Build_docker_image.md

Run Sport Gesture Classification on Docker

• I did not install nvidia runtime to let docker access the local gpu. Now it would only use cpu device.

Install Docker Server & Client

# remove the installed docker
sudo apt-get purge docker-ce docker-ce-cli containerd.io docker-buildx-plugin docker-compose-plugin docker-ce-rootless-extras
sudo rm -rf /var/lib/docker
sudo rm -rf /var/lib/containerd

# install
sudo apt-get install docker-ce docker-ce-cli containerd.io docker-compose-plugin

# check docker version
docker version

• If meeting any errors, you can check Installation.md

Build Docker

# build image 
docker build -t sports_api:<tag> . 
# --no-cache (normally when it failed, it would build from the failed layer, with `--no-cache`, it would remove all the cache and build from the start)

# run docker without gpu
docker run \
-it \
--rm \
-p 12000:12000 \
-p 6006:6006 \
-v /home/linlin/dataset/sports_kaggle:/home/linlin/dataset/sports_kaggle \
sports_api:v1
# -t gives the sudo terminal
# -i: give the interacting interface
# -p host_port:container_port, 12000 for flask, 6006 for tensorboard
# -v $host_path:$container_path

Train in Docker Container

python3 train.py

Tensorboard Events

• Here the local logging events would be saved into docker with the same directory as /home/linlin/ll_docker/sportsnoma-deep-learning/sports_events
• The new events file when training on docker would be saved on this folder in docker, but files are still on docker container, not locally exist

Access the Tensorboard in Docker

• Run python3 -m tensorboard.main --logdir=. --bind_all on container
• In local pc: localhost:6006 ?? video

Data Transfer between container and local

sudo docker cp container-id:/path/filename.txt ~/Desktop/filename.txt
sudo docker cp foo.txt container_id:/foo.txt

Run API in Docker Container

python3 api.py 

• In local pc: go to localhost:12000 to go to flask api

Acess

• Modify the file inside the docker (so docker image would be updated)

  • terminal 1

  docker run -v /home/linlin/dataset/sports_kaggle/:/home/linlin/dataset/sports_kaggle/ \
  -it \
  --rm \
  sports_api:<tag>  # docker run: create a new container    
  

  • terminal 2

  docker exec -it <container_id> /bin/bash  # docker exec: run command on running container
  # modify the file
  # ctrl + d exit
  docker commit -m "message" <container id> sports_api:<tag>
  

Save Image

docker save sports_api:<tag> > docker_sports_api.tar
docker save myimage:<tag> | gzip > docker_image_sports_api.tar.gz
docker save sports_api:<tag> --output docker_image_sports_api.tar

docker load --input *.tar  #  It restores both images and tags.
docker load < *.tar