[miscellaneous] docker usage
기타 유용한
Docker명령어에 대해 작성합니다.
Make Dockerfile
- Get docker image
docker pull nvidia/cuda:11.6.1-devel-ubuntu20.04
- Dockerfile
- 반드시 파일의 제목은
Dockerfile이어야 한다. (확장자는 없음.)
- 반드시 파일의 제목은
Example Dockerfile
# 기존 이미지를 기반으로 사용
# Use nvidia/cuda version matches your server
FROM nvidia/cuda:11.6.1-cudnn8-devel-ubuntu20.04
# 필요한 패키지 설치
# Install ubuntu apt packages. Do not remove default packages.
RUN apt-get update
# opencv-python error
RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install\
libgl1\
libgl1-mesa-glx \
libglib2.0-0 -y && \
rm -rf /var/lib/apt/lists/*
RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y \
wget \
apt-utils \
build-essential \
ca-certificates \
curl \
git \
htop \
sudo \
vim \
python3-dev \
python3-pip \
&& rm -rf /var/lib/apt/lists/*
# Miniconda 설치
RUN wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh -O /miniconda.sh
RUN bash /miniconda.sh -b -p /miniconda
ENV PATH="/miniconda/bin:${PATH}"
# Conda를 이용해 Python3 설치 및 Python 패키지 설치
RUN conda install -y python=3.9 && \
pip3 --no-cache-dir install --upgrade \
pip \
setuptools \
ipython \
ipdb \
matplotlib \
pandas \
scipy \
torch \
jupyter \
torchvision \
torchtext \
torchsummary \
slacker \
tqdm
# # 사용자 설정
# ARG UNAME
# ARG UID
# ARG GID
# # Ensure that the group and user are created successfully
# RUN if [ -z "$UNAME" ] || [ -z "$UID" ] || [ -z "$GID" ]; then echo "UNAME, UID, and GID arguments are required" && exit 1; fi && \
# addgroup --gid ${GID} ${UNAME} && \
# useradd -m -u ${UID} -g ${GID} -s /bin/bash ${UNAME} && \
# adduser ${UNAME} sudo
# USER ${UNAME}
# WORKDIR /home/${UNAME}
CMD [ "/bin/bash" ]
Build Log
joonhyung@devbox:~/dockers/joonh_cu116$ docker build -t joonh_cu116 .
[+] Building 260.0s (11/11) FINISHED
=> [internal] load .dockerignore 0.0s
=> => transferring context: 2B 0.0s
=> [internal] load build definition from Dockerfile 0.0s
=> => transferring dockerfile: 1.72kB 0.0s
=> [internal] load metadata for docker.io/nvidia/cuda:11.6.1-cudnn8-devel-ubuntu20.04 0.0s
=> [1/7] FROM docker.io/nvidia/cuda:11.6.1-cudnn8-devel-ubuntu20.04 0.0s
=> CACHED [2/7] RUN apt-get update 0.0s
=> [3/7] RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install libgl1 libgl1-mesa-glx libglib2.0-0 -y && rm -rf /var/lib/apt/lists/* 19.2s
=> [4/7] RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y wget apt-utils build-essential ca-certificates curl git htop s 24.1s
=> [5/7] RUN wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh -O /miniconda.sh 3.0s
=> [6/7] RUN bash /miniconda.sh -b -p /miniconda 6.4s
=> [7/7] RUN conda install -y python=3.9 && pip3 --no-cache-dir install --upgrade pip setuptools ipython ipdb matplotlib pandas scipy torc 177.4s
=> exporting to image 29.8s
=> => exporting layers 29.8s
=> => writing image sha256:48e3f2be75424e71be48c75d5472c90a9bf5ac9c94586bd723127d5bed67714b 0.0s
=> => naming to docker.io/library/joonh_cu116
Build docker image
docker build -t [이미지 이름:이미지 버전] [Dockerfile의 경로]
docker build -t joonh_cu116 .
docker build --build-arg UNAME=joonhyung-lee --build-arg UID=1001 --build-arg GID=1001 -t joonh_cu116 .
Get built docker images
docker images
>>> joonh_cu116 latest 7c1bae5ab933 33 seconds ago 9.56GB
Start Docker Container
docker run -v [로컬_경로]:[컨테이너_경로] -it --gpus all [이미지_이름]:[태그] /bin/bash
docker run -v /home/joonhyung/python/:/home/root/python -it --gpus all joonh_cu116:latest /bin/bash
Commit and Push Docker Container
get information about container
joonhyung@devbox:~/dockers/joonh_cu116$ docker ps
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
79b602e29923 c2e91d98d3ed "/opt/nvidia/nvidia_…" About an hour ago Up About an hour magical_neumann
Commit
docker commit CONTAINER IMAGE_NAME
joonhyung@devbox:~/dockers/joonh_cu116$ docker commit 79b602e29923 joonh_cu116:v00
sha256:f50aabee25697dc96bc80f76fd231c74d1fbfe77cb47698a4ec89e0b84c5ba81
- docker images로 image가 적절하게 생성되었는지 확인 가능.
joonhyung@devbox:~/dockers/joonh_cu116$ docker images
REPOSITORY TAG IMAGE ID CREATED SIZE
joonh_cu116 v00 f50aabee2569 14 seconds ago 31GB
Build Dockerfile
- Add lines
-
FROM joonh_cu116:v00 # Install ubuntu apt packages. RUN sudo apt update && sudo apt install -y <ubuntu-packages>
-
- Build New version
-
docker build -t joonh_cu116:v01 .
-
Login Docker server
joonhyung@devbox:~/dockers/joonh_cu116_v00$ docker login -u joonhyunglee
Password:
WARNING! Your password will be stored unencrypted in /home/joonhyung/.docker/config.json.
Configure a credential helper to remove this warning. See
https://docs.docker.com/engine/reference/commandline/login/#credentials-store
Login Succeeded
push on Docker
joonhyung@devbox:~/dockers/joonh_cu116_v00$ docker push joonhyunglee/joonh_cu116:v01
Enjoy Reading This Article?
Here are some more articles you might like to read next: