lpgonzalez/docker_streamlit_app repository overview

⁠docker-streamlit-app 🚀

A compact, Dockerized Streamlit demo that loads a pre-packed scikit-learn model and serves a prediction UI (Iris dataset). Built for teaching and fast iteration.

⁠Authors

• Lisardo Prieto González
• Ricardo Aler Mur
• Luis Mario García Lafuente

⁠Table of Contents

• docker-streamlit-app 🚀⁠
  • Authors⁠
  • Table of Contents⁠
  • What it does ⚙️⁠
  • Quick links 🔗⁠
  • Prerequisites ✅⁠
  • Quick start (recommended) ⚡⁠
  • Development / Iteration 🛠️⁠
  • Makefile targets (summary) 📦⁠
  • Files of interest 📁⁠
  • Tips & Troubleshooting 🩺⁠
  • License 📜⁠

⁠What it does ⚙️

• Runs a Streamlit app on port 8501.
• Loads a serialized "pack" (model + metadata) from app/pack_for_streamlit.joblib via app/myst_v4.py.
• Presents a small UI to enter features and obtain predicted class + probabilities.

⁠Quick links 🔗

• App entry: app/myst_v4.py
• Pack: app/pack_for_streamlit.joblib
• Requirements: app/requirements.txt
• Notebook used to create the pack: app/p1_sol_deployment.ipynb
• Docker config: Dockerfile
• Helpers: Makefile

⁠Prerequisites ✅

• Docker Engine
• GNU Make (optional but recommended)
• (Optional for local dev) Python 3.12

⁠Quick start (recommended) ⚡

1. Build the production runtime image:

make build-prod

1. Run the packaged container (detached):

make deploy

1. Open the app in your browser:

http://localhost:8501

To stop and remove the container:

make stop
make clean

⁠Development / Iteration 🛠️

• Mount local app/ into the running container (no image rebuild needed):

make deploy-devel

• Run interactively (attach to logs, removed on exit):

make deploy-devel-interactive

• Enter a shell inside a running container:

make bash

⁠Makefile targets (summary) 📦

• make build-prod — Build the production/runtime image (multi-stage Dockerfile).
• make deploy — Run packaged container (detached).
• make deploy-devel — Run container with local app/ mounted for fast dev.
• make deploy-devel-interactive — Run container interactively (attach to stdout/stderr).
• make stop — Stop the running container.
• make clean — Stop and remove the container.
• make build-prod-all — Alias to build-prod (also tags :latest).
• make push-to-repo — Push images to a registry (if tagged).
• make save-image / load-image — Save/load tarball of the image.

⁠Files of interest 📁

• app/myst_v4.py — Streamlit entrypoint (loads the joblib pack).
• app/pack_for_streamlit.joblib — Pre-built model + metadata.
• app/requirements.txt — Python deps installed during build.
• Dockerfile — Multi-stage build (builder + runtime).
• Makefile — Convenience tasks to build, run and debug.

⁠Tips & Troubleshooting 🩺

• Healthcheck endpoint (inside container): http://localhost:8501/_stcore/health⁠
• If port 8501 is busy, either stop the other process or change port mapping in the Makefile.
• Inspect container logs:

docker logs docker-streamlit-app

• If dependencies change, rebuild the image:

make build-prod

• For UI-only edits, prefer make deploy-devel to avoid rebuilds.

⁠License 📜

MIT — see LICENSE⁠