Docker-MLflow

This repo contains a very simple Docker container to easily run MLflow.

Docker

Build

1. To specify a Dockerfile

docker build -f ./mlflow.dockerfile  -t mlflow . 

Run

To run

docker run -p 5000:5000 mlflow

Stop

To stop immediately

docker stop <containerId>

or after an allocated time, let's say 60s

docker stop -t 60 <containerId>

Check

1. Check running containers

docker ps

2. Filter for container id given image name

docker ps -q --filter ancestor=mlflow

3. Chaining filtering and stopping

docker stop $(docker ps -q --filter ancestor=mlflow)

4. To check for docker images

docker images

Remove

1. To remove tag

docker rmi <REPOSITORY>:<TAG>

2. To remove image

docker rmi <IMAGE ID>

Amwazon ECR

Checking

1. Check ECR repositories

aws ecr describe-repositories --region ap-northeast-1

Push Docker Image

1. Login to AWS ECR

aws ecr get-login-password --region ap-northeast-1 | docker login --username AWS --password-stdin $HOSTNAME

If you see Login Succeeded, you're good to go.

2. Tag the image that will be pushed to ECR

docker tag <IMAGE ID> $HOSTNAME/$REPOSITORY_NAME:mlflow

3. Push the tagged docker image to ECR

Note: Please don't forget the repository name.

docker push $HOSTNAME/$REPOSITORY_NAME:mlflow