Some Docker commands often used

After install docker, to try if it works:

docker run hello-world

docker run --rm docker/whalesay cowsay "hello hello hello"

To generate a docker image using a docker file

docker build --network host -t image_name:tag_whatever -f dockerfilename .

To build a container and use tensorboard and then destroy it

docker run -it --rm --user $(id -u):$(id -g) --name containername -v /some_directory:/directory_under_docker --shm-size=1g --gpus=all --ulimit memlock=-1 --network host kirinspeech:tf1.15.2 tensorboard --host 0.0.0.0 --logdir ckpt_dirs --port 3456

To build a container from the above image

docker run -it --user $(id -u):$(id -g) --name containername -v /some_directory:/directory_under_docker --shm-size=1g --ulimit memlock=-1 --network host image_name:tag_whatever bash

docker restart container_name ## the first time when it was built

docker run --name=lite_test -it --user $(id -u):$(id -g) -v /some_directory:/directory_under_docker --shm-size=1g --ulimit memlock=-1 docker.io/library/python:3 /bin/bash

docker run --name=containername -it --user $(id -u):$(id -g) --runtime=nvidia -v /some_directory:/directory_under_docker --shm-size=2g --ulimit memlock=-1 --gpus all nvcr.io/nvidia/pytorch:21.12-py3 /bin/bash

Go to a saved container

docker exec -it container_name bash

docker exec -it --user root container_name bash

Other common docker operations

docker ps -a

docker images -a

docker container stop a_container_id

docker container rm a_container_id

docker image rm an_image_id

# stop all containers from certain image

sudo docker ps -a -q --filter="name=face_tracker/mofapp"

# remove all stopped container

sudo docker container prune

# remove old containers on linux

sudo docker ps -a | grep "weeks ago" | awk "{print $1}" | xargs --no-run-if-empty sudo docker rm

sudo docker ps -a | grep 'weeks ago\|months ago' | awk '{print $1}' | xargs --no-run-if-empty sudo docker rm

To save the docker image from a docker container

docker commit [NAME] [TAG]

docker save -o [DIR]/abcde.tar [TAG]

docker load -i abcde.tar

## or

docker load -i < abcde.tar.gz

To use pycharm professional installed at local machine to remotely debug projects at server using remote docker container (from zixing):

At the server side, set up a docker container:

#### build a new docker with port map

docker run -p 1234:22 --name=somename -it --user $(id -u):$(id -g) --runtime=nvidia -v /aa/bb:/cc -v /dd:/ee --shm-size=1g --ulimit memlock=-1 --ulimit stack=67108864 --gpus all dockerimage:tag /bin/bash

§ (disable --network host)

§ 22: container port for sftp; 1234: host port, self-defined

§ Use `docker ps` to check whether the port map works

docker port somename

Within the docker container, apply the following changes

## exec --user root -it somename bash

§ apt-get update

§ apt-get install openssh-server openssh-client

§ nano /etc/ssh/sshd_config

change #PermitRootLogin without-password -----> PermitRootLogin yes

change #PasswordAuthentication yes -----> PasswordAuthentication yes

change #UsePAM yes -----> UsePAM no

§ passwd root

§ service ssh restart

## test if the ssh access to this container is built

ssh root@ipaddress -p 1234

#### Setup the pycharm at local machine

§ Deployment--> configuration --> sftp

§ Setting --> project interprater --> ssh interpreter --> existing --> (others)

https://www.cnblogs.com/ruiyang-/p/10158658.html

https://blog.csdn.net/u012620515/article/details/89856072

An example of dockerfile:

FROM nvcr.io/nvidia/tensorflow:20.06-tf1-py3

# Run and updates and cleans.

RUN apt-get clean -y

RUN apt-get install -f -y

RUN dpkg --configure -a

RUN apt-get update -y

RUN apt-get install cmake build-essential pkg-config

# Install ffmpeg

RUN apt install ffmpeg -y

# Install pip packages.

RUN pip install --upgrade pip

RUN pip install \

num2words \

pypinyin \

subword-nmt==0.3.6 \

tqdm==4.34.0

RUN pip list

# Finishing touches.

WORKDIR .