspace
If JupyterLab is used instead of Jupyter notebook use juypter lab list instead of jupyter notebook list
docker composePlease checkout on how to use the docker compose following the examples mentiond in the main README.md - Docker Compose examples with GPU support section.