Docker for ML workloads in MLOps - Practice Problems & Coding Challenges

Challenge - 5 Problems

💻 Command Output

intermediate

Output of Docker command with ML model container

You run this command to list running Docker containers for your ML model deployment:

docker ps --filter "ancestor=ml-model:latest" --format "{{.Names}}: {{.Status}}"

What is the expected output if one container named model-serving-1 is running for 5 minutes?

A model-serving-1: Created 5 minutes ago

B No containers found

C ml-model:latest: Up 5 minutes

D model-serving-1: Up 5 minutes

💡 Hint

The command filters containers by image and formats output to show container names and status.

The command lists running containers from the image 'ml-model:latest' and shows their names and uptime. 'Up 5 minutes' means the container is running for 5 minutes.

❓ Configuration

intermediate

Correct Dockerfile snippet for ML dependencies

Which Dockerfile snippet correctly installs Python 3.12 and the ML package scikit-learn in a Debian-based image?

A

FROM python:3.12
RUN pip install sklearn

B

FROM debian
RUN apt-get install python3.12 scikit-learn

C

FROM python:3.12-slim
RUN pip install scikit-learn

D

FROM python:3.12
RUN apt-get install scikit-learn

💡 Hint

Use official Python images and pip for Python packages.

Option C uses the official Python 3.12 slim image and installs scikit-learn via pip, which is the correct way to install Python ML packages in Docker.

❓ Troubleshoot

advanced

Diagnosing Docker container crash for ML model

Your ML model container crashes immediately after starting. The Docker logs show:

ModuleNotFoundError: No module named 'tensorflow'

Which Dockerfile change will fix this error?

A Add <code>ENV TENSORFLOW_HOME=/usr/local/tensorflow</code>

B Add <code>RUN pip install tensorflow</code> before the command that runs the model

C Change base image to <code>tensorflow/tensorflow:latest</code>

D Add <code>RUN apt-get install tensorflow</code>

💡 Hint

The error means the Python package tensorflow is missing.

Installing tensorflow via pip inside the Dockerfile ensures the Python environment has the required package. Changing base image might work but is not guaranteed. apt-get does not install Python packages.

🔀 Workflow

advanced

Correct order to build and deploy ML Docker container

Arrange these steps in the correct order to build and deploy an ML model Docker container:
1. Push image to Docker registry
2. Write Dockerfile with ML dependencies
3. Run container on cloud server
4. Build Docker image locally

A 1,2,3,4

B 2,1,3,4

C 1,3,2,4

D 2,3,1,4

💡 Hint

You must write the Dockerfile before building the image.

First write the Dockerfile, then build the image locally, push it to a registry, and finally run the container on the cloud server.

✅ Best Practice

expert

Best practice for managing large ML model files in Docker

You have a large ML model file (2GB) that changes frequently. What is the best practice to handle this in your Docker workflow?

A Mount the model file as a volume at container runtime to avoid rebuilding image

B Include the model file inside the Docker image to keep everything together

C Download the model file inside the container during build time

D Store the model file in environment variables for easy access

💡 Hint

Think about avoiding rebuilding large images when the model changes.

Mounting the model as a volume allows updating the model without rebuilding the Docker image, saving time and storage.